Datasets:

NeurIPSConference

/

NeurIPS2024-papers

like 0

Follow

Conference on Neural Information Processing Systems 1

https://openreview.net/forum?id=psG4LXlDNs

@inproceedings{ jiang2024achieving, title={Achieving \${\textbackslash}tilde\{O\}(1/{\textbackslash}epsilon)\$ Sample Complexity for Constrained Markov Decision Process}, author={Jiashuo Jiang and Yinyu Ye}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=psG4LXlDNs} }

We consider the reinforcement learning problem for the constrained Markov decision process (CMDP), which plays a central role in satisfying safety or resource constraints in sequential learning and decision-making. In this problem, we are given finite resources and a MDP with unknown transition probabilities. At each stage, we take an action, collecting a reward and consuming some resources, all assumed to be unknown and need to be learned over time. In this work, we take the first step towards deriving optimal problem-dependent guarantees for the CMDP problems. We derive a logarithmic regret bound, which translates into a $O(\frac{1}{\Delta\cdot\epsilon}\cdot\log^2(1/\epsilon))$ sample complexity bound, with $\Delta$ being a problem-dependent parameter, yet independent of $\epsilon$. Our sample complexity bound improves upon the state-of-art $O(1/\epsilon^2)$ sample complexity for CMDP problems established in the previous literature, in terms of the dependency on $\epsilon$. To achieve this advance, we develop a new framework for analyzing CMDP problems. To be specific, our algorithm operates in the primal space and we resolve the primal LP for the CMDP problem at each period in an online manner, with \textit{adaptive} remaining resource capacities. The key elements of our algorithm are: i) a characterization of the instance hardness via LP basis, ii) an eliminating procedure that identifies one optimal basis of the primal LP, and; iii) a resolving procedure that is adaptive to the remaining resources and sticks to the characterized optimal basis.

Achieving Õ(1/ϵ) Sample Complexity for Constrained Markov Decision Process

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=psDrko9v1D

@inproceedings{ ma2024efficient, title={Efficient Combinatorial Optimization via Heat Diffusion}, author={Hengyuan Ma and Wenlian Lu and Jianfeng Feng}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=psDrko9v1D} }

Combinatorial optimization problems are widespread but inherently challenging due to their discrete nature. The primary limitation of existing methods is that they can only access a small fraction of the solution space at each iteration, resulting in limited efficiency for searching the global optimal. To overcome this challenge, diverging from conventional efforts of expanding the solver's search scope, we focus on enabling information to actively propagate to the solver through heat diffusion. By transforming the target function while preserving its optima, heat diffusion facilitates information flow from distant regions to the solver, providing more efficient navigation. Utilizing heat diffusion, we propose a framework for solving general combinatorial optimization problems. The proposed methodology demonstrates superior performance across a range of the most challenging and widely encountered combinatorial optimizations. Echoing recent advancements in harnessing thermodynamics for generative artificial intelligence, our study further reveals its significant potential in advancing combinatorial optimization.

Efficient Combinatorial Optimization via Heat Diffusion

NeurIPS.cc/2024/Conference

2403.08757

poster

https://openreview.net/forum?id=prgxz9fYbf

@inproceedings{ milsom2024stochastic, title={Stochastic Kernel Regularisation Improves Generalisation in Deep Kernel Machines}, author={Edward Milsom and Ben Anson and Laurence Aitchison}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=prgxz9fYbf} }

Recent work developed convolutional deep kernel machines, achieving 92.7% test accuracy on CIFAR-10 using a ResNet-inspired architecture, which is SOTA for kernel methods. However, this still lags behind neural networks, which easily achieve over 94% test accuracy with similar architectures. In this work we introduce several modifications to improve the convolutional deep kernel machine’s generalisation, including stochastic kernel regularisation, which adds noise to the learned Gram matrices during training. The resulting model achieves 94.5% test accuracy on CIFAR-10. This finding has important theoretical and practical implications, as it demonstrates that the ability to perform well on complex tasks like image classification is not unique to neural networks. Instead, other approaches including deep kernel methods can achieve excellent performance on such tasks, as long as they have the capacity to learn representations from data.

Stochastic Kernel Regularisation Improves Generalisation in Deep Kernel Machines

NeurIPS.cc/2024/Conference

2410.06171

poster

https://openreview.net/forum?id=preo49P1VY

@inproceedings{ hwang2024hydra, title={Hydra: Bidirectional State Space Models Through Generalized Matrix Mixers}, author={Sukjun Hwang and Aakash Lahoti and Ratish Puduppully and Tri Dao and Albert Gu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=preo49P1VY} }

A wide array of sequence models are built on a framework modeled after Transformers, comprising alternating sequence mixer and channel mixer layers. This paper studies a unifying *matrix mixer* view of sequence mixers that can be conceptualized as a linear map on the input sequence. This framework encompasses a broad range of well-known sequence models, including the self-attention of Transformers as well as recent strong alternatives such as structured state space models (SSMs), and allows understanding downstream characteristics such as efficiency and expressivity through properties of their structured matrix class. We identify a key axis of matrix parameterizations termed *sequence alignment*, which increases the flexibility and performance of matrix mixers, providing insights into the strong performance of Transformers and recent SSMs such as Mamba. Furthermore, the matrix mixer framework offers a systematic approach to developing sequence mixers with desired properties, allowing us to develop several new sub-quadratic sequence models. In particular, we propose a natural bidirectional extension of the Mamba model (**Hydra**), parameterized as a *quasiseparable matrix mixer*, which demonstrates superior performance over other sequence models including Transformers on non-causal tasks. As a drop-in replacement for attention layers, \name outperforms BERT by 0.8 points on the GLUE benchmark and ViT by 2% Top-1 accuracy on ImageNet.

Hydra: Bidirectional State Space Models Through Generalized Matrix Mixers

NeurIPS.cc/2024/Conference

2407.09941

https://huggingface.co/papers/2407.09941

poster

https://openreview.net/forum?id=prXfM5X2Db

@inproceedings{ wang2024frieren, title={Frieren: Efficient Video-to-Audio Generation Network with Rectified Flow Matching}, author={Yongqi Wang and Wenxiang Guo and Rongjie Huang and Jiawei Huang and Zehan Wang and Fuming You and Ruiqi Li and Zhou Zhao}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=prXfM5X2Db} }

Video-to-audio (V2A) generation aims to synthesize content-matching audio from silent video, and it remains challenging to build V2A models with high generation quality, efficiency, and visual-audio temporal synchrony. We propose Frieren, a V2A model based on rectified flow matching. Frieren regresses the conditional transport vector field from noise to spectrogram latent with straight paths and conducts sampling by solving ODE, outperforming autoregressive and score-based models in terms of audio quality. By employing a non-autoregressive vector field estimator based on a feed-forward transformer and channel-level cross-modal feature fusion with strong temporal alignment, our model generates audio that is highly synchronized with the input video. Furthermore, through reflow and one-step distillation with guided vector field, our model can generate decent audio in a few, or even only one sampling step. Experiments indicate that Frieren achieves state-of-the-art performance in both generation quality and temporal alignment on VGGSound, with alignment accuracy reaching 97.22\%, and 6.2\% improvement in inception score over the strong diffusion-based baseline. Audio samples and code are available at http://frieren-v2a.github.io.

Frieren: Efficient Video-to-Audio Generation Network with Rectified Flow Matching

NeurIPS.cc/2024/Conference

2406.00320

poster

https://openreview.net/forum?id=pqi4vqBYXW

@inproceedings{ wang2024hallod, title={Hallo3D: Multi-Modal Hallucination Detection and Mitigation for Consistent 3D Content Generation}, author={Hongbo Wang and Jie Cao and Jin Liu and Xiaoqiang Zhou and Huaibo Huang and Ran He}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pqi4vqBYXW} }

Recent advancements in 3D content generation have been significant, primarily due to the visual priors provided by pretrained diffusion models. However, large 2D visual models exhibit spatial perception hallucinations, leading to multi-view inconsistency in 3D content generated through Score Distillation Sampling (SDS). This phenomenon, characterized by overfitting to specific views, is referred to as the "Janus Problem". In this work, we investigate the hallucination issues of pretrained models and find that large multimodal models without geometric constraints possess the capability to infer geometric structures, which can be utilized to mitigate multi-view inconsistency. Building on this, we propose a novel tuning-free method. We represent the multimodal inconsistency query information to detect specific hallucinations in 3D content, using this as an enhanced prompt to re-consist the 2D renderings of 3D and jointly optimize the structure and appearance across different views. Our approach does not require 3D training data and can be implemented plug-and-play within existing frameworks. Extensive experiments demonstrate that our method significantly improves the consistency of 3D content generation and specifically mitigates hallucinations caused by pretrained large models, achieving state-of-the-art performance compared to other optimization methods.

Hallo3D: Multi-Modal Hallucination Detection and Mitigation for Consistent 3D Content Generation

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pqYceEa87j

@inproceedings{ qiu2024spectral, title={Spectral Editing of Activations for Large Language Model Alignment}, author={Yifu QIU and Zheng Zhao and Yftah Ziser and Anna Korhonen and Edoardo Ponti and Shay B Cohen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pqYceEa87j} }

Large language models (LLMs) often exhibit undesirable behaviours, such as generating untruthful or biased content. Editing their internal representations has been shown to be effective in mitigating such behaviours on top of the existing alignment methods. We propose a novel inference-time editing method, namely spectral editing of activations (SEA), to project the input representations into directions with maximal covariance with the positive demonstrations (e.g., truthful) while minimising covariance with the negative demonstrations (e.g., hallucinated). We also extend our method to non-linear editing using feature functions. We run extensive experiments on benchmarks concerning truthfulness and bias with six open-source LLMs of different sizes and model families. The results demonstrate the superiority of SEA in effectiveness, generalisation to similar tasks, as well as computation and data efficiency. We also show that SEA editing only has a limited negative impact on other model capabilities.

Spectral Editing of Activations for Large Language Model Alignment

NeurIPS.cc/2024/Conference

2405.09719

poster

https://openreview.net/forum?id=pqD7ckR8AF

@inproceedings{ abdolahpourrostam2024superdeepfool, title={SuperDeepFool: a new fast and accurate minimal adversarial attack}, author={Alireza Abdolahpourrostam and Mahed Abroshan and Seyed-Mohsen Moosavi-Dezfooli}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pqD7ckR8AF} }

Deep neural networks have been known to be vulnerable to adversarial examples, which are inputs that are modified slightly to fool the network into making incorrect predictions. This has led to a significant amount of research on evaluating the robustness of these networks against such perturbations. One particularly important robustness metric is the robustness to minimal $\ell_{2}$ adversarial perturbations. However, existing methods for evaluating this robustness metric are either computationally expensive or not very accurate. In this paper, we introduce a new family of adversarial attacks that strike a balance between effectiveness and computational efficiency. Our proposed attacks are generalizations of the well-known DeepFool (DF) attack, while they remain simple to understand and implement. We demonstrate that our attacks outperform existing methods in terms of both effectiveness and computational efficiency. Our proposed attacks are also suitable for evaluating the robustness of large models and can be used to perform adversarial training (AT) to achieve state-of-the-art robustness to minimal $\ell_{2}$ adversarial perturbations.

SuperDeepFool: a new fast and accurate minimal adversarial attack

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=poE54GOq2l

@inproceedings{ li2024snapkv, title={Snap{KV}: {LLM} Knows What You are Looking for Before Generation}, author={Yuhong Li and Yingbing Huang and Bowen Yang and Bharat Venkitesh and Acyr Locatelli and Hanchen Ye and Tianle Cai and Patrick Lewis and Deming Chen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=poE54GOq2l} }

Large Language Models (LLMs) have made remarkable progress in processing extensive contexts, with the Key-Value (KV) cache playing a vital role in enhancing their performance. However, the growth of the KV cache in response to increasing input length poses challenges to memory and time efficiency. To address this problem, this paper introduces SnapKV, an innovative and fine-tuning-free approach that efficiently minimizes KV cache size while still delivering comparable performance in real-world applications. We discover that each attention head in the model consistently focuses on specific prompt attention features during generation. Meanwhile, this robust pattern can be obtained from an `observation' window located at the end of the prompts. Drawing on this insight, SnapKV automatically compresses KV caches by selecting clustered important KV positions for each attention head. Our approach significantly reduces the growing computational overhead and memory footprint when processing long input sequences. Specifically, SnapKV achieves a consistent decoding speed with a 3.6x increase in generation speed and an 8.2x enhancement in memory efficiency compared to baseline when processing inputs of 16K tokens. At the same time, it maintains comparable performance to baseline models across 16 long sequence datasets. Moreover, SnapKV can process up to 380K context tokens on a single A100-80GB GPU using HuggingFace implementation with minor changes, exhibiting only a negligible accuracy drop in the Needle-in-a-Haystack test. Further comprehensive studies suggest SnapKV's potential for practical applications.

SnapKV: LLM Knows What You are Looking for Before Generation

NeurIPS.cc/2024/Conference

2404.14469

https://huggingface.co/papers/2404.14469

poster

https://openreview.net/forum?id=pnmUiVAGnv

@inproceedings{ huang2024cat, title={{CAT}: Coordinating Anatomical-Textual Prompts for Multi-Organ and Tumor Segmentation}, author={Zhongzhen Huang and Yankai Jiang and Rongzhao Zhang and Shaoting Zhang and Xiaofan Zhang}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pnmUiVAGnv} }

Existing promptable segmentation methods in the medical imaging field primarily consider either textual or visual prompts to segment relevant objects, yet they often fall short when addressing anomalies in medical images, like tumors, which may vary greatly in shape, size, and appearance. Recognizing the complexity of medical scenarios and the limitations of textual or visual prompts, we propose a novel dual-prompt schema that leverages the complementary strengths of visual and textual prompts for segmenting various organs and tumors. Specifically, we introduce $\textbf{\textit{CAT}}$, an innovative model that $\textbf{C}$oordinates $\textbf{A}$natomical prompts derived from 3D cropped images with $\textbf{T}$extual prompts enriched by medical domain knowledge. The model architecture adopts a general query-based design, where prompt queries facilitate segmentation queries for mask prediction. To synergize two types of prompts within a unified framework, we implement a ShareRefiner, which refines both segmentation and prompt queries while disentangling the two types of prompts. Trained on a consortium of 10 public CT datasets, $\textbf{\textit{CAT}}$ demonstrates superior performance in multiple segmentation tasks. Further validation on a specialized in-house dataset reveals the remarkable capacity of segmenting tumors across multiple cancer stages. This approach confirms that coordinating multimodal prompts is a promising avenue for addressing complex scenarios in the medical domain.

CAT: Coordinating Anatomical-Textual Prompts for Multi-Organ and Tumor Segmentation

NeurIPS.cc/2024/Conference

2406.07085

poster

https://openreview.net/forum?id=plH8gW7tPQ

@inproceedings{ zhong2024algorithmic, title={Algorithmic Capabilities of Random Transformers}, author={Ziqian Zhong and Jacob Andreas}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=plH8gW7tPQ} }

Trained transformer models have been found to implement interpretable procedures for tasks like arithmetic and associative recall, but little is understood about how the circuits that implement these procedures originate during training. To what extent do they depend on the supervisory signal provided to models, and to what extent are they attributable to behavior already present in models at the beginning of training? To investigate these questions, we investigate what functions can be learned by randomly initialized transformers in which only the embedding layers are optimized, so that the only input--output mappings learnable from data are those already implemented (up to a choice of encoding scheme) by the randomly initialized model. We find that these random transformers can perform a wide range of meaningful algorithmic tasks, including modular arithmetic, in-weights and in-context associative recall, decimal addition, parenthesis balancing, and even some aspects of natural language text generation. Our results indicate that some algorithmic capabilities are present in transformers (and accessible via appropriately structured inputs) even before these models are trained.

Algorithmic Capabilities of Random Transformers

NeurIPS.cc/2024/Conference

2410.04368

poster

https://openreview.net/forum?id=pje1Y71jad

@inproceedings{ dong2024costefficient, title={Cost-efficient Knowledge-based Question Answering with Large Language Models}, author={Junnan Dong and Qinggang Zhang and Chuang Zhou and Hao Chen and Daochen Zha and Xiao Huang}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pje1Y71jad} }

Knowledge-based question answering (KBQA) is widely used in many scenarios that necessitate domain knowledge. Large language models (LLMs) bring opportunities to KBQA, while their costs are significantly higher and absence of domain-specific knowledge during pre-training. We are motivated to combine LLMs and prior small models on knowledge graphs (KGMs) for both inferential accuracy and cost saving. However, it remains challenging since accuracy and cost are not readily combined in the optimization as two distinct metrics. It is also laborious for model selection since different models excel in diverse knowledge. To this end, we propose Coke, a novel cost-efficient strategy for KBQA with LLMs, modeled as a tailored multi-armed bandit problem to minimize calls to LLMs within limited budgets. We first formulate the accuracy expectation with a cluster-level Thompson Sampling for either KGMs or LLMs. A context-aware policy is optimized to further distinguish the expert model subject to the question semantics. The overall decision is bounded by the cost regret according to historical expenditure on failures. Extensive experiments showcase the superior performance of Coke, which moves the Pareto frontier with up to 20.89% saving of GPT-4 fees while achieving a 2.74% higher accuracy on the benchmark datasets.

Cost-efficient Knowledge-based Question Answering with Large Language Models

NeurIPS.cc/2024/Conference

2405.17337

poster

https://openreview.net/forum?id=pjD08dtAh0

@inproceedings{ xu2024humanvla, title={Human{VLA}: Towards Vision-Language Directed Object Rearrangement by Physical Humanoid}, author={Xinyu Xu and Yizheng Zhang and Yong-Lu Li and Lei Han and Cewu Lu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pjD08dtAh0} }

Physical Human-Scene Interaction (HSI) plays a crucial role in numerous applications. However, existing HSI techniques are limited to specific object dynamics and privileged information, which prevents the development of more comprehensive applications. To address this limitation, we introduce HumanVLA for general object rearrangement directed by practical vision and language. A teacher-student framework is utilized to develop HumanVLA. A state-based teacher policy is trained first using goal-conditioned reinforcement learning and adversarial motion prior. Then, it is distilled into a vision-language-action model via behavior cloning. We propose several key insights to facilitate the large-scale learning process. To support general object rearrangement by physical humanoid, we introduce a novel Human-in-the-Room dataset encompassing various rearrangement tasks. Through extensive experiments and analysis, we demonstrate the effectiveness of our approach.

HumanVLA: Towards Vision-Language Directed Object Rearrangement by Physical Humanoid

NeurIPS.cc/2024/Conference

2406.19972

poster

https://openreview.net/forum?id=piOzFx9whU

@inproceedings{ ehyaei2024wasserstein, title={Wasserstein Distributionally Robust Optimization through the Lens of Structural Causal Models and Individual Fairness}, author={Ahmad Reza Ehyaei and Golnoosh Farnadi and Samira Samadi}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=piOzFx9whU} }

In recent years, Wasserstein Distributionally Robust Optimization (DRO) has garnered substantial interest for its efficacy in data-driven decision-making under distributional uncertainty. However, limited research has explored the application of DRO to address individual fairness concerns, particularly when considering causal structures and discrete sensitive attributes in learning problems. To address this gap, we first formulate the DRO problem from the perspectives of causality and individual fairness. We then present the DRO dual formulation as an efficient tool to convert the main problem into a more tractable and computationally efficient form. Next, we characterize the closed form of the approximate worst-case loss quantity as a regularizer, eliminating the max-step in the Min-Max DRO problem. We further estimate the regularizer in more general cases and explore the relationship between DRO and classical robust optimization. Finally, by removing the assumption of a known structural causal model, we provide finite sample error bounds when designing DRO with empirical distributions and estimated causal structures to ensure efficiency and robust learning.

Wasserstein Distributionally Robust Optimization through the Lens of Structural Causal Models and Individual Fairness

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pgUQFIJ6BE

@inproceedings{ zhou2024nearoptimal, title={Near-Optimal Distributed Minimax Optimization under the Second-Order Similarity}, author={Qihao Zhou and Haishan Ye and Luo Luo}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pgUQFIJ6BE} }

This paper considers the distributed convex-concave minimax optimization under the second-order similarity. We propose stochastic variance-reduced optimistic gradient sliding (SVOGS) method, which takes the advantage of the finite-sum structure in the objective by involving the mini-batch client sampling and variance reduction. We prove SVOGS can achieve the $\varepsilon$-duality gap within communication rounds of ${\mathcal O}(\delta D^2/\varepsilon)$, communication complexity of ${\mathcal O}(n+\sqrt{n}\delta D^2/\varepsilon)$, and local gradient calls of $\tilde{\mathcal O}(n+(\sqrt{n}\delta+L)D^2/\varepsilon\log(1/\varepsilon))$, where $n$ is the number of nodes, $\delta$ is the degree of the second-order similarity, $L$ is the smoothness parameter and $D$ is the diameter of the constraint set. We can verify that all of above complexity (nearly) matches the corresponding lower bounds. For the specific $\mu$-strongly-convex-$\mu$-strongly-convex case, our algorithm has the upper bounds on communication rounds, communication complexity, and local gradient calls of $\mathcal O(\delta/\mu\log(1/\varepsilon))$, ${\mathcal O}((n+\sqrt{n}\delta/\mu)\log(1/\varepsilon))$, and $\tilde{\mathcal O}(n+(\sqrt{n}\delta+L)/\mu)\log(1/\varepsilon))$ respectively, which are also nearly tight. Furthermore, we conduct the numerical experiments to show the empirical advantages of proposed method.

Near-Optimal Distributed Minimax Optimization under the Second-Order Similarity

NeurIPS.cc/2024/Conference

2405.16126

poster

https://openreview.net/forum?id=pgQCsyKdpN

@inproceedings{ wang2024adaptiveisp, title={Adaptive{ISP}: Learning an Adaptive Image Signal Processor for Object Detection}, author={Yujin Wang and Xu Tian yi and Zhang Fan and Tianfan Xue and Jinwei Gu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pgQCsyKdpN} }

Image Signal Processors (ISPs) convert raw sensor signals into digital images, which significantly influence the image quality and the performance of downstream computer vision tasks. Designing ISP pipeline and tuning ISP parameters are two key steps for building an imaging and vision system. To find optimal ISP configurations, recent works use deep neural networks as a proxy to search for ISP parameters or ISP pipelines. However, these methods are primarily designed to maximize the image quality, which are sub-optimal in the performance of high-level computer vision tasks such as detection, recognition, and tracking. Moreover, after training, the learned ISP pipelines are mostly fixed at the inference time, whose performance degrades in dynamic scenes. To jointly optimize ISP structures and parameters, we propose AdaptiveISP, a task-driven and scene-adaptive ISP. One key observation is that for the majority of input images, only a few processing modules are needed to improve the performance of downstream recognition tasks, and only a few inputs require more processing. Based on this, AdaptiveISP utilizes deep reinforcement learning to automatically generate an optimal ISP pipeline and the associated ISP parameters to maximize the detection performance. Experimental results show that AdaptiveISP not only surpasses the prior state-of-the-art methods for object detection but also dynamically manages the trade-off between detection performance and computational cost, especially suitable for scenes with large dynamic range variations. Project website: https://openimaginglab.github.io/AdaptiveISP/.

AdaptiveISP: Learning an Adaptive Image Signal Processor for Object Detection

NeurIPS.cc/2024/Conference

2410.22939

poster

https://openreview.net/forum?id=pfvcsgFrJ6

@inproceedings{ li2024on, title={On Causal Discovery in the Presence of Deterministic Relations}, author={Loka Li and Haoyue Dai and Hanin Al Ghothani and Biwei Huang and Jiji Zhang and Shahar Harel and Isaac Bentwich and Guangyi Chen and Kun Zhang}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pfvcsgFrJ6} }

Many causal discovery methods typically rely on the assumption of independent noise, yet real-life situations often involve deterministic relationships. In these cases, observed variables are represented as deterministic functions of their parental variables without noise. When determinism is present, constraint-based methods encounter challenges due to the violation of the faithfulness assumption. In this paper, we find, supported by both theoretical analysis and empirical evidence, that score-based methods with exact search can naturally address the issues of deterministic relations under rather mild assumptions. Nonetheless, exact score-based methods can be computationally expensive. To enhance the efficiency and scalability, we develop a novel framework for causal discovery that can detect and handle deterministic relations, called Determinism-aware Greedy Equivalent Search (DGES). DGES comprises three phases: (1) identify minimal deterministic clusters (i.e., a minimal set of variables with deterministic relationships), (2) run modified Greedy Equivalent Search (GES) to obtain an initial graph, and (3) perform exact search exclusively on the deterministic cluster and its neighbors. The proposed DGES accommodates both linear and nonlinear causal relationships, as well as both continuous and discrete data types. Furthermore, we investigate the identifiability conditions of DGES. We conducted extensive experiments on both simulated and real-world datasets to show the efficacy of our proposed method.

On Causal Discovery in the Presence of Deterministic Relations

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pf4OuJyn4Q

@inproceedings{ rafailov2024scaling, title={Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms}, author={Rafael Rafailov and Yaswanth Chittepu and Ryan Park and Harshit Sikchi and Joey Hejna and W. Bradley Knox and Chelsea Finn and Scott Niekum}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pf4OuJyn4Q} }

Reinforcement Learning from Human Feedback (RLHF)has been crucial to the recent success of Large Language Models (LLMs), however it is often a complex and brittle process. In the classical RLHF framework, a reward model is first trained to represent human preferences, which is in turn used by an online reinforcement learning (RL) algorithm to optimized the LLM. A prominent issue with such methods is reward over-optimization or reward hacking, where the performance as measured by the learned proxy reward model increases, but the true model quality plateaus or even deteriorates. Direct Alignment Algorithms (DDAs), such as Direct Preference Optimization (DPO) have emerged as alternatives to the classical RLHF pipeline. However, despite not training a separate proxy reward model or using RL, they still commonly deteriorate from over-optimization. While the so-called reward hacking phenomenon is not well-defined for DAAs, we still uncover similar trends: at higher KL-budgets, DAA algorithms exhibit similar degradation patters to their classic RLHF counterparts. In particular, we find that DAA methods deteriorate not only across a wide range of KL-budgets, but also often before even a single epoch of the dataset is completed. Through extensive empirical experimentation this work formulates the reward over-optimization or hacking problem for DAAs and explores its consequences across objectives, training regimes, and model scales.

Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms

NeurIPS.cc/2024/Conference

2406.02900

https://huggingface.co/papers/2406.02900

poster

https://openreview.net/forum?id=pebP89l4v6

@inproceedings{ ren2024sharing, title={Sharing Key Semantics in Transformer Makes Efficient Image Restoration}, author={Bin Ren and Yawei Li and Jingyun Liang and Rakesh Ranjan and Mengyuan Liu and Rita Cucchiara and Luc Van Gool and Ming-Hsuan Yang and Nicu Sebe}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pebP89l4v6} }

Image Restoration (IR), a classic low-level vision task, has witnessed significant advancements through deep models that effectively model global information. Notably, the emergence of Vision Transformers (ViTs) has further propelled these advancements. When computing, the self-attention mechanism, a cornerstone of ViTs, tends to encompass all global cues, even those from semantically unrelated objects or regions. This inclusivity introduces computational inefficiencies, particularly noticeable with high input resolution, as it requires processing irrelevant information, thereby impeding efficiency. Additionally, for IR, it is commonly noted that small segments of a degraded image, particularly those closely aligned semantically, provide particularly relevant information to aid in the restoration process, as they contribute essential contextual cues crucial for accurate reconstruction. To address these challenges, we propose boosting IR's performance by sharing the key semantics via Transformer for IR (i.e., SemanIR) in this paper. Specifically, SemanIR initially constructs a sparse yet comprehensive key-semantic dictionary within each transformer stage by establishing essential semantic connections for every degraded patch. Subsequently, this dictionary is shared across all subsequent transformer blocks within the same stage. This strategy optimizes attention calculation within each block by focusing exclusively on semantically related components stored in the key-semantic dictionary. As a result, attention calculation achieves linear computational complexity within each window. Extensive experiments across 6 IR tasks confirm the proposed SemanIR's state-of-the-art performance, quantitatively and qualitatively showcasing advancements. The visual results, code, and trained models are available at: https://github.com/Amazingren/SemanIR.

Sharing Key Semantics in Transformer Makes Efficient Image Restoration

NeurIPS.cc/2024/Conference

2405.20008

poster

https://openreview.net/forum?id=pc4GSBi1Hx

@inproceedings{ wu2024lotlip, title={Lo{TLIP}: Improving Language-Image Pre-training for Long Text Understanding}, author={Wei Wu and Kecheng Zheng and Shuailei Ma and Fan Lu and Yuxin Guo and Yifei Zhang and Wei Chen and Qingpei Guo and Yujun Shen and Zheng-Jun Zha}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pc4GSBi1Hx} }

In this work, we empirically confirm that the key reason causing such an issue is that the training images are usually paired with short captions, leaving certain tokens easily overshadowed by salient tokens. Towards this problem, our initial attempt is to relabel the data with long captions, however, directly learning with which may lead to performance degradation in understanding short text (e.g., in the image classification task). Then, after incorporating corner tokens to aggregate diverse textual information, we manage to help the model catch up to its original level of short text understanding yet greatly enhance its capability of long text understanding. We further look into whether the model can continuously benefit from longer captions and notice a clear trade-off between the performance and the efficiency. Finally, we validate the effectiveness of our approach using a self-constructed large-scale dataset, which consists of 100M long caption oriented text-image pairs. It is noteworthy that, on the task of long-text image retrieval, we beat the competitor using long captions with 11.1% improvement (i.e., from 72.62% to 83.72%). The project page is available at https://wuw2019.github.io/lot-lip.

LoTLIP: Improving Language-Image Pre-training for Long Text Understanding

NeurIPS.cc/2024/Conference

2410.05249

poster

https://openreview.net/forum?id=paobkszgIA

@inproceedings{ yang2024endtoend, title={End-to-End Video Semantic Segmentation in Adverse Weather using Fusion Blocks and Temporal-Spatial Teacher-Student Learning}, author={Xin Yang and YAN WENDING and Michael Bi Mi and Yuan Yuan and Robby T. Tan}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=paobkszgIA} }

Adverse weather conditions can significantly degrade the video frames, causing existing video semantic segmentation methods to produce erroneous predictions. In this work, we target adverse weather conditions and introduce an end-to-end domain adaptation strategy that leverages a fusion block, temporal-spatial teacher-student learning, and a temporal weather degradation augmentation approach. The fusion block integrates temporal information from adjacent frames at the feature level, trained end-to-end, eliminating the need for pretrained optical flow, distinguishing our method from existing approaches. Our teacher-student approach involves two teachers: one focuses on exploring temporal information from adjacent frames, and the other harnesses spatial information from the current frame. Finally, we apply temporal weather degradation augmentation to consecutive frames to more accurately represent adverse weather degradations. Our method achieves a performance of 25.4 and 33.0 mIoU on the adaptation from VIPER and Synthia to MVSS, respectively, representing an improvement of 4.3 and 5.8 mIoU over the existing state-of-the-art method.

End-to-End Video Semantic Segmentation in Adverse Weather using Fusion Blocks and Temporal-Spatial Teacher-Student Learning

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=paYwtPBpyZ

@inproceedings{ huguet2024sequenceaugmented, title={Sequence-Augmented {SE}(3)-Flow Matching For Conditional Protein Generation}, author={Guillaume Huguet and James Vuckovic and Kilian FATRAS and Eric Thibodeau-Laufer and Pablo Lemos and Riashat Islam and Cheng-Hao Liu and Jarrid Rector-Brooks and Tara Akhound-Sadegh and Michael M. Bronstein and Alexander Tong and Joey Bose}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=paYwtPBpyZ} }

Proteins are essential for almost all biological processes and derive their diverse functions from complex $3 \rm D$ structures, which are in turn determined by their amino acid sequences. In this paper, we exploit the rich biological inductive bias of amino acid sequences and introduce FoldFlow++, a novel sequence-conditioned $\text{SE}(3)$-equivariant flow matching model for protein structure generation. FoldFlow++ presents substantial new architectural features over the previous FoldFlow family of models including a protein large language model to encode sequence, a new multi-modal fusion trunk that combines structure and sequence representations, and a geometric transformer based decoder. To increase diversity and novelty of generated samples -- crucial for de-novo drug design -- we train FoldFlow++ at scale on a new dataset that is an order of magnitude larger than PDB datasets of prior works, containing both known proteins in PDB and high-quality synthetic structures achieved through filtering. We further demonstrate the ability to align FoldFlow++ to arbitrary rewards, e.g. increasing secondary structures diversity, by introducing a Reinforced Finetuning (ReFT) objective. We empirically observe that FoldFlow++ outperforms previous state-of-the-art protein structure-based generative models, improving over RFDiffusion in terms of unconditional generation across all metrics including designability, diversity, and novelty across all protein lengths, as well as exhibiting generalization on the task of equilibrium conformation sampling. Finally, we demonstrate that a fine-tuned FoldFlow++ makes progress on challenging conditional design tasks such as designing scaffolds for the VHH nanobody.

Sequence-Augmented SE(3)-Flow Matching For Conditional Protein Generation

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pXFiHHySEw

@inproceedings{ hu2024multistage, title={Multi-Stage Predict+Optimize for (Mixed Integer) Linear Programs}, author={Xinyi HU and Jasper C.H. Lee and Jimmy H.M. Lee and Peter J. Stuckey}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pXFiHHySEw} }

The recently-proposed framework of Predict+Optimize tackles optimization problems with parameters that are unknown at solving time, in a supervised learning setting. Prior frameworks consider only the scenario where all unknown parameters are (eventually) revealed simultaneously. In this work, we propose Multi-Stage Predict+Optimize, a novel extension catering to applications where unknown parameters are revealed in sequential stages, with optimization decisions made in between. We further develop three training algorithms for neural networks (NNs) for our framework as proof of concept, both of which handle all mixed integer linear programs. The first baseline algorithm is a natural extension of prior work, training a single NN which makes a single prediction of unknown parameters. The second and third algorithms instead leverage the possibility of updating parameter predictions between stages, and trains one NN per stage. To handle the interdependency between the neural networks, we adopt sequential and parallelized versions of coordinate descent for training. Experimentation on three benchmarks demonstrates the superior learning performance of our methods over classical approaches.

Multi-Stage Predict+Optimize for (Mixed Integer) Linear Programs

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pX71TM2MLh

@inproceedings{ cao2024data, title={Data Free Backdoor Attacks}, author={Bochuan Cao and Jinyuan Jia and Chuxuan Hu and Wenbo Guo and Zhen Xiang and Jinghui Chen and Bo Li and Dawn Song}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pX71TM2MLh} }

Backdoor attacks aim to inject a backdoor into a classifier such that it predicts any input with an attacker-chosen backdoor trigger as an attacker-chosen target class. Existing backdoor attacks require either retraining the classifier with some clean data or modifying the model's architecture. As a result, they are 1) not applicable when clean data is unavailable, 2) less efficient when the model is large, and 3) less stealthy due to architecture changes. In this work, we propose DFBA, a novel retraining-free and data-free backdoor attack without changing the model architecture. Technically, our proposed method modifies a few parameters of a classifier to inject a backdoor. Through theoretical analysis, we verify that our injected backdoor is provably undetectable and unremovable by various state-of-the-art defenses under mild assumptions. Our evaluation on multiple datasets further demonstrates that our injected backdoor: 1) incurs negligible classification loss, 2) achieves 100\% attack success rates, and 3) bypasses six existing state-of-the-art defenses. Moreover, our comparison with a state-of-the-art non-data-free backdoor attack shows our attack is more stealthy and effective against various defenses while achieving less classification accuracy loss. We will release our code upon paper acceptance.

Data Free Backdoor Attacks

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pWowK7jqok

@inproceedings{ wu2024emotion, title={E-Motion: Future Motion Simulation via Event Sequence Diffusion}, author={Song Wu and Zhiyu Zhu and Junhui Hou and Guangming Shi and Jinjian Wu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pWowK7jqok} }

Forecasting a typical object's future motion is a critical task for interpreting and interacting with dynamic environments in computer vision. Event-based sensors, which could capture changes in the scene with exceptional temporal granularity, may potentially offer a unique opportunity to predict future motion with a level of detail and precision previously unachievable. Inspired by that, we propose to integrate the strong learning capacity of the video diffusion model with the rich motion information of an event camera as a motion simulation framework. Specifically, we initially employ pre-trained stable video diffusion models to adapt the event sequence dataset. This process facilitates the transfer of extensive knowledge from RGB videos to an event-centric domain. Moreover, we introduce an alignment mechanism that utilizes reinforcement learning techniques to enhance the reverse generation trajectory of the diffusion model, ensuring improved performance and accuracy. Through extensive testing and validation, we demonstrate the effectiveness of our method in various complex scenarios, showcasing its potential to revolutionize motion flow prediction in computer vision applications such as autonomous vehicle guidance, robotic navigation, and interactive media. Our findings suggest a promising direction for future research in enhancing the interpretative power and predictive accuracy of computer vision systems. The source code is publicly available at https://github.com/p4r4mount/E-Motion.

E-Motion: Future Motion Simulation via Event Sequence Diffusion

NeurIPS.cc/2024/Conference

2410.08649

poster

https://openreview.net/forum?id=pW9Jwim918

@inproceedings{ lee2024remodetect, title={ReMoDetect: Reward Models Recognize Aligned {LLM}'s Generations}, author={Hyunseok Lee and Jihoon Tack and Jinwoo Shin}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pW9Jwim918} }

The remarkable capabilities and easy accessibility of large language models (LLMs) have significantly increased societal risks (e.g., fake news generation), necessitating the development of LLM-generated text (LGT) detection methods for safe usage. However, detecting LGTs is challenging due to the vast number of LLMs, making it impractical to account for each LLM individually; hence, it is crucial to identify the common characteristics shared by these models. In this paper, we draw attention to a common feature of recent powerful LLMs, namely the alignment training, i.e., training LLMs to generate human-preferable texts. Our key finding is that as these aligned LLMs are trained to maximize the human preferences, they generate texts with higher estimated preferences even than human-written texts; thus, such texts are easily detected by using the reward model (i.e., an LLM trained to model human preference distribution). Based on this finding, we propose two training schemes to further improve the detection ability of the reward model, namely (i) continual preference fine-tuning to make reward model prefer aligned LGTs even further and (ii) reward modeling of Human/LLM mixed texts (a rephrased texts from human-written texts using aligned LLMs), which serves as a median preference text corpus between LGTs and human-written texts to learn the decision boundary better. We provide an extensive evaluation by considering six text domains across twelve aligned LLMs, where our method demonstrates state-of-the-art results.

ReMoDetect: Reward Models Recognize Aligned LLM's Generations

NeurIPS.cc/2024/Conference

2405.17382

https://huggingface.co/papers/2405.17382

poster

https://openreview.net/forum?id=pVPyCgXv57

@inproceedings{ lee2024learning, title={Learning to Merge Tokens via Decoupled Embedding for Efficient Vision Transformers}, author={Dong Hoon Lee and Seunghoon Hong}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pVPyCgXv57} }

Recent token reduction methods for Vision Transformers (ViTs) incorporate token merging, which measures the similarities between token embeddings and combines the most similar pairs. However, their merging policies are directly dependent on intermediate features in ViTs, which prevents exploiting features tailored for merging and requires end-to-end training to improve token merging. In this paper, we propose Decoupled Token Embedding for Merging (DTEM) that enhances token merging through a decoupled embedding learned via a continuously relaxed token merging process. Our method introduces a lightweight embedding module decoupled from the ViT forward pass to extract dedicated features for token merging, thereby addressing the restriction from using intermediate features. The continuously relaxed token merging, applied during training, enables us to learn the decoupled embeddings in a differentiable manner. Thanks to the decoupled structure, our method can be seamlessly integrated into existing ViT backbones and trained either modularly by learning only the decoupled embeddings or end-to-end by fine-tuning. We demonstrate the applicability of DTEM on various tasks, including classification, captioning, and segmentation, with consistent improvement in token merging. Especially in the ImageNet-1k classification, DTEM achieves a 37.2\% reduction in FLOPs while maintaining a top-1 accuracy of 79.85\% with DeiT-small.

Learning to Merge Tokens via Decoupled Embedding for Efficient Vision Transformers

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pU0z2sNM1M

@inproceedings{ loftus2024causal, title={Causal Dependence Plots}, author={Joshua R. Loftus and Lucius E.J. Bynum and Sakina Hansen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pU0z2sNM1M} }

To use artificial intelligence and machine learning models wisely we must understand how they interact with the world, including how they depend causally on data inputs. In this work we develop Causal Dependence Plots (CDPs) to visualize how a model's predicted outcome depends on changes in a given predictor *along with consequent causal changes in other predictor variables*. Crucially, this differs from standard methods based on independence or holding other predictors constant, such as regression coefficients or Partial Dependence Plots (PDPs). Our explanatory framework generalizes PDPs, including them as a special case, as well as a variety of other interpretive plots that show, for example, the total, direct, and indirect effects of causal mediation. We demonstrate with simulations and real data experiments how CDPs can be combined in a modular way with methods for causal learning or sensitivity analysis. Since people often think causally about input-output dependence, CDPs can be powerful tools in the xAI or interpretable machine learning toolkit and contribute to applications like scientific machine learning and algorithmic fairness.

Causal Dependence Plots

NeurIPS.cc/2024/Conference

2303.04209

poster

https://openreview.net/forum?id=pRSgf5VdD0

@inproceedings{ buchholz2024learning, title={Learning Partitions from Context}, author={Simon Buchholz}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pRSgf5VdD0} }

In this paper, we study the problem of learning the structure of a discrete set of $N$ tokens based on their interactions with other tokens. We focus on a setting where the tokens can be partitioned into a small number of classes, and there exists a real-valued function $f$ defined on certain sets of tokens. This function, which captures the interactions between tokens, depends only on the class memberships of its arguments. The goal is to recover the class memberships of all tokens from a finite number of samples of $f$. We begin by analyzing this problem from both complexity-theoretic and information-theoretic viewpoints. We prove that it is NP-complete in general, and for random instances, we show that on the order of $N\ln(N)$ samples, implying very sparse interactions, suffice to identify the partition. We then investigate the conditions under which gradient flow dynamics of token embeddings can reveal the class structure, finding that this is achievable in certain settings when given on the order of $N^2\ln^2(N)$ samples.

Learning Partitions from Context

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pRQmRaonxf

@inproceedings{ shi2024transformers, title={Transformers as Game Players: Provable In-context Game-playing Capabilities of Pre-trained Models}, author={Chengshuai Shi and Kun Yang and Jing Yang and Cong Shen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pRQmRaonxf} }

The in-context learning (ICL) capability of pre-trained models based on the transformer architecture has received growing interest in recent years. While theoretical understanding has been obtained for ICL in reinforcement learning (RL), the previous results are largely confined to the single-agent setting. This work proposes to further explore the in-context learning capabilities of pre-trained transformer models in competitive multi-agent games, i.e., in-context game-playing (ICGP). Focusing on the classical two-player zero-sum games, theoretical guarantees are provided to demonstrate that pre-trained transformers can provably learn to approximate Nash equilibrium in an in-context manner for both decentralized and centralized learning settings. As a key part of the proof, constructional results are established to demonstrate that the transformer architecture is sufficiently rich to realize celebrated multi-agent game-playing algorithms, in particular, decentralized V-learning and centralized VI-ULCB.

Transformers as Game Players: Provable In-context Game-playing Capabilities of Pre-trained Models

NeurIPS.cc/2024/Conference

2410.09701

poster

https://openreview.net/forum?id=pR5g1bBqoV

@inproceedings{ haas2024boldsymbolmumathbfp, title={\${\textbackslash}boldsymbol\{{\textbackslash}mu\}{\textbackslash}mathbf\{P{\textasciicircum}2\}\$: Effective Sharpness Aware Minimization Requires Layerwise Perturbation Scaling}, author={Moritz Haas and Jin Xu and Volkan Cevher and Leena Chennuru Vankadara}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pR5g1bBqoV} }

Sharpness Aware Minimization (SAM) enhances performance across various neural architectures and datasets. As models are continually scaled up to improve performance, a rigorous understanding of SAM’s scaling behaviour is paramount. To this end, we study the infinite-width limit of neural networks trained with SAM, using the Tensor Programs framework. Our findings reveal that the dynamics of standard SAM effectively reduce to applying SAM solely in the last layer in wide neural networks, even with optimal hyperparameters. In contrast, we identify a stable parameterization with layerwise perturbation scaling, which we call *Maximal Update and Perturbation Parameterization* ($\mu$P$^2$), that ensures all layers are both feature learning and effectively perturbed in the limit. Through experiments with MLPs, ResNets and Vision Transformers, we empirically demonstrate that $\mu$P$^2$ is the first parameterization to achieve hyperparameter transfer of the joint optimum of learning rate and perturbation radius across model scales. Moreover, we provide an intuitive condition to derive $\mu$P$^2$ for other perturbation rules like Adaptive SAM and SAM-ON, also ensuring balanced perturbation effects across all layers.

μ𝐏^2: Effective Sharpness Aware Minimization Requires Layerwise Perturbation Scaling

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pR37AmwbOt

@inproceedings{ pan2024leveraging, title={Leveraging Catastrophic Forgetting to Develop Safe Diffusion Models against Malicious Finetuning}, author={Jiadong Pan and Hongcheng Gao and Zongyu Wu and taihang Hu and Li Su and Qingming Huang and Liang Li}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pR37AmwbOt} }

Diffusion models (DMs) have demonstrated remarkable proficiency in producing images based on textual prompts. Numerous methods have been proposed to ensure these models generate safe images. Early methods attempt to incorporate safety filters into models to mitigate the risk of generating harmful images but such external filters do not inherently detoxify the model and can be easily bypassed. Hence, model unlearning and data cleaning are the most essential methods for maintaining the safety of models, given their impact on model parameters. However, malicious fine-tuning can still make models prone to generating harmful or undesirable images even with these methods. Inspired by the phenomenon of catastrophic forgetting, we propose a training policy using contrastive learning to increase the latent space distance between clean and harmful data distribution, thereby protecting models from being fine-tuned to generate harmful images due to forgetting. The experimental results demonstrate that our methods not only maintain clean image generation capabilities before malicious fine-tuning but also effectively prevent DMs from producing harmful images after malicious fine-tuning. Our method can also be combined with other safety methods to maintain their safety against malicious fine-tuning further.

Leveraging Catastrophic Forgetting to Develop Safe Diffusion Models against Malicious Finetuning

NeurIPS.cc/2024/Conference

oral

https://openreview.net/forum?id=pPeXYByHNd

@inproceedings{ chen2024msagpt, title={{MSAGPT}: Neural Prompting Protein Structure Prediction via {MSA} Generative Pre-Training}, author={Bo Chen and Zhilei Bei and Xingyi Cheng and Pan Li and Jie Tang and Le Song}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pPeXYByHNd} }

Multiple Sequence Alignment (MSA) plays a pivotal role in unveiling the evolutionary trajectories of protein families. The accuracy of protein structure predictions is often compromised for protein sequences that lack sufficient homologous information to construct high-quality MSA. Although various methods have been proposed to generate high-quality MSA under these conditions, they fall short in comprehensively capturing the intricate co-evolutionary patterns within MSA or require guidance from external oracle models. Here we introduce MSAGPT, a novel approach to prompt protein structure predictions via MSA generative pre-training in a low-MSA regime. MSAGPT employs a simple yet effective 2D evolutionary positional encoding scheme to model the complex evolutionary patterns. Endowed by this, the flexible 1D MSA decoding framework facilitates zero- or few-shot learning. Moreover, we demonstrate leveraging the feedback from AlphaFold2 (AF2) can further enhance the model’s capacity via Rejective Fine-tuning (RFT) and Reinforcement Learning from AF2 Feedback (RLAF). Extensive experiments confirm the efficacy of MSAGPT in generating faithful and informative MSA (up to +8.5% TM-Score on few-shot scenarios). The transfer learning also demonstrates its great potential for the wide range of tasks resorting to the quality of MSA.

MSAGPT: Neural Prompting Protein Structure Prediction via MSA Generative Pre-Training

NeurIPS.cc/2024/Conference

2406.05347

https://huggingface.co/papers/2406.05347

poster

https://openreview.net/forum?id=pPSWHsgqRp

@inproceedings{ guha2024smoothie, title={Smoothie: Label Free Language Model Routing}, author={Neel Guha and Mayee F Chen and Trevor Chow and Ishan S. Khare and Christopher Re}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pPSWHsgqRp} }

Large language models (LLMs) are increasingly used in applications where LLM inputs may span many different tasks. Recent work has found that the choice of LLM is consequential, and different LLMs may be good for different input samples. Prior approaches have thus explored how engineers might select an LLM to use for each sample (i.e. _routing_). While existing routing methods mostly require training auxiliary models on human-annotated data, our work explores whether it is possible to perform _unsupervised_ routing. We propose Smoothie, a weak supervision-inspired routing approach that requires no labeled data. Given a set of outputs from different LLMs, Smoothie constructs a latent variable graphical model over embedding representations of observable LLM outputs and unknown “true” outputs. Using this graphical model, we estimate sample-dependent quality scores for each LLM, and route each sample to the LLM with the highest corresponding score. We find that Smoothie's LLM quality-scores correlate with ground-truth model quality (correctly identifying the optimal model on 9/14 tasks), and that Smoothie outperforms baselines for routing by up to 10 points accuracy.

Smoothie: Label Free Language Model Routing

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pOXgdFEB7q

@inproceedings{ harun2024what, title={What Variables Affect Out-of-Distribution Generalization in Pretrained Models?}, author={Md Yousuf Harun and Kyungbok Lee and Jhair Gallardo and Giri Prashanth and Christopher Kanan}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pOXgdFEB7q} }

Embeddings produced by pre-trained deep neural networks (DNNs) are widely used; however, their efficacy for downstream tasks can vary widely. We study the factors influencing transferability and out-of-distribution (OOD) generalization of pre-trained DNN embeddings through the lens of the tunnel effect hypothesis, which is closely related to intermediate neural collapse. This hypothesis suggests that deeper DNN layers compress representations and hinder OOD generalization. Contrary to earlier work, our experiments show this is not a universal phenomenon. We comprehensively investigate the impact of DNN architecture, training data, image resolution, and augmentations on transferability. We identify that training with high-resolution datasets containing many classes greatly reduces representation compression and improves transferability. Our results emphasize the danger of generalizing findings from toy datasets to broader contexts.

What Variables Affect Out-of-Distribution Generalization in Pretrained Models?

NeurIPS.cc/2024/Conference

2405.15018

poster

https://openreview.net/forum?id=pNnvzQsS4P

@inproceedings{ zhang2024kv, title={{KV} Cache is 1 Bit Per Channel: Efficient Large Language Model Inference with Coupled Quantization}, author={Tianyi Zhang and Jonah Wonkyu Yi and Zhaozhuo Xu and Anshumali Shrivastava}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pNnvzQsS4P} }

Efficient deployment of Large Language Models (LLMs) requires batching multiple requests together to improve throughput. As batch size, context length, or model size increases, the size of key and value (KV) cache quickly becomes the main contributor to GPU memory usage and the bottleneck of inference latency and throughput. Quantization has emerged as an effective technique for KV cache compression, but existing methods still fail at very low bit widths. Currently, KV cache quantization is performed per-channel or per-token independently. Our analysis shows that distinct channels of a key/value activation embedding are highly interdependent, and the joint entropy of multiple channels grows at a slower rate than the sum of their marginal entropy, which implies that per-channel independent quantization is sub-optimal. To mitigate this sub-optimality, we propose Coupled Quantization (CQ), which couples multiple key/value channels together for quantization to exploit their interdependence and encode the activations in a more information-efficient manner. Extensive experiments reveal that CQ compares favorably with existing baselines in preserving model quality, and improves inference throughput by 1.4–3.5$\times$ relative to the uncompressed baseline. Furthermore, we demonstrate that CQ can preserve model quality reasonably with KV cache quantized down to 1 bit.

KV Cache is 1 Bit Per Channel: Efficient Large Language Model Inference with Coupled Quantization

NeurIPS.cc/2024/Conference

2405.03917

https://huggingface.co/papers/2405.03917

poster

https://openreview.net/forum?id=pMaCRgu8GV

@inproceedings{ cook2024artificial, title={Artificial Generational Intelligence: Cultural Accumulation in Reinforcement Learning}, author={Jonathan Cook and Chris Lu and Edward Hughes and Joel Z Leibo and Jakob Nicolaus Foerster}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pMaCRgu8GV} }

Cultural accumulation drives the open-ended and diverse progress in capabilities spanning human history. It builds an expanding body of knowledge and skills by combining individual exploration with inter-generational information transmission. Despite its widespread success among humans, the capacity for artificial learning agents to accumulate culture remains under-explored. In particular, approaches to reinforcement learning typically strive for improvements over only a single lifetime. Generational algorithms that do exist fail to capture the open-ended, emergent nature of cultural accumulation, which allows individuals to trade-off innovation and imitation. Building on the previously demonstrated ability for reinforcement learning agents to perform social learning, we find that training setups which balance this with independent learning give rise to cultural accumulation. These accumulating agents outperform those trained for a single lifetime with the same cumulative experience. We explore this accumulation by constructing two models under two distinct notions of a generation: episodic generations, in which accumulation occurs via in-context learning and train-time generations, in which accumulation occurs via in-weights learning. In-context and in-weights cultural accumulation can be interpreted as analogous to knowledge and skill accumulation, respectively. To the best of our knowledge, this work is the first to present general models that achieve emergent cultural accumulation in reinforcement learning, opening up new avenues towards more open-ended learning systems, as well as presenting new opportunities for modelling human culture.

Artificial Generational Intelligence: Cultural Accumulation in Reinforcement Learning

NeurIPS.cc/2024/Conference

2406.00392

https://huggingface.co/papers/2406.00392

poster

https://openreview.net/forum?id=pMPBxMf8T3

@inproceedings{ xu2024the, title={The Implicit Bias of Heterogeneity towards Invariance: A Study of Multi-Environment Matrix Sensing}, author={Yang Xu and Yihong Gu and Cong Fang}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pMPBxMf8T3} }

Models are expected to engage in invariance learning, which involves distinguishing the core relations that remain consistent across varying environments to ensure the predictions are safe, robust and fair. While existing works consider specific algorithms to realize invariance learning, we show that model has the potential to learn invariance through standard training procedures. In other words, this paper studies the implicit bias of Stochastic Gradient Descent (SGD) over heterogeneous data and shows that the implicit bias drives the model learning towards an invariant solution. We call the phenomenon the implicit invariance learning. Specifically, we theoretically investigate the multi-environment low-rank matrix sensing problem where in each environment, the signal comprises (i) a lower-rank invariant part shared across all environments; and (ii) a significantly varying environment-dependent spurious component. The key insight is, through simply employing the large step size large-batch SGD sequentially in each environment without any explicit regularization, the oscillation caused by heterogeneity can provably prevent model learning spurious signals. The model reaches the invariant solution after certain iterations. In contrast, model learned using pooled SGD over all data would simultaneously learn both the invariant and spurious signals. Overall, we unveil another implicit bias that is a result of the symbiosis between the heterogeneity of data and modern algorithms, which is, to the best of our knowledge, first in the literature.

The Implicit Bias of Heterogeneity towards Invariance: A Study of Multi-Environment Matrix Sensing

NeurIPS.cc/2024/Conference

2403.01420

poster

https://openreview.net/forum?id=pMJFaBzoG3

@inproceedings{ guo2024otp, title={{OT}4P: Unlocking Effective Orthogonal Group Path for Permutation Relaxation}, author={Yaming Guo and Chen Zhu and Hengshu Zhu and Tieru Wu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pMJFaBzoG3} }

Optimization over permutations is typically an NP-hard problem that arises extensively in ranking, matching, tracking, etc. Birkhoff polytope-based relaxation methods have made significant advancements, particularly in penalty-free optimization and probabilistic inference. Relaxation onto the orthogonal group offers unique potential advantages such as a lower representation dimension and preservation of inner products; however, equally effective approaches remain unexplored. To bridge the gap, we present a temperature-controlled differentiable transformation that maps unconstrained vector space to the orthogonal group, where the temperature, in the limit, concentrates orthogonal matrices near permutation matrices. This transformation naturally implements a parameterization for the relaxation of permutation matrices, allowing for gradient-based optimization of problems involving permutations. Additionally, by deriving a re-parameterized gradient estimator, this transformation also provides efficient stochastic optimization over the latent permutations. Extensive experiments involving the optimization over permutation matrices validate the effectiveness of the proposed method.

OT4P: Unlocking Effective Orthogonal Group Path for Permutation Relaxation

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pLoX8Og3bH

@inproceedings{ ji2024unleashing, title={Unleashing Multispectral Video's Potential in Semantic Segmentation: A Semi-supervised Viewpoint and New {UAV}-View Benchmark}, author={Wei Ji and Jingjing Li and Wenbo Li and Yilin Shen and Li cheng and Hongxia Jin}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pLoX8Og3bH} }

Thanks to the rapid progress in RGB & thermal imaging, also known as multispectral imaging, the task of multispectral video semantic segmentation, or MVSS in short, has recently drawn significant attentions. Noticeably, it offers new opportunities in improving segmentation performance under unfavorable visual conditions such as poor light or overexposure. Unfortunately, there are currently very few datasets available, including for example MVSeg dataset that focuses purely toward eye-level view; and it features the sparse annotation nature due to the intensive demands of labeling process. To address these key challenges of the MVSS task, this paper presents two major contributions: the introduction of MVUAV, a new MVSS benchmark dataset, and the development of a dedicated semi-supervised MVSS baseline - SemiMV. Our MVUAV dataset is captured via Unmanned Aerial Vehicles (UAV), which offers a unique oblique bird’s-eye view complementary to the existing MVSS datasets; it also encompasses a broad range of day/night lighting conditions and over 30 semantic categories. In the meantime, to better leverage the sparse annotations and extra unlabeled RGB-Thermal videos, a semi-supervised learning baseline, SemiMV, is proposed to enforce consistency regularization through a dedicated Cross-collaborative Consistency Learning (C3L) module and a denoised temporal aggregation strategy. Comprehensive empirical evaluations on both MVSeg and MVUAV benchmark datasets have showcased the efficacy of our SemiMV baseline.

Unleashing Multispectral Video's Potential in Semantic Segmentation: A Semi-supervised Viewpoint and New UAV-View Benchmark

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pK2qGRY2Hv

@inproceedings{ sam2024the, title={The Limits of Transfer Reinforcement Learning with Latent Low-rank Structure}, author={Tyler Sam and Yudong Chen and Christina Yu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pK2qGRY2Hv} }

Many reinforcement learning (RL) algorithms are too costly to use in practice due to the large sizes $S,A$ of the problem's state and action space. To resolve this issue, we study transfer RL with latent low rank structure. We consider the problem of transferring a latent low rank representation when the source and target MDPs have transition kernels with Tucker rank $(S, d, A)$, $(S ,S , d), (d, S , A )$, or $(d , d , d )$. In each setting, we introduce the transfer-ability coefficient $\alpha$ that measures the difficulty of representational transfer. Our algorithm learns latent representations in each source MDP and then exploits the linear structure to remove the dependence on $S , A $, or $SA $ in the target MDP regret bound. We complement our positive results with information theoretic lower bounds that show our algorithms (excluding the ($d, d, d$) setting) are minimax-optimal with respect to $\alpha$.

The Limits of Transfer Reinforcement Learning with Latent Low-rank Structure

NeurIPS.cc/2024/Conference

2410.21601

poster

https://openreview.net/forum?id=pJlFURyTG5

@inproceedings{ zhang2024scalable, title={Scalable Constrained Policy Optimization for Safe Multi-agent Reinforcement Learning}, author={Lijun Zhang and Lin Li and Wei Wei and Huizhong Song and Yaodong Yang and Jiye Liang}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pJlFURyTG5} }

A challenging problem in seeking to bring multi-agent reinforcement learning (MARL) techniques into real-world applications, such as autonomous driving and drone swarms, is how to control multiple agents safely and cooperatively to accomplish tasks. Most existing safe MARL methods learn the centralized value function by introducing a global state to guide safety cooperation. However, the global coupling arising from agents’ safety constraints and the exponential growth of the state-action space size limit their applicability in instant communication or computing resource-constrained systems and larger multi-agent systems. In this paper, we develop a novel scalable and theoretically-justified multi-agent constrained policy optimization method. This method utilizes the rigorous bounds of the trust region method and the bounds of the truncated advantage function to provide a new local policy optimization objective for each agent. Also, we prove that the safety constraints and the joint policy improvement can be met when each agent adopts a sequential update scheme to optimize a $\kappa$-hop policy. Then, we propose a practical algorithm called Scalable MAPPO-Lagrangian (Scal-MAPPO-L). The proposed method’s effectiveness is verified on a collection of benchmark tasks, and the results support our theory that decentralized training with local interactions can still improve reward performance and satisfy safe constraints.

Scalable Constrained Policy Optimization for Safe Multi-agent Reinforcement Learning

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pHiTmEsAfZ

@inproceedings{ liang2024makeanagent, title={Make-An-Agent: A Generalizable Policy Network Generator with Behavior-Prompted Diffusion}, author={Yongyuan Liang and Tingqiang Xu and Kaizhe Hu and Guangqi Jiang and Furong Huang and Huazhe Xu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pHiTmEsAfZ} }

Can we generate a control policy for an agent using just one demonstration of desired behaviors as a prompt, as effortlessly as creating an image from a textual description? In this paper, we present **Make-An-Agent**, a novel policy parameter generator that leverages the power of conditional diffusion models for behavior-to-policy generation. Guided by behavior embeddings that encode trajectory information, our policy generator synthesizes latent parameter representations, which can then be decoded into policy networks. Trained on policy network checkpoints and their corresponding trajectories, our generation model demonstrates remarkable versatility and scalability on multiple tasks and has a strong generalization ability on unseen tasks to output well-performed policies with only few-shot demonstrations as inputs. We showcase its efficacy and efficiency on various domains and tasks, including varying objectives, behaviors, and even across different robot manipulators. Beyond simulation, we directly deploy policies generated by **Make-An-Agent** onto real-world robots on locomotion tasks. Project page: https://cheryyunl.github.io/make-an-agent/.

Make-An-Agent: A Generalizable Policy Network Generator with Behavior-Prompted Diffusion

NeurIPS.cc/2024/Conference

2407.10973

https://huggingface.co/papers/2407.10973

poster

https://openreview.net/forum?id=pH3XAQME6c

@inproceedings{ arditi2024refusal, title={Refusal in Language Models Is Mediated by a Single Direction}, author={Andy Arditi and Oscar Balcells Obeso and Aaquib Syed and Daniel Paleka and Nina Rimsky and Wes Gurnee and Neel Nanda}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pH3XAQME6c} }

Conversational large language models are fine-tuned for both instruction-following and safety, resulting in models that obey benign requests but refuse harmful ones. While this refusal behavior is widespread across chat models, its underlying mechanisms remain poorly understood. In this work, we show that refusal is mediated by a one-dimensional subspace, across 13 popular open-source chat models up to 72B parameters in size. Specifically, for each model, we find a single direction such that erasing this direction from the model's residual stream activations prevents it from refusing harmful instructions, while adding this direction elicits refusal on even harmless instructions. Leveraging this insight, we propose a novel white-box jailbreak method that surgically disables a model's ability to refuse, with minimal effect on other capabilities. This interpretable rank-one weight edit results in an effective jailbreak technique that is simpler and more efficient than fine-tuning. Finally, we mechanistically analyze how adversarial suffixes suppress propagation of the refusal-mediating direction. Our findings underscore the brittleness of current safety fine-tuning methods. More broadly, our work showcases how an understanding of model internals can be leveraged to develop practical methods for controlling model behavior.

Refusal in Language Models Is Mediated by a Single Direction

NeurIPS.cc/2024/Conference

2406.11717

https://huggingface.co/papers/2406.11717

poster

https://openreview.net/forum?id=pGeAcYhnN5

@inproceedings{ wen2024speculative, title={Speculative Decoding with {CTC}-based Draft Model for {LLM} Inference Acceleration}, author={Zhuofan Wen and Shangtong Gui and Yang Feng}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pGeAcYhnN5} }

Inference acceleration of large language models (LLMs) has been put forward in many application scenarios and speculative decoding has shown its advantage in addressing inference acceleration. Speculative decoding usually introduces a draft model to assist the base LLM where the draft model produces drafts and the base LLM verifies the draft for acceptance or rejection. In this framework, the final inference speed is decided by the decoding speed of the draft model and the acceptance rate of the draft provided by the draft model. Currently the widely used draft models usually generate draft tokens for the next several positions in a non-autoregressive way without considering the correlations between draft tokens. Therefore, it has a high decoding speed but an unsatisfactory acceptance rate. In this paper, we focus on how to improve the performance of the draft model and aim to accelerate inference via a high acceptance rate. To this end, we propose a CTC-based draft model which strengthens the correlations between draft tokens during the draft phase, thereby generating higher-quality draft candidate sequences. Experiment results show that compared to strong baselines, the proposed method can achieve a higher acceptance rate and hence a faster inference speed.

Speculative Decoding with CTC-based Draft Model for LLM Inference Acceleration

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pGR5X4e1gy

@inproceedings{ finkelshtein2024learning, title={Learning on Large Graphs using Intersecting Communities}, author={Ben Finkelshtein and Ismail Ilkan Ceylan and Michael M. Bronstein and Ron Levie}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pGR5X4e1gy} }

Message Passing Neural Networks (MPNNs) are a staple of graph machine learning. MPNNs iteratively update each node’s representation in an input graph by aggregating messages from the node’s neighbors, which necessitates a memory complexity of the order of the __number of graph edges__. This complexity might quickly become prohibitive for large graphs provided they are not very sparse. In this paper, we propose a novel approach to alleviate this problem by approximating the input graph as an intersecting community graph (ICG) -- a combination of intersecting cliques. The key insight is that the number of communities required to approximate a graph __does not depend on the graph size__. We develop a new constructive version of the Weak Graph Regularity Lemma to efficiently construct an approximating ICG for any input graph. We then devise an efficient graph learning algorithm operating directly on ICG in linear memory and time with respect to the __number of nodes__ (rather than edges). This offers a new and fundamentally different pipeline for learning on very large non-sparse graphs, whose applicability is demonstrated empirically on node classification tasks and spatio-temporal data processing.

Learning on Large Graphs using Intersecting Communities

NeurIPS.cc/2024/Conference

2405.20724

poster

https://openreview.net/forum?id=pGOBEYcXzs

@inproceedings{ jo2024mixture, title={Mixture of Scales: Memory-Efficient Token-Adaptive Binarization for Large Language Models}, author={Dongwon Jo and Taesu Kim and Yulhwa Kim and Jae-Joon Kim}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pGOBEYcXzs} }

Binarization, which converts weight parameters to binary values, has emerged as an effective strategy to reduce the size of large language models (LLMs). However, typical binarization techniques significantly diminish linguistic effectiveness of LLMs. To address this issue, we introduce a novel binarization technique called Mixture of Scales (BinaryMoS). Unlike conventional methods, BinaryMoS employs multiple scaling experts for binary weights, dynamically merging these experts for each token to adaptively generate scaling factors. This token-adaptive approach boosts the representational power of binarized LLMs by enabling contextual adjustments to the values of binary weights. Moreover, because this adaptive process only involves the scaling factors rather than the entire weight matrix, BinaryMoS maintains compression efficiency similar to traditional static binarization methods. Our experimental results reveal that BinaryMoS surpasses conventional binarization techniques in various natural language processing tasks and even outperforms 2-bit quantization methods, all while maintaining similar model size to static binarization techniques.

Mixture of Scales: Memory-Efficient Token-Adaptive Binarization for Large Language Models

NeurIPS.cc/2024/Conference

2406.12311

https://huggingface.co/papers/2406.12311

poster

https://openreview.net/forum?id=pGEY8JQ3qx

@inproceedings{ zurek2024spanbased, title={Span-Based Optimal Sample Complexity for Weakly Communicating and General Average Reward {MDP}s}, author={Matthew Zurek and Yudong Chen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pGEY8JQ3qx} }

We study the sample complexity of learning an $\varepsilon$-optimal policy in an average-reward Markov decision process (MDP) under a generative model. For weakly communicating MDPs, we establish the complexity bound $\widetilde{O}\left(SA\frac{\mathsf{H}}{\varepsilon^2} \right)$, where $\mathsf{H}$ is the span of the bias function of the optimal policy and $SA$ is the cardinality of the state-action space. Our result is the first that is minimax optimal (up to log factors) in all parameters $S,A,\mathsf{H}$, and $\varepsilon$, improving on existing work that either assumes uniformly bounded mixing times for all policies or has suboptimal dependence on the parameters. We also initiate the study of sample complexity in general (multichain) average-reward MDPs. We argue a new transient time parameter $\mathsf{B}$ is necessary, establish an $\widetilde{O}\left(SA\frac{\mathsf{B} + \mathsf{H}}{\varepsilon^2} \right)$ complexity bound, and prove a matching (up to log factors) minimax lower bound. Both results are based on reducing the average-reward MDP to a discounted MDP, which requires new ideas in the general setting. To optimally analyze this reduction, we develop improved bounds for $\gamma$-discounted MDPs, showing that $\widetilde{O}\left(SA\frac{\mathsf{H}}{(1-\gamma)^2\varepsilon^2} \right)$ and $\widetilde{O}\left(SA\frac{\mathsf{B} + \mathsf{H}}{(1-\gamma)^2\varepsilon^2} \right)$ samples suffice to learn $\varepsilon$-optimal policies in weakly communicating and in general MDPs, respectively. Both these results circumvent the well-known minimax lower bound of $\widetilde{\Omega}\left(SA\frac{1}{(1-\gamma)^3\varepsilon^2} \right)$ for $\gamma$-discounted MDPs, and establish a quadratic rather than cubic horizon dependence for a fixed MDP instance.

Span-Based Optimal Sample Complexity for Weakly Communicating and General Average Reward MDPs

NeurIPS.cc/2024/Conference

2403.11477

oral

https://openreview.net/forum?id=pG380vLYRU

@inproceedings{ lin2024faster, title={Faster Accelerated First-order Methods for Convex Optimization with Strongly Convex Function Constraints}, author={Zhenwei Lin and Qi Deng}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pG380vLYRU} }

In this paper, we introduce faster accelerated primal-dual algorithms for minimizing a convex function subject to strongly convex function constraints. Prior to our work, the best complexity bound was $\mathcal{O}(1/{\varepsilon})$, regardless of the strong convexity of the constraint function. It is unclear whether the strong convexity assumption can enable even better convergence results. To address this issue, we have developed novel techniques to progressively estimate the strong convexity of the Lagrangian function. Our approach, for the first time, effectively leverages the constraint strong convexity, obtaining an improved complexity of $\mathcal{O}(1/\sqrt{\varepsilon})$. This rate matches the complexity lower bound for strongly-convex-concave saddle point optimization and is therefore order-optimal. We show the superior performance of our methods in sparsity-inducing constrained optimization, notably Google's personalized PageRank problem. Furthermore, we show that a restarted version of the proposed methods can effectively identify the optimal solution's sparsity pattern within a finite number of steps, a result that appears to have independent significance.

Faster Accelerated First-order Methods for Convex Optimization with Strongly Convex Function Constraints

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=pEhvscmSgG

@inproceedings{ alles2024constrained, title={Constrained Latent Action Policies for Model-Based Offline Reinforcement Learning}, author={Marvin Alles and Philip Becker-Ehmck and Patrick van der Smagt and Maximilian Karl}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pEhvscmSgG} }

In offline reinforcement learning, a policy is learned using a static dataset in the absence of costly feedback from the environment. In contrast to the online setting, only using static datasets poses additional challenges, such as policies generating out-of-distribution samples. Model-based offline reinforcement learning methods try to overcome these by learning a model of the underlying dynamics of the environment and using it to guide policy search. It is beneficial but, with limited datasets, errors in the model and the issue of value overestimation among out-of-distribution states can worsen performance. Current model-based methods apply some notion of conservatism to the Bellman update, often implemented using uncertainty estimation derived from model ensembles. In this paper, we propose Constrained Latent Action Policies (C-LAP) which learns a generative model of the joint distribution of observations and actions. We cast policy learning as a constrained objective to always stay within the support of the latent action distribution, and use the generative capabilities of the model to impose an implicit constraint on the generated actions. Thereby eliminating the need to use additional uncertainty penalties on the Bellman update and significantly decreasing the number of gradient steps required to learn a policy. We empirically evaluate C-LAP on the D4RL and V-D4RL benchmark, and show that C-LAP is competitive to state-of-the-art methods, especially outperforming on datasets with visual observations.

Constrained Latent Action Policies for Model-Based Offline Reinforcement Learning

NeurIPS.cc/2024/Conference

2411.04562

poster

https://openreview.net/forum?id=pCVxYw6FKg

@inproceedings{ lim2024the, title={The Empirical Impact of Neural Parameter Symmetries, or Lack Thereof}, author={Derek Lim and Theo Putterman and Robin Walters and Haggai Maron and Stefanie Jegelka}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pCVxYw6FKg} }

Many algorithms and observed phenomena in deep learning appear to be affected by parameter symmetries --- transformations of neural network parameters that do not change the underlying neural network function. These include linear mode connectivity, model merging, Bayesian neural network inference, metanetworks, and several other characteristics of optimization or loss-landscapes. However, theoretical analysis of the relationship between parameter space symmetries and these phenonmena is difficult. In this work, we empirically investigate the impact of neural parameter symmetries by introducing new neural network architectures that have reduced parameter space symmetries. We develop two methods, with some provable guarantees, of modifying standard neural networks to reduce parameter space symmetries. With these new methods, we conduct a comprehensive experimental study consisting of multiple tasks aimed at assessing the effect of removing parameter symmetries. Our experiments reveal several interesting observations on the empirical impact of parameter symmetries; for instance, we observe linear mode connectivity between our networks without alignment of weight spaces, and we find that our networks allow for faster and more effective Bayesian neural network training.

The Empirical Impact of Neural Parameter Symmetries, or Lack Thereof

NeurIPS.cc/2024/Conference

2405.20231

poster

https://openreview.net/forum?id=pCJ0l1JVUX

@inproceedings{ dong2024hamba, title={Hamba: Single-view 3D Hand Reconstruction with Graph-guided Bi-Scanning Mamba}, author={Haoye Dong and Aviral Chharia and Wenbo Gou and Francisco Vicente Carrasco and Fernando De la Torre}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pCJ0l1JVUX} }

3D Hand reconstruction from a single RGB image is challenging due to the articulated motion, self-occlusion, and interaction with objects. Existing SOTA methods employ attention-based transformers to learn the 3D hand pose and shape, yet they do not fully achieve robust and accurate performance, primarily due to inefficiently modeling spatial relations between joints. To address this problem, we propose a novel graph-guided Mamba framework, named Hamba, which bridges graph learning and state space modeling. Our core idea is to reformulate Mamba's scanning into graph-guided bidirectional scanning for 3D reconstruction using a few effective tokens. This enables us to efficiently learn the spatial relationships between joints for improving reconstruction performance. Specifically, we design a Graph-guided State Space (GSS) block that learns the graph-structured relations and spatial sequences of joints and uses 88.5\% fewer tokens than attention-based methods. Additionally, we integrate the state space features and the global features using a fusion module. By utilizing the GSS block and the fusion module, Hamba effectively leverages the graph-guided state space features and jointly considers global and local features to improve performance. Experiments on several benchmarks and in-the-wild tests demonstrate that Hamba significantly outperforms existing SOTAs, achieving the PA-MPVPE of 5.3mm and F@15mm of 0.992 on FreiHAND. At the time of this paper's acceptance, Hamba holds the top position, Rank 1, in two competition leaderboards on 3D hand reconstruction.

Hamba: Single-view 3D Hand Reconstruction with Graph-guided Bi-Scanning Mamba

NeurIPS.cc/2024/Conference

2407.09646

poster

https://openreview.net/forum?id=pC44UMwy2v

@inproceedings{ chen2024unlocking, title={Unlocking the Capabilities of Thought: A Reasoning Boundary Framework to Quantify and Optimize Chain-of-Thought}, author={Qiguang Chen and Libo Qin and Jiaqi WANG and Jingxuan Zhou and Wanxiang Che}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pC44UMwy2v} }

Chain-of-Thought (CoT) reasoning has emerged as a promising approach for enhancing the performance of large language models (LLMs) on complex reasoning tasks. Recently, a series of studies attempt to explain the mechanisms underlying CoT, aiming to deepen the understanding of its efficacy. Nevertheless, the existing research faces two major challenges: (1) a lack of quantitative metrics to assess CoT capabilities and (2) a dearth of guidance on optimizing CoT performance. Motivated by this, in this work, we introduce a novel reasoning boundary framework (RBF) to address these challenges. To solve the lack of quantification, we first define a reasoning boundary (RB) to quantify the upper-bound of CoT and establish a combination law for RB, enabling a practical quantitative approach applicable to various real-world CoT tasks. To address the lack of optimization, we propose three categories of RBs. We further optimize these categories with combination laws focused on RB promotion and reasoning path optimization for CoT improvement. Through extensive experiments on 27 models and 5 tasks, the study validates the existence and rationality of the proposed framework. Furthermore, it explains the effectiveness of 10 CoT strategies and guides optimization from two perspectives. We hope this work can provide a comprehensive understanding of the boundaries and optimization strategies for reasoning in LLMs. Our code and data are available at https://github.com/LightChen233/reasoning-boundary.

Unlocking the Capabilities of Thought: A Reasoning Boundary Framework to Quantify and Optimize Chain-of-Thought

NeurIPS.cc/2024/Conference

2410.05695

https://huggingface.co/papers/2410.05695

oral

https://openreview.net/forum?id=pASJxzMJb7

@inproceedings{ yokoi2024zipfian, title={Zipfian Whitening}, author={Sho Yokoi and Han Bao and Hiroto Kurita and Hidetoshi Shimodaira}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=pASJxzMJb7} }

The word embedding space in neural models is skewed, and correcting this can improve task performance. We point out that most approaches for modeling, correcting, and measuring the symmetry of an embedding space implicitly assume that the word frequencies are *uniform*; in reality, word frequencies follow a highly non-uniform distribution, known as *Zipf's law*. Surprisingly, simply performing PCA whitening weighted by the empirical word frequency that follows Zipf's law significantly improves task performance, surpassing established baselines. From a theoretical perspective, both our approach and existing methods can be clearly categorized: word representations are distributed according to an exponential family with either uniform or Zipfian base measures. By adopting the latter approach, we can naturally emphasize informative low-frequency words in terms of their vector norm, which becomes evident from the information-geometric perspective, and in terms of the loss functions for imbalanced classification. Additionally, our theory corroborates that popular natural language processing methods, such as skip-gram negative sampling, WhiteningBERT, and headless language models, work well just because their word embeddings encode the empirical word frequency into the underlying probabilistic model.

Zipfian Whitening

NeurIPS.cc/2024/Conference

2411.00680

https://huggingface.co/papers/2411.00680

poster

https://openreview.net/forum?id=p54CYwdjVP

@inproceedings{ lin2024a, title={A Globally Optimal Portfolio for m-Sparse Sharpe Ratio Maximization}, author={Yizun Lin and Zhao-Rong Lai and Cheng Li}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p54CYwdjVP} }

The Sharpe ratio is an important and widely-used risk-adjusted return in financial engineering. In modern portfolio management, one may require an m-sparse (no more than m active assets) portfolio to save managerial and financial costs. However, few existing methods can optimize the Sharpe ratio with the m-sparse constraint, due to the nonconvexity and the complexity of this constraint. We propose to convert the m-sparse fractional optimization problem into an equivalent m-sparse quadratic programming problem. The semi-algebraic property of the resulting objective function allows us to exploit the Kurdyka-Lojasiewicz property to develop an efficient Proximal Gradient Algorithm (PGA) that leads to a portfolio which achieves the globally optimal m-sparse Sharpe ratio under certain conditions. The convergence rates of PGA are also provided. To the best of our knowledge, this is the first proposal that achieves a globally optimal m-sparse Sharpe ratio with a theoretically-sound guarantee.

A Globally Optimal Portfolio for m-Sparse Sharpe Ratio Maximization

NeurIPS.cc/2024/Conference

2410.21100

poster

https://openreview.net/forum?id=p50Dyqk0GX

@inproceedings{ li2024dual, title={Dual Risk Minimization: Towards Next-Level Robustness in Fine-tuning Zero-Shot Models}, author={Kaican Li and Weiyan Xie and Yongxiang Huang and Didan Deng and Lanqing HONG and Zhenguo Li and Ricardo Silva and Nevin L. Zhang}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p50Dyqk0GX} }

Fine-tuning foundation models often compromises their robustness to distribution shifts. To remedy this, most robust fine-tuning methods aim to preserve the pre-trained features. However, not all pre-trained features are robust and those methods are largely indifferent to which ones to preserve. We propose dual risk minimization (DRM), which combines empirical risk minimization with worst-case risk minimization, to better preserve the core features of downstream tasks. In particular, we utilize core-feature descriptions generated by LLMs to induce core-based zero-shot predictions which then serve as proxies to estimate the worst-case risk. DRM balances two crucial aspects of model robustness: expected performance and worst-case performance, establishing a new state of the art on various real-world benchmarks. DRM significantly improves the out-of-distribution performance of CLIP ViT-L/14@336 on ImageNet (75.9$\to$77.1), WILDS-iWildCam (47.1$\to$51.8), and WILDS-FMoW (50.7$\to$53.1); opening up new avenues for robust fine-tuning. Our code is available at https://github.com/vaynexie/DRM.

Dual Risk Minimization: Towards Next-Level Robustness in Fine-tuning Zero-Shot Models

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=p4a1nSvwD7

@inproceedings{ gu2024from, title={From Dictionary to Tensor: A Scalable Multi-View Subspace Clustering Framework with Triple Information Enhancement}, author={Zhibin Gu and Songhe Feng}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p4a1nSvwD7} }

While Tensor-based Multi-view Subspace Clustering (TMSC) has garnered significant attention for its capacity to effectively capture high-order correlations among multiple views, three notable limitations in current TMSC methods necessitate consideration: 1) high computational complexity and reliance on dictionary completeness resulting from using observed data as the dictionary, 2) inaccurate subspace representation stemming from the oversight of local geometric information and 3) under-penalization of noise-related singular values within tensor data caused by treating all singular values equally. To address these limitations, this paper presents a \textbf{S}calable TMSC framework with \textbf{T}riple inf\textbf{O}rmatio\textbf{N} \textbf{E}nhancement (\textbf{STONE}). Notably, an enhanced anchor dictionary learning mechanism has been utilized to recover the low-rank anchor structure, resulting in reduced computational complexity and increased resilience, especially in scenarios with inadequate dictionaries. Additionally, we introduce an anchor hypergraph Laplacian regularizer to preserve the inherent geometry of the data within the subspace representation. Simultaneously, an improved hyperbolic tangent function has been employed as a precise approximation for tensor rank, effectively capturing the significant variations in singular values. Extensive experimentation on a variety of datasets demonstrates that our approach surpasses SOTA methods in both effectiveness and efficiency.

From Dictionary to Tensor: A Scalable Multi-View Subspace Clustering Framework with Triple Information Enhancement

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=p43ObIwJFW

@inproceedings{ chen2024learning, title={Learning to Solve Quadratic Unconstrained Binary Optimization in a Classification Way}, author={Ming Chen and Jie Chun and Shang Xiang and Luona Wei and Yonghao Du and Qian Wan and Yuning Chen and Yingwu Chen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p43ObIwJFW} }

The quadratic unconstrained binary optimization (QUBO) is a well-known NP-hard problem that takes an $n\times n$ matrix $Q$ as input and decides an $n$-dimensional 0-1 vector $x$, to optimize a quadratic function. Existing learning-based models that always formulate the solution process as sequential decisions suffer from high computational overload. To overcome this issue, we propose a neural solver called the Value Classification Model (VCM) that formulates the solution process from a classification perspective. It applies a Depth Value Network (DVN) based on graph convolution that exploits the symmetry property in $Q$ to auto-grasp value features. These features are then fed into a Value Classification Network (VCN) which directly generates classification solutions. Trained by a highly efficient model-tailored Greedy-guided Self Trainer (GST) which does not require any priori optimal labels, VCM significantly outperforms competitors in both computational efficiency and solution quality with a remarkable generalization ability. It can achieve near-optimal solutions in milliseconds with an average optimality gap of just 0.362\% on benchmarks with up to 2500 variables. Notably, a VCM trained at a specific DVN depth can steadily find better solutions by simply extending the testing depth, which narrows the gap to 0.034\% on benchmarks. To our knowledge, this is the first learning-based model to reach such a performance.

Learning to Solve Quadratic Unconstrained Binary Optimization in a Classification Way

NeurIPS.cc/2024/Conference

oral

https://openreview.net/forum?id=p3tSEFMwpG

@inproceedings{ helli2024driftresilient, title={Drift-Resilient Tab{PFN}: In-Context Learning Temporal Distribution Shifts on Tabular Data}, author={Kai Helli and David Schnurr and Noah Hollmann and Samuel M{\"u}ller and Frank Hutter}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p3tSEFMwpG} }

While most ML models expect independent and identically distributed data, this assumption is often violated in real-world scenarios due to distribution shifts, resulting in the degradation of machine learning model performance. Until now, no tabular method has consistently outperformed classical supervised learning, which ignores these shifts. To address temporal distribution shifts, we present Drift-Resilient TabPFN, a fresh approach based on In-Context Learning with a Prior-Data Fitted Network that learns the learning algorithm itself: it accepts the entire training dataset as input and makes predictions on the test set in a single forward pass. Specifically, it learns to approximate Bayesian inference on synthetic datasets drawn from a prior that specifies the model's inductive bias. This prior is based on structural causal models (SCM), which gradually shift over time. To model shifts of these causal models, we use a secondary SCM, that specifies changes in the primary model parameters. The resulting Drift-Resilient TabPFN can be applied to unseen data, runs in seconds on small to moderately sized datasets and needs no hyperparameter tuning. Comprehensive evaluations across 18 synthetic and real-world datasets demonstrate large performance improvements over a wide range of baselines, such as XGB, CatBoost, TabPFN, and applicable methods featured in the Wild-Time benchmark. Compared to the strongest baselines, it improves accuracy from 0.688 to 0.744 and ROC AUC from 0.786 to 0.832 while maintaining stronger calibration. This approach could serve as significant groundwork for further research on out-of-distribution prediction.

Drift-Resilient TabPFN: In-Context Learning Temporal Distribution Shifts on Tabular Data

NeurIPS.cc/2024/Conference

2411.10634

poster

https://openreview.net/forum?id=p3nPHMpx04

@inproceedings{ sakai2024a, title={A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation}, author={Tomoya Sakai and Haoxiang Qiu and Takayuki Katsuki and Daiki Kimura and Takayuki Osogami and Tadanobu Inoue}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p3nPHMpx04} }

The goal of *generalized* few-shot semantic segmentation (GFSS) is to recognize *novel-class* objects through training with a few annotated examples and the *base-class* model that learned the knowledge about the base classes. Unlike the classic few-shot semantic segmentation, GFSS aims to classify pixels into both base and novel classes, meaning it is a more practical setting. Current GFSS methods rely on several techniques such as using combinations of customized modules, carefully designed loss functions, meta-learning, and transductive learning. However, we found that a simple rule and standard supervised learning substantially improve the GFSS performance. In this paper, we propose a simple yet effective method for GFSS that does not use the techniques mentioned above. Also, we theoretically show that our method perfectly maintains the segmentation performance of the base-class model over most of the base classes. Through numerical experiments, we demonstrated the effectiveness of our method. It improved in novel-class segmentation performance in the $1$-shot scenario by $6.1$% on the PASCAL-$5^i$ dataset, $4.7$% on the PASCAL-$10^i$ dataset, and $1.0$% on the COCO-$20^i$ dataset.

A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=p3hNrpeWMe

@inproceedings{ alam2024a, title={A Walsh Hadamard Derived Linear Vector Symbolic Architecture}, author={Mohammad Mahmudul Alam and Alexander Oberle and Edward Raff and Stella Biderman and Tim Oates and James Holt}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p3hNrpeWMe} }

Vector Symbolic Architectures (VSAs) are one approach to developing Neuro-symbolic AI, where two vectors in $\mathbb{R}^d$ are 'bound' together to produce a new vector in the same space. VSAs support the commutativity and associativity of this binding operation, along with an inverse operation, allowing one to construct symbolic-style manipulations over real-valued vectors. Most VSAs were developed before deep learning and automatic differentiation became popular and instead focused on efficacy in hand-designed systems. In this work, we introduce the Hadamard-derived linear Binding (HLB), which is designed to have favorable computational efficiency, and efficacy in classic VSA tasks, and perform well in differentiable systems.

A Walsh Hadamard Derived Linear Vector Symbolic Architecture

NeurIPS.cc/2024/Conference

2410.22669

poster

https://openreview.net/forum?id=p3gMGkHMkM

@inproceedings{ lim2024particle, title={Particle Semi-Implicit Variational Inference}, author={Jen Ning Lim and Adam Michael Johansen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p3gMGkHMkM} }

Semi-implicit variational inference (SIVI) enriches the expressiveness of variational families by utilizing a kernel and a mixing distribution to hierarchically define the variational distribution. Existing SIVI methods parameterize the mixing distribution using implicit distributions, leading to intractable variational densities. As a result, directly maximizing the evidence lower bound (ELBO) is not possible, so they resort to one of the following: optimizing bounds on the ELBO, employing costly inner-loop Markov chain Monte Carlo runs, or solving minimax objectives. In this paper, we propose a novel method for SIVI called Particle Variational Inference (PVI) which employs empirical measures to approximate the optimal mixing distributions characterized as the minimizer of a free energy functional. PVI arises naturally as a particle approximation of a Euclidean–Wasserstein gradient flow and, unlike prior works, it directly optimizes the ELBO whilst making no parametric assumption about the mixing distribution. Our empirical results demonstrate that PVI performs favourably compared to other SIVI methods across various tasks. Moreover, we provide a theoretical analysis of the behaviour of the gradient flow of a related free energy functional: establishing the existence and uniqueness of solutions as well as propagation of chaos results.

Particle Semi-Implicit Variational Inference

NeurIPS.cc/2024/Conference

2407.00649

oral

https://openreview.net/forum?id=p37NlKi9vl

@inproceedings{ buffelli2024exact, title={Exact, Tractable Gauss-Newton Optimization in Deep Reversible Architectures Reveal Poor Generalization}, author={Davide Buffelli and Jamie McGowan and Wangkun Xu and Alexandru Cioba and Da-shan Shiu and Guillaume Hennequin and Alberto Bernacchia}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p37NlKi9vl} }

Second-order optimization has been shown to accelerate the training of deep neural networks in many applications, often yielding faster progress per iteration on the training loss compared to first-order optimizers. However, the generalization properties of second-order methods are still being debated. Theoretical investigations have proved difficult to carry out outside the tractable settings of heavily simplified model classes - thus, the relevance of existing theories to practical deep learning applications remains unclear. Similarly, empirical studies in large-scale models and real datasets are significantly confounded by the necessity to approximate second-order updates in practice. It is often unclear whether the observed generalization behaviour arises specifically from the second-order nature of the parameter updates, or instead reflects the specific structured (e.g. Kronecker) approximations used or any damping-based interpolation towards first-order updates. Here, we show for the first time that exact Gauss-Newton (GN) updates take on a tractable form in a class of deep reversible architectures that are sufficiently expressive to be meaningfully applied to common benchmark datasets. We exploit this novel setting to study the training and generalization properties of the GN optimizer. We find that exact GN generalizes poorly. In the mini-batch training setting, this manifests as rapidly saturating progress even on the training loss, with parameter updates found to overfit each mini-batch without producing the features that would support generalization to other mini-batches. In contrast to previous work, we show that our experiments run in the feature learning regime, in which the neural tangent kernel (NTK) changes during the course of training. However, changes in the NTK are not associated with any significant change in neural representations, explaining the lack of generalization.

Exact, Tractable Gauss-Newton Optimization in Deep Reversible Architectures Reveal Poor Generalization

NeurIPS.cc/2024/Conference

2411.07979

poster

https://openreview.net/forum?id=p2PO2PUPFY

@inproceedings{ wang2024textditextpose, title={\${\textbackslash}text\{Di\}{\textasciicircum}2{\textbackslash}text\{Pose\}\$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation}, author={Weiquan Wang and Jun Xiao and Chunping Wang and Wei Liu and Zhao Wang and Long Chen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p2PO2PUPFY} }

Diffusion models have demonstrated their effectiveness in addressing the inherent uncertainty and indeterminacy in monocular 3D human pose estimation (HPE). Despite their strengths, the need for large search spaces and the corresponding demand for substantial training data make these models prone to generating biomechanically unrealistic poses. This challenge is particularly noticeable in occlusion scenarios, where the complexity of inferring 3D structures from 2D images intensifies. In response to these limitations, we introduce the **Di**screte **Di**ffusion **Pose** (**$\text{Di}^2\text{Pose}$**), a novel framework designed for occluded 3D HPE that capitalizes on the benefits of a discrete diffusion model. Specifically, **$\text{Di}^2\text{Pose}$** employs a two-stage process: it first converts 3D poses into a discrete representation through a pose quantization step, which is subsequently modeled in latent space through a discrete diffusion process. This methodological innovation restrictively confines the search space towards physically viable configurations and enhances the model’s capability to comprehend how occlusions affect human pose within the latent space. Extensive evaluations conducted on various benchmarks (e.g., Human3.6M, 3DPW, and 3DPW-Occ) have demonstrated its effectiveness.

Di^2Pose: Discrete Diffusion Model for Occluded 3D Human Pose Estimation

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=p1ft33Mu3J

@inproceedings{ vladymyrov2024linear, title={Linear Transformers are Versatile In-Context Learners}, author={Max Vladymyrov and Johannes Von Oswald and Mark Sandler and Rong Ge}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p1ft33Mu3J} }

Recent research has demonstrated that transformers, particularly linear attention models, implicitly execute gradient-descent-like algorithms on data provided in-context during their forward inference step. However, their capability in handling more complex problems remains unexplored. In this paper, we prove that each layer of a linear transformer maintains a weight vector for an implicit linear regression problem and can be interpreted as performing a variant of preconditioned gradient descent. We also investigate the use of linear transformers in a challenging scenario where the training data is corrupted with different levels of noise. Remarkably, we demonstrate that for this problem linear transformers discover an intricate and highly effective optimization algorithm, surpassing or matching in performance many reasonable baselines. We analyze this algorithm and show that it is a novel approach incorporating momentum and adaptive rescaling based on noise levels. Our findings show that even linear transformers possess the surprising ability to discover sophisticated optimization strategies.

Linear Transformers are Versatile In-Context Learners

NeurIPS.cc/2024/Conference

2402.14180

https://huggingface.co/papers/2402.14180

poster

https://openreview.net/forum?id=p1LpXNPmIa

@inproceedings{ yu2024promptfix, title={PromptFix: You Prompt and We Fix the Photo}, author={Yongsheng Yu and Ziyun Zeng and Hang Hua and Jianlong Fu and Jiebo Luo}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p1LpXNPmIa} }

Diffusion models equipped with language models demonstrate excellent controllability in image generation tasks, allowing image processing to adhere to human instructions. However, the lack of diverse instruction-following data hampers the development of models that effectively recognize and execute user-customized instructions, particularly in low-level tasks. Moreover, the stochastic nature of the diffusion process leads to deficiencies in image generation or editing tasks that require the detailed preservation of the generated images. To address these limitations, we propose PromptFix, a comprehensive framework that enables diffusion models to follow human instructions to perform a wide variety of image-processing tasks. First, we construct a large-scale instruction-following dataset that covers comprehensive image-processing tasks, including low-level tasks, image editing, and object creation. Next, we propose a high-frequency guidance sampling method to explicitly control the denoising process and preserve high-frequency details in unprocessed areas. Finally, we design an auxiliary prompting adapter, utilizing Vision-Language Models (VLMs) to enhance text prompts and improve the model's task generalization. Experimental results show that PromptFix outperforms previous methods in various image-processing tasks. Our proposed model also achieves comparable inference efficiency with these baseline models and exhibits superior zero-shot capabilities in blind restoration and combination tasks.

PromptFix: You Prompt and We Fix the Photo

NeurIPS.cc/2024/Conference

2405.16785

https://huggingface.co/papers/2405.16785

poster

https://openreview.net/forum?id=p0BBKhD5aI

@inproceedings{ bordelon2024infinite, title={Infinite Limits of Multi-head Transformer Dynamics}, author={Blake Bordelon and Hamza Tahir Chaudhry and Cengiz Pehlevan}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=p0BBKhD5aI} }

In this work we analyze various scaling limits of the training dynamics of transformer models in the feature learning regime. We identify the set of parameterizations which admit well defined infinite width and depth limits that allow the attention layers to update throughout training, a relevant notion of feature learning in these models. We then use tools from dynamical mean field theory (DMFT) to analyze various infinite limits (infinite heads, infinite key/query dimension, and infinite depth) which have different statistical descriptions depending on which infinite limit is taken and how attention layers are scaled. We provide numerical evidence of convergence to the limits and show they maintain the correct scale of updates for both SGD and Adam.

Infinite Limits of Multi-head Transformer Dynamics

NeurIPS.cc/2024/Conference

2405.15712

poster

https://openreview.net/forum?id=oyl2Fnzune

@inproceedings{ zhu2024unimed, title={Uni-Med: A Unified Medical Generalist Foundation Model For Multi-Task Learning Via Connector-MoE}, author={Xun Zhu and Ying Hu and Fanbin Mo and Miao Li and Ji Wu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=oyl2Fnzune} }

Multi-modal large language models (MLLMs) have shown impressive capabilities as a general-purpose interface for various visual and linguistic tasks. However, building a unified MLLM for multi-task learning in the medical field remains a thorny challenge. To mitigate the tug-of-war problem of multi-modal multi-task optimization in MLLMs, recent advances primarily focus on improving the LLM components, while neglecting the connector that bridges the gap between modalities. In this paper, we introduce Uni-Med, a novel medical generalist foundation model which consists of a universal visual feature extraction module, a connector mixture-of-experts (CMoE) module, and an LLM. Benefiting from the proposed CMoE that leverages a well-designed router with a mixture of projection experts at the connector, Uni-Med achieves efficient solution to the tug-of-war problem and can perform six different medical tasks including question answering, visual question answering, report generation, referring expression comprehension, referring expression generation and image classification. To the best of our knowledge, Uni-Med is the first effort to tackle multi-task interference at the connector in MLLMs. Extensive ablation experiments validate the effectiveness of introducing CMoE under any configuration, with up to an average 8% performance gains. We further provide interpretation analysis of the tug-of-war problem from the perspective of gradient optimization and parameter statistics. Compared to previous state-of-the-art medical MLLMs, Uni-Med achieves competitive or superior evaluation metrics on diverse tasks. Code and resources are available at https://github.com/tsinghua-msiip/Uni-Med.

Uni-Med: A Unified Medical Generalist Foundation Model For Multi-Task Learning Via Connector-MoE

NeurIPS.cc/2024/Conference

2409.17508

poster

https://openreview.net/forum?id=oyiBLfNJvY

@inproceedings{ tolguenec2024exploration, title={Exploration by Learning Diverse Skills through Successor State Representations}, author={Paul-Antoine LE TOLGUENEC and Yann BESSE and Florent Teichteil-K{\"o}nigsbuch and Dennis George Wilson and Emmanuel Rachelson}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=oyiBLfNJvY} }

The ability to perform different skills can encourage agents to explore. In this work, we aim to construct a set of diverse skills that uniformly cover the state space. We propose a formalization of this search for diverse skills, building on a previous definition based on the mutual information between states and skills. We consider the distribution of states reached by a policy conditioned on each skill and leverage the successor state representation to maximize the difference between these skill distributions. We call this approach LEADS: Learning Diverse Skills through Successor State Representations. We demonstrate our approach on a set of maze navigation and robotic control tasks which show that our method is capable of constructing a diverse set of skills which exhaustively cover the state space without relying on reward or exploration bonuses. Our findings demonstrate that this new formalization promotes more robust and efficient exploration by combining mutual information maximization and exploration bonuses.

Exploration by Learning Diverse Skills through Successor State Representations

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=owuEcT6BTl

@inproceedings{ park2024emergence, title={Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space}, author={Core Francisco Park and Maya Okawa and Andrew Lee and Ekdeep Singh Lubana and Hidenori Tanaka}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=owuEcT6BTl} }

Modern generative models demonstrate impressive capabilities, likely stemming from an ability to identify and manipulate abstract concepts underlying their training data. However, fundamental questions remain: what determines the concepts a model learns, the order in which it learns them, and its ability to manipulate those concepts? To address these questions, we propose analyzing a model’s learning dynamics via a framework we call the concept space, where each axis represents an independent concept underlying the data generating process. By characterizing learning dynamics in this space, we identify how the speed at which a concept is learned, and hence the order of concept learning, is controlled by properties of the data we term concept signal. Further, we observe moments of sudden turns in the direction of a model’s learning dynamics in concept space. Surprisingly, these points precisely correspond to the emergence of hidden capabilities, i.e., where latent interventions show the model possesses the capability to manipulate a concept, but these capabilities cannot yet be elicited via naive input prompting. While our results focus on synthetically defined toy datasets, we hypothesize a general claim on emergence of hidden capabilities may hold: generative models possess latent capabilities that emerge suddenly and consistently during training, though a model might not exhibit these capabilities under naive input prompting.

Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space

NeurIPS.cc/2024/Conference

2406.19370

https://huggingface.co/papers/2406.19370

oral

https://openreview.net/forum?id=owHj0G15cd

@inproceedings{ huang2024direct, title={Direct Preference-Based Evolutionary Multi-Objective Optimization with Dueling Bandits}, author={Tian Huang and Shengbo Wang and Ke Li}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=owHj0G15cd} }

The ultimate goal of multi-objective optimization (MO) is to assist human decision-makers (DMs) in identifying solutions of interest (SOI) that optimally reconcile multiple objectives according to their preferences. Preference-based evolutionary MO (PBEMO) has emerged as a promising framework that progressively approximates SOI by involving human in the optimization-cum-decision-making process. Yet, current PBEMO approaches are prone to be inefficient and misaligned with the DM’s true aspirations, especially when inadvertently exploiting mis-calibrated reward models. This is further exacerbated when considering the stochastic nature of human feedback. This paper proposes a novel framework that navigates MO to SOI by directly leveraging human feedback without being restricted by a predefined reward model nor cumbersome model selection. Specifically, we developed a clustering-based stochastic dueling bandits algorithm that strategically scales well to high-dimensional dueling bandits, and achieves a regret of $\mathcal{O}(K^2\log T)$, where $K$ is the number of clusters and $T$ is the number of rounds. The learned preferences are then transformed into a unified probabilistic format that can be readily adapted to prevalent EMO algorithms. This also leads to a principled termination criterion that strategically manages human cognitive loads and computational budget. Experiments on $48$ benchmark test problems, including synthetic problems, RNA inverse design and protein structure prediction, fully demonstrate the effectiveness of our proposed approach.

Direct Preference-Based Evolutionary Multi-Objective Optimization with Dueling Bandits

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=ouoBW2PXFQ

@inproceedings{ park2024calanet, title={{CALAN}et: Cheap All-Layer Aggregation for Human Activity Recognition}, author={Jaegyun Park and Dae-Won Kim and Jaesung Lee}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ouoBW2PXFQ} }

With the steady growth of sensing technology and wearable devices, sensor-based human activity recognition has become essential in widespread applications, such as healthcare monitoring and fitness tracking, where accurate and real-time systems are required. To achieve real-time response, recent studies have focused on lightweight neural network models. Specifically, they designed the network architectures by restricting the number of layers shallowly or connections of each layer. However, these approaches suffer from limited accuracy because the classifier only uses the features at the last layer. In this study, we propose a cheap all-layer aggregation network, CALANet, for accuracy improvement while maintaining the efficiency of existing real-time HAR models. Specifically, CALANet allows the classifier to aggregate the features for all layers, resulting in a performance gain. In addition, this work proves that the theoretical computation cost of CALANet is equivalent to that of conventional networks. Evaluated on seven publicly available datasets, CALANet outperformed existing methods, achieving state-of-the-art performance. The source codes of the CALANet are publicly available at https://github.com/jgpark92/CALANet.

CALANet: Cheap All-Layer Aggregation for Human Activity Recognition

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=otxOtsWCMb

@inproceedings{ wallingford2024from, title={From an Image to a Scene: Learning to Imagine the World from a Million 360{\textdegree} Videos}, author={Matthew Wallingford and Anand Bhattad and Aditya Kusupati and Vivek Ramanujan and Matt Deitke and Aniruddha Kembhavi and Roozbeh Mottaghi and Wei-Chiu Ma and Ali Farhadi}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=otxOtsWCMb} }

Three-dimensional (3D) understanding of objects and scenes play a key role in humans' ability to interact with the world and has been an active area of research in computer vision, graphics, and robotics. Large scale synthetic and object-centric 3D datasets have shown to be effective in training models that have 3D understanding of objects. However, applying a similar approach to real-world objects and scenes is difficult due to a lack of large-scale data. Videos are a potential source for real-world 3D data, but finding diverse yet corresponding views of the same content have shown to be difficult at scale. Furthermore, standard videos come with fixed viewpoints, determined at the time of capture. This restricts the ability to access scenes from a variety of more diverse and potentially useful perspectives. We argue that large scale ODIN videos can address these limitations to provide scalable corresponding frames from diverse views. In this paper we introduce 360-1M, a 360° video dataset consisting of 1 million videos, and a process for efficiently finding corresponding frames from diverse viewpoints at scale. We train our diffusion-based model, ODIN, on 360-1M. Empowered by the largest real-world, multi-view dataset to date, ODIN is able to freely generate novel views of real-world scenes. Unlike previous methods, ODIN can move the camera through the environment, enabling the model to infer the geometry and layout of the scene. Additionally, we show improved performance on standard novel view synthesis and 3D reconstruction benchmarks.

From an Image to a Scene: Learning to Imagine the World from a Million 360° Videos

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=otZPBS0un6

@inproceedings{ li2024freqblender, title={FreqBlender: Enhancing DeepFake Detection by Blending Frequency Knowledge}, author={Hanzhe LI and Jiaran Zhou and Yuezun Li and Baoyuan Wu and Bin Li and Junyu Dong}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=otZPBS0un6} }

Generating synthetic fake faces, known as pseudo-fake faces, is an effective way to improve the generalization of DeepFake detection. Existing methods typically generate these faces by blending real or fake faces in spatial domain. While these methods have shown promise, they overlook the simulation of frequency distribution in pseudo-fake faces, limiting the learning of generic forgery traces in-depth. To address this, this paper introduces {\em FreqBlender}, a new method that can generate pseudo-fake faces by blending frequency knowledge. Concretely, we investigate the major frequency components and propose a Frequency Parsing Network to adaptively partition frequency components related to forgery traces. Then we blend this frequency knowledge from fake faces into real faces to generate pseudo-fake faces. Since there is no ground truth for frequency components, we describe a dedicated training strategy by leveraging the inner correlations among different frequency knowledge to instruct the learning process. Experimental results demonstrate the effectiveness of our method in enhancing DeepFake detection, making it a potential plug-and-play strategy for other methods.

FreqBlender: Enhancing DeepFake Detection by Blending Frequency Knowledge

NeurIPS.cc/2024/Conference

2404.13872

poster

https://openreview.net/forum?id=os14qXhy55

@inproceedings{ lu2024octreeocc, title={OctreeOcc: Efficient and Multi-Granularity Occupancy Prediction Using Octree Queries}, author={Yuhang Lu and Xinge ZHU and Tai Wang and Yuexin Ma}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=os14qXhy55} }

Occupancy prediction has increasingly garnered attention in recent years for its fine-grained understanding of 3D scenes. Traditional approaches typically rely on dense, regular grid representations, which often leads to excessive computational demands and a loss of spatial details for small objects. This paper introduces OctreeOcc, an innovative 3D occupancy prediction framework that leverages the octree representation to adaptively capture valuable information in 3D, offering variable granularity to accommodate object shapes and semantic regions of varying sizes and complexities. In particular, we incorporate image semantic information to improve the accuracy of initial octree structures and design an effective rectification mechanism to refine the octree structure iteratively. Our extensive evaluations show that OctreeOcc not only surpasses state-of-the-art methods in occupancy prediction, but also achieves a 15%-24% reduction in computational overhead compared to dense-grid-based methods.

OctreeOcc: Efficient and Multi-Granularity Occupancy Prediction Using Octree Queries

NeurIPS.cc/2024/Conference

2312.03774

poster

https://openreview.net/forum?id=orxQccN8Fm

@inproceedings{ li2024getting, title={Getting More Juice Out of the {SFT} Data: Reward Learning from Human Demonstration Improves {SFT} for {LLM} Alignment}, author={Jiaxiang Li and Siliang Zeng and Hoi To Wai and Chenliang Li and Alfredo Garcia and Mingyi Hong}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=orxQccN8Fm} }

Aligning human preference and value is an important requirement for contemporary foundation models. State-of-the-art techniques such as Reinforcement Learning from Human Feedback (RLHF) often consist of two stages: 1) supervised fine-tuning (SFT), where the model is fine-tuned by learning from human demonstration data; 2) Preference learning, where preference data is used to learn a reward model, which is in turn used by a reinforcement learning (RL) step to fine-tune the model. Such reward model serves as a proxy to human preference, and it is critical to guide the RL step towards improving the model quality. In this work, we argue that the SFT stage significantly benefits from learning a reward model as well. Instead of using the human demonstration data directly via supervised learning, we propose to leverage an Inverse Reinforcement Learning (IRL) technique to {\it simultaneously} build an reward model and a policy model. This approach leads to new SFT algorithms that are not only efficient to implement, but are robust to the presence of low-quality supervised learning data. Moreover, we discover a connection between the proposed IRL based approach, and a recent line of works called Self-Play Fine-tune (SPIN, \cite{chen2024self}). Theoretically, we show that the proposed algorithms converge to the stationary solutions of the IRL problem. Empirically, we align 1B and 7B models using proposed methods and evaluate them on a reward benchmark model and the HuggingFace Open LLM Leaderboard. The proposed methods show significant performance improvement over existing SFT approaches. Our results indicate that it is beneficial to leverage reward learning throughout the entire alignment process. Our code is available at \url{https://github.com/JasonJiaxiangLi/Reward_learning_SFT}.

Getting More Juice Out of the SFT Data: Reward Learning from Human Demonstration Improves SFT for LLM Alignment

NeurIPS.cc/2024/Conference

2405.17888

poster

https://openreview.net/forum?id=oqdy2EFrja

@inproceedings{ zhu2024compositional, title={Compositional 3D-aware Video Generation with {LLM} Director}, author={Hanxin Zhu and Tianyu He and Anni Tang and Junliang Guo and Zhibo Chen and Jiang Bian}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=oqdy2EFrja} }

Significant progress has been made in text-to-video generation through the use of powerful generative models and large-scale internet data. However, substantial challenges remain in precisely controlling individual elements within the generated video, such as the movement and appearance of specific characters and the manipulation of viewpoints. In this work, we propose a novel paradigm that generates each element in 3D representation separately and then composites them with priors from Large Language Models (LLMs) and 2D diffusion models. Specifically, given an input textual query, our scheme consists of four stages: 1) we leverage the LLMs as the director to first decompose the complex query into several sub-queries, where each sub-query describes each element of the generated video; 2) to generate each element, pre-trained models are invoked by the LLMs to obtain the corresponding 3D representation; 3) to composite the generated 3D representations, we prompt multi-modal LLMs to produce coarse guidance on the scale, location, and trajectory of different objects; 4) to make the results adhere to natural distribution, we further leverage 2D diffusion priors and use score distillation sampling to refine the composition. Extensive experiments demonstrate that our method can generate high-fidelity videos from text with flexible control over each element.

Compositional 3D-aware Video Generation with LLM Director

NeurIPS.cc/2024/Conference

2409.00558

https://huggingface.co/papers/2409.00558

poster

https://openreview.net/forum?id=opt72TYzwZ

@inproceedings{ li2024optimal, title={Optimal ablation for interpretability}, author={Maximilian Li and Lucas Janson}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=opt72TYzwZ} }

Interpretability studies often involve tracing the flow of information through machine learning models to identify specific model components that perform relevant computations for tasks of interest. Prior work quantifies the importance of a model component on a particular task by measuring the impact of performing ablation on that component, or simulating model inference with the component disabled. We propose a new method, optimal ablation (OA), and show that OA-based component importance has theoretical and empirical advantages over measuring importance via other ablation methods. We also show that OA-based component importance can benefit several downstream interpretability tasks, including circuit discovery, localization of factual recall, and latent prediction.

Optimal ablation for interpretability

NeurIPS.cc/2024/Conference

2409.09951

oral

https://openreview.net/forum?id=opaRhDvQRD

@inproceedings{ xinrui2024forgetting, title={Forgetting, Ignorance or Myopia: Revisiting Key Challenges in Online Continual Learning}, author={Wang Xinrui and Chuanxing Geng and Wenhai Wan and Shao-Yuan Li and Songcan Chen}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=opaRhDvQRD} }

Online continual learning (OCL) requires the models to learn from constant, endless streams of data. While significant efforts have been made in this field, most were focused on mitigating the \textit{catastrophic forgetting} issue to achieve better classification ability, at the cost of a much heavier training workload. They overlooked that in real-world scenarios, e.g., in high-speed data stream environments, data do not pause to accommodate slow models. In this paper, we emphasize that \textit{model throughput}-- defined as the maximum number of training samples that a model can process within a unit of time -- is equally important. It directly limits how much data a model can utilize and presents a challenging dilemma for current methods. With this understanding, we revisit key challenges in OCL from both empirical and theoretical perspectives, highlighting two critical issues beyond the well-documented catastrophic forgetting: (\romannumeral1) Model's ignorance: the single-pass nature of OCL challenges models to learn effective features within constrained training time and storage capacity, leading to a trade-off between effective learning and model throughput; (\romannumeral2) Model's myopia: the local learning nature of OCL on the current task leads the model to adopt overly simplified, task-specific features and \textit{excessively sparse classifier}, resulting in the gap between the optimal solution for the current task and the global objective. To tackle these issues, we propose the Non-sparse Classifier Evolution framework (NsCE) to facilitate effective global discriminative feature learning with minimal time cost. NsCE integrates non-sparse maximum separation regularization and targeted experience replay techniques with the help of pre-trained models, enabling rapid acquisition of new globally discriminative features. Extensive experiments demonstrate the substantial improvements of our framework in performance, throughput and real-world practicality.

Forgetting, Ignorance or Myopia: Revisiting Key Challenges in Online Continual Learning

NeurIPS.cc/2024/Conference

2409.19245

poster

https://openreview.net/forum?id=omyzrkacme

@inproceedings{ scheid2024learning, title={Learning to Mitigate Externalities: the Coase Theorem with Hindsight Rationality}, author={Antoine Scheid and Aymeric Capitaine and Etienne Boursier and Eric Moulines and Michael Jordan and Alain Oliviero Durmus}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=omyzrkacme} }

In Economics, the concept of externality refers to any indirect effect resulting from an interaction between players and affecting a third party without compensation. Most of the models within which externality has been studied assume that agents have perfect knowledge of their environment and preferences. This is a major hindrance to the practical implementation of many proposed solutions. To adress this issue, we consider a two-players bandit game setting where the actions of one of the player affect the other one. Building upon this setup, we extend the Coase theorem [Coase, 2013], which suggests that the optimal approach for maximizing the social welfare in the presence of externality is to establish property rights, i.e., enabling transfers and bargaining between the players. Nonetheless, this fundamental result relies on the assumption that bargainers possess perfect knowledge of the underlying game. We first demonstrate that in the absence of property rights in the considered online scenario, the social welfare breaks down. We then provide a policy for the players, which allows them to learn a bargaining strategy which maximizes the total welfare, recovering the Coase theorem under uncertainty.

Learning to Mitigate Externalities: the Coase Theorem with Hindsight Rationality

NeurIPS.cc/2024/Conference

2406.19824

oral

https://openreview.net/forum?id=om2Aa0gUha

@inproceedings{ protopapas2024policy, title={Policy Mirror Descent with Lookahead}, author={Kimon Protopapas and Anas Barakat}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=om2Aa0gUha} }

Policy Mirror Descent (PMD) stands as a versatile algorithmic framework encompassing several seminal policy gradient algorithms such as natural policy gradient, with connections with state-of-the-art reinforcement learning (RL) algorithms such as TRPO and PPO. PMD can be seen as a soft Policy Iteration algorithm implementing regularized 1-step greedy policy improvement. However, 1-step greedy policies might not be the best choice and recent remarkable empirical successes in RL such as AlphaGo and AlphaZero have demonstrated that greedy approaches with respect to multiple steps outperform their 1-step counterpart. In this work, we propose a new class of PMD algorithms called $h$-PMD which incorporates multi-step greedy policy improvement with lookahead depth $h$ to the PMD update rule. To solve discounted infinite horizon Markov Decision Processes with discount factor $\gamma$, we show that $h$-PMD which generalizes the standard PMD enjoys a faster dimension-free $\gamma^h$-linear convergence rate, contingent on the computation of multi-step greedy policies. We propose an inexact version of $h$-PMD where lookahead action values are estimated. Under a generative model, we establish a sample complexity for $h$-PMD which improves over prior work. Finally, we extend our result to linear function approximation to scale to large state spaces. Under suitable assumptions, our sample complexity only involves dependence on the dimension of the feature map space instead of the state space size.

Policy Mirror Descent with Lookahead

NeurIPS.cc/2024/Conference

2403.14156

poster

https://openreview.net/forum?id=ojLIEQ0j9T

@inproceedings{ lipshutz2024shaping, title={Shaping the distribution of neural responses with interneurons in a recurrent circuit model}, author={David Lipshutz and Eero P Simoncelli}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ojLIEQ0j9T} }

Efficient coding theory posits that sensory circuits transform natural signals into neural representations that maximize information transmission subject to resource constraints. Local interneurons are thought to play an important role in these transformations, shaping patterns of circuit activity to facilitate and direct information flow. However, the relationship between these coordinated, nonlinear, circuit-level transformations and the properties of interneurons (e.g., connectivity, activation functions) remains unknown. Here, we propose a normative computational model that establishes such a relationship. Our model is derived from an optimal transport objective that conceptualizes the circuit's input-response function as transforming the inputs to achieve a target response distribution. The circuit, which is comprised of primary neurons that are recurrently connected to a set of local interneurons, continuously optimizes this objective by dynamically adjusting both the synaptic connections between neurons as well as the interneuron activation functions. In an application motivated by redundancy reduction theory, we demonstrate that when the inputs are natural image statistics and the target distribution is a spherical Gaussian, the circuit learns a nonlinear transformation that significantly reduces statistical dependencies in neural responses. Overall, our results provide a framework in which the distribution of circuit responses is systematically and nonlinearly controlled by adjustment of interneuron connectivity and activation functions.

Shaping the distribution of neural responses with interneurons in a recurrent circuit model

NeurIPS.cc/2024/Conference

2405.17745

poster

https://openreview.net/forum?id=ojIhvhQBAQ

@inproceedings{ chandrasekaran2024efficient, title={Efficient Discrepancy Testing for Learning with Distribution Shift}, author={Gautam Chandrasekaran and Adam Klivans and Vasilis Kontonis and Konstantinos Stavropoulos and Arsen Vasilyan}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ojIhvhQBAQ} }

A fundamental notion of distance between train and test distributions from the field of domain adaptation is discrepancy distance. While in general hard to compute, here we provide the first set of provably efficient algorithms for testing *localized* discrepancy distance, where discrepancy is computed with respect to a fixed output classifier. These results imply a broad set of new, efficient learning algorithms in the recently introduced model of Testable Learning with Distribution Shift (TDS learning) due to Klivans et al. (2023). Our approach generalizes and improves all prior work on TDS learning: (1) we obtain *universal* learners that succeed simultaneously for large classes of test distributions, (2) achieve near-optimal error rates, and (3) give exponential improvements for constant depth circuits. Our methods further extend to semi-parametric settings and imply the first positive results for low-dimensional convex sets. Additionally, we separate learning and testing phases and obtain algorithms that run in fully polynomial time at test time.

Efficient Discrepancy Testing for Learning with Distribution Shift

NeurIPS.cc/2024/Conference

2406.09373

poster

https://openreview.net/forum?id=ohvXBIPV7e

@inproceedings{ yang2024cspg, title={{CSPG}: Crossing Sparse Proximity Graphs for Approximate Nearest Neighbor Search}, author={Ming Yang and Yuzheng Cai and Weiguo Zheng}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ohvXBIPV7e} }

The state-of-the-art approximate nearest neighbor search (ANNS) algorithm builds a large proximity graph on the dataset and performs a greedy beam search, which may bring many unnecessary explorations. We develop a novel framework, namely *corssing sparse proximity graph (CSPG)*, based on random partitioning of the dataset. It produces a smaller sparse proximity graph for each partition and routing vectors that bind all the partitions. An efficient two-staged approach is designed for exploring *CSPG*, with fast approaching and cross-partition expansion. We theoretically prove that *CSPG* can accelerate the existing graph-based ANNS algorithms by reducing unnecessary explorations. In addition, we conduct extensive experiments on benchmark datasets. The experimental results confirm that the existing graph-based methods can be significantly outperformed by incorporating *CSPG*, achieving 1.5x to 2x speedups of *QPS* in almost all recalls.

CSPG: Crossing Sparse Proximity Graphs for Approximate Nearest Neighbor Search

NeurIPS.cc/2024/Conference

poster

https://openreview.net/forum?id=ohi00YhT3T

@inproceedings{ shen2024neurovision, title={Neuro-Vision to Language: Enhancing Brain Recording-based Visual Reconstruction and Language Interaction}, author={Guobin Shen and Dongcheng Zhao and Xiang He and Linghao Feng and Yiting Dong and Jihang Wang and Qian Zhang and Yi Zeng}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ohi00YhT3T} }

Decoding non-invasive brain recordings is pivotal for advancing our understanding of human cognition but faces challenges due to individual differences and complex neural signal representations. Traditional methods often require customized models and extensive trials, lacking interpretability in visual reconstruction tasks. Our framework integrates 3D brain structures with visual semantics using a *Vision Transformer 3D*. This unified feature extractor efficiently aligns fMRI features with multiple levels of visual embeddings, eliminating the need for subject-specific models and allowing extraction from single-trial data. The extractor consolidates multi-level visual features into one network, simplifying integration with Large Language Models (LLMs). Additionally, we have enhanced the fMRI dataset with diverse fMRI-image-related textual data to support multimodal large model development. Integrating with LLMs enhances decoding capabilities, enabling tasks such as brain captioning, complex reasoning, concept localization, and visual reconstruction. Our approach demonstrates superior performance across these tasks, precisely identifying language-based concepts within brain signals, enhancing interpretability, and providing deeper insights into neural processes. These advances significantly broaden the applicability of non-invasive brain decoding in neuroscience and human-computer interaction, setting the stage for advanced brain-computer interfaces and cognitive models.

Neuro-Vision to Language: Enhancing Brain Recording-based Visual Reconstruction and Language Interaction

NeurIPS.cc/2024/Conference

2404.19438

https://huggingface.co/papers/2404.19438

poster

https://openreview.net/forum?id=ogk236hsJM

@inproceedings{ luo2024onestep, title={One-Step Diffusion Distillation through Score Implicit Matching}, author={Weijian Luo and Zemin Huang and Zhengyang Geng and J Zico Kolter and Guo-Jun Qi}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ogk236hsJM} }

Despite their strong performances on many generative tasks, diffusion models require a large number of sampling steps in order to generate realistic samples. This has motivated the community to develop effective methods to distill pre-trained diffusion models into more efficient models, but these methods still typically require few-step inference or perform substantially worse than the underlying model. In this paper, we present Score Implicit Matching (SIM) a new approach to distilling pre-trained diffusion models into single-step generator models, while maintaining almost the same sample generation ability as the original model as well as being data-free with no need of training samples for distillation. The method rests upon the fact that, although the traditional score-based loss is intractable to minimize for generator models, under certain conditions we \emph{can} efficiently compute the \emph{gradients} for a wide class of score-based divergences between a diffusion model and a generator. SIM shows strong empirical performances for one-step generators: on the CIFAR10 dataset, it achieves an FID of 2.06 for unconditional generation and 1.96 for class-conditional generation. Moreover, by applying SIM to a leading transformer-based diffusion model, we distill a single-step generator for text-to-image (T2I) generation that attains an aesthetic score of 6.42 with no performance decline over the original multi-step counterpart, clearly outperforming the other one-step generators including SDXL-TURBO of 5.33, SDXL-LIGHTNING of 5.34 and HYPER-SDXL of 5.85. We will release this industry-ready one-step transformer-based T2I generator along with this paper.

One-Step Diffusion Distillation through Score Implicit Matching

NeurIPS.cc/2024/Conference

2410.16794

https://huggingface.co/papers/2410.16794

poster

https://openreview.net/forum?id=ogaeChzbKu

@inproceedings{ xian2024raw, title={{RAW}: A Robust and Agile Plug-and-Play Watermark Framework for {AI}-Generated Images with Provable Guarantees}, author={Xun Xian and Ganghua Wang and Xuan Bi and Jayanth Srinivasa and Ashish Kundu and Mingyi Hong and Jie Ding}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ogaeChzbKu} }

Safeguarding intellectual property and preventing potential misuse of AI-generated images are of paramount importance. This paper introduces a robust and agile plug-and-play watermark detection framework, referred to as RAW. As a departure from existing encoder-decoder methods, which incorporate fixed binary codes as watermarks within latent representations, our approach introduces learnable watermarks directly into the original image data. Subsequently, we employ a classifier that is jointly trained with the watermark to detect the presence of the watermark. The proposed framework is compatible with various generative architectures and supports on-the-fly watermark injection after training. By incorporating state-of-the-art smoothing techniques, we show that the framework also provides provable guarantees regarding the false positive rate for misclassifying a watermarked image, even in the presence of adversarial attacks targeting watermark removal. Experiments on a diverse range of images generated by state-of-the-art diffusion models demonstrate substantially improved watermark encoding speed and watermark detection performance, under adversarial attacks, while maintaining image quality. Our code is publicly available [here](https://github.com/jeremyxianx/RAWatermark).

RAW: A Robust and Agile Plug-and-Play Watermark Framework for AI-Generated Images with Provable Guarantees

NeurIPS.cc/2024/Conference

2403.18774

poster

https://openreview.net/forum?id=ofjTu2ktxO

@inproceedings{ chen2024carrot, title={Carrot and Stick: Eliciting Comparison Data and Beyond}, author={Yiling Chen and Shi Feng and Fang-Yi Yu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=ofjTu2ktxO} }

Comparison data elicited from people are fundamental to many machine learning tasks, including reinforcement learning from human feedback for large language models and estimating ranking models. They are typically subjective and not directly verifiable. How to truthfully elicit such comparison data from rational individuals? We design peer prediction mechanisms for eliciting comparison data using a bonus-penalty payment. Our design leverages on the strong stochastic transitivity for comparison data to create symmetrically strongly truthful mechanisms such that truth-telling 1) forms a strict Bayesian Nash equilibrium, and 2) yields the highest payment among all symmetric equilibria. Each individual only needs to evaluate one pair of items and report her comparison in our mechanism. We further extend the bonus-penalty payment concept to eliciting networked data, designing a symmetrically strongly truthful mechanism when agents’ private signals are sampled according to the Ising models. We provide the necessary and sufficient conditions for our bonus-penalty payment to have truth-telling as a strict Bayesian Nash equilibrium. Experiments on two real-world datasets further support our theoretical discoveries.

Carrot and Stick: Eliciting Comparison Data and Beyond

NeurIPS.cc/2024/Conference

2410.23243

poster

https://openreview.net/forum?id=oe7MfqFK1M

@inproceedings{ liu2024recovering, title={Recovering Complete Actions for Cross-dataset Skeleton Action Recognition}, author={Hanchao Liu and Yujiang Li and Tai-Jiang Mu and Shi-min Hu}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=oe7MfqFK1M} }

Despite huge progress in skeleton-based action recognition, its generalizability to different domains remains a challenging issue. In this paper, to solve the skeleton action generalization problem, we present a recover-and-resample augmentation framework based on a novel complete action prior. We observe that human daily actions are confronted with temporal mismatch across different datasets, as they are usually partial observations of their complete action sequences. By recovering complete actions and resampling from these full sequences, we can generate strong augmentations for unseen domains. At the same time, we discover the nature of general action completeness within large datasets, indicated by the per-frame diversity over time. This allows us to exploit two assets of transferable knowledge that can be shared across action samples and be helpful for action completion: boundary poses for determining the action start, and linear temporal transforms for capturing global action patterns. Therefore, we formulate the recovering stage as a two-step stochastic action completion with boundary pose-conditioned extrapolation followed by smooth linear transforms. Both the boundary poses and linear transforms can be efficiently learned from the whole dataset via clustering. We validate our approach on a cross-dataset setting with three skeleton action datasets, outperforming other domain generalization approaches by a considerable margin.

Recovering Complete Actions for Cross-dataset Skeleton Action Recognition

NeurIPS.cc/2024/Conference

2410.23641

poster

https://openreview.net/forum?id=oe5ZEqTOaz

@inproceedings{ guo2024classifierguided, title={Classifier-guided Gradient Modulation for Enhanced Multimodal Learning}, author={Zirun Guo and Tao Jin and Jingyuan Chen and Zhou Zhao}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, year={2024}, url={https://openreview.net/forum?id=oe5ZEqTOaz} }

Multimodal learning has developed very fast in recent years. However, during the multimodal training process, the model tends to rely on only one modality based on which it could learn faster, thus leading to inadequate use of other modalities. Existing methods to balance the training process always have some limitations on the loss functions, optimizers and the number of modalities and only consider modulating the magnitude of the gradients while ignoring the directions of the gradients. To solve these problems, in this paper, we present a novel method to balance multimodal learning with **C**lassifier-**G**uided **G**radient **M**odulation (CGGM), considering both the magnitude and directions of the gradients. We conduct extensive experiments on four multimodal datasets: UPMC-Food 101, CMU-MOSI, IEMOCAP and BraTS 2021, covering classification, regression and segmentation tasks. The results show that CGGM outperforms all the baselines and other state-of-the-art methods consistently, demonstrating its effectiveness and versatility. Our code is available at https://github.com/zrguo/CGGM.

Classifier-guided Gradient Modulation for Enhanced Multimodal Learning

NeurIPS.cc/2024/Conference

2411.01409