Datasets:

sswt

/

arxiv_sample_50K

like 0

Robust Transient Stability Constrained Optimal Power Flow with Power Flow Routers Considering Renewable Uncertainties ; This paper proposes a robust transient stability constrained optimal power flow problem that addresses renewable uncertainties by the coordination of generation redispatch and power flow router PFR tuning.PFR refers to a general type of networkside controller that enlarges the feasible region of the OPF problem. The coordination between networkside and generatorside control in the proposed model is more general than the traditional methods which focus on generation dispatch only. An offlineonline solution framework is developed to solve the problem efficiently. Under this framework the original problem is significantly simplified, so that we only need to solve a lowdimensional deterministic problem at the online stage to achieve realtime implementation with a high robustness level. The proposed method is verified on the modified New England 39bus system. Numerical results demonstrate that the proposed method is efficient and shows good performance on economy and robustness.

The Convolution Exponential and Generalized Sylvester Flows ; This paper introduces a new method to build linear flows, by taking the exponential of a linear transformation. This linear transformation does not need to be invertible itself, and the exponential has the following desirable properties it is guaranteed to be invertible, its inverse is straightforward to compute and the log Jacobian determinant is equal to the trace of the linear transformation. An important insight is that the exponential can be computed implicitly, which allows the use of convolutional layers. Using this insight, we develop new invertible transformations named convolution exponentials and graph convolution exponentials, which retain the equivariance of their underlying transformations. In addition, we generalize Sylvester Flows and propose Convolutional Sylvester Flows which are based on the generalization and the convolution exponential as basis change. Empirically, we show that the convolution exponential outperforms other linear transformations in generative flows on CIFAR10 and the graph convolution exponential improves the performance of graph normalizing flows. In addition, we show that Convolutional Sylvester Flows improve performance over residual flows as a generative flow model measured in loglikelihood.

Noise robustness and experimental demonstration of a quantum generative adversarial network for continuous distributions ; The potential advantage of machine learning in quantum computers is a topic of intense discussion in the literature. Theoretical, numerical and experimental explorations will most likely be required to understand its power. There has been different algorithms proposed to exploit the probabilistic nature of variational quantum circuits for generative modelling. In this paper, we employ a hybrid architecture for quantum generative adversarial networks QGANs and study their robustness in the presence of noise. We devise a simple way of adding different types of noise to the quantum generator circuit, and numerically simulate the noisy hybrid quantum generative adversarial networks HQGANs to learn continuous probability distributions, and show that the performance of HQGANs remain unaffected. We also investigate the effect of different parameters on the training time to reduce the computational scaling of the algorithm and simplify its deployment on a quantum computer. We then perform the training on Rigetti's Aspen42QA quantum processing unit, and present the results from the training. Our results pave the way for experimental exploration of different quantum machine learning algorithms on noisy intermediate scale quantum devices.

Image Augmentations for GAN Training ; Data augmentations have been widely studied to improve the accuracy and robustness of classifiers. However, the potential of image augmentation in improving GAN models for image synthesis has not been thoroughly investigated in previous studies. In this work, we systematically study the effectiveness of various existing augmentation techniques for GAN training in a variety of settings. We provide insights and guidelines on how to augment images for both vanilla GANs and GANs with regularizations, improving the fidelity of the generated images substantially. Surprisingly, we find that vanilla GANs attain generation quality on par with recent stateoftheart results if we use augmentations on both real and generated images. When this GAN training is combined with other augmentationbased regularization techniques, such as contrastive loss and consistency regularization, the augmentations further improve the quality of generated images. We provide new stateoftheart results for conditional generation on CIFAR10 with both consistency loss and contrastive loss as additional regularizations.

CoCon A SelfSupervised Approach for Controlled Text Generation ; Pretrained Transformerbased language models LMs display remarkable natural language generation capabilities. With their immense potential, controlling text generation of such LMs is getting attention. While there are studies that seek to control highlevel attributes such as sentiment and topic of generated text, there is still a lack of more precise control over its content at the word and phraselevel. Here, we propose ContentConditioner CoCon to control an LM's output text with a content input, at a finegrained level. In our selfsupervised approach, the CoCon block learns to help the LM complete a partiallyobserved text sequence by conditioning with content inputs that are withheld from the LM. Through experiments, we show that CoCon can naturally incorporate target content into generated texts and control highlevel text attributes in a zeroshot manner.

Experiments on route choice set generation using a large GPS trajectory set ; Several route choice models developed in the literature were based on a relatively small number of observations. With the extensive use of tracking devices in recent surveys, there is a possibility to obtain insights with respect to the traveler's choice behavior. In this paper, different path generation algorithms are evaluated using a large GPS trajectory dataset. The dataset contains 6,000 observations from TelAviv metropolitan area. An initial analysis is performed by generating a single route based on the shortest path. Almost 60 percent of the 6,000 observations can be covered assuming a threshold of 80 overlap using a single path. This result significantly contrasts previous literature findings. Link penalty, link elimination, simulation and vianode methods are applied to generate route sets, and the consistency of the algorithms are compared. A modified link penalty method, which accounts for preference of using higher hierarchical roads, provides a route set with 97 coverage 80 overlap threshold. The vianode method produces route set with satisfying coverage, and generates routes that are more heterogeneous in terms number of links and routes ratio.

Learning disconnected manifolds a no GANs land ; Typical architectures of Generative AdversarialNetworks make use of a unimodal latent distribution transformed by a continuous generator. Consequently, the modeled distribution always has connected support which is cumbersome when learning a disconnected set of manifolds. We formalize this problem by establishing a no free lunch theorem for the disconnected manifold learning stating an upper bound on the precision of the targeted distribution. This is done by building on the necessary existence of a lowquality region where the generator continuously samples data between two disconnected modes. Finally, we derive a rejection sampling method based on the norm of generators Jacobian and show its efficiency on several generators including BigGAN.

Endtoend Sinkhorn Autoencoder with Noise Generator ; In this work, we propose a novel endtoend sinkhorn autoencoder with noise generator for efficient data collection simulation. Simulating processes that aim at collecting experimental data is crucial for multiple reallife applications, including nuclear medicine, astronomy and high energy physics. Contemporary methods, such as Monte Carlo algorithms, provide highfidelity results at a price of high computational cost. Multiple attempts are taken to reduce this burden, e.g. using generative approaches based on Generative Adversarial Networks or Variational Autoencoders. Although such methods are much faster, they are often unstable in training and do not allow sampling from an entire data distribution. To address these shortcomings, we introduce a novel method dubbed endtoend Sinkhorn Autoencoder, that leverages sinkhorn algorithm to explicitly align distribution of encoded real data examples and generated noise. More precisely, we extend autoencoder architecture by adding a deterministic neural network trained to map noise from a known distribution onto autoencoder latent space representing data distribution. We optimise the entire model jointly. Our method outperforms competing approaches on a challenging dataset of simulation data from Zero Degree Calorimeters of ALICE experiment in LHC. as well as standard benchmarks, such as MNIST and CelebA.

Generating Master Faces for Use in Performing Wolf Attacks on Face Recognition Systems ; Due to its convenience, biometric authentication, especial face authentication, has become increasingly mainstream and thus is now a prime target for attackers. Presentation attacks and face morphing are typical types of attack. Previous research has shown that fingervein and fingerprintbased authentication methods are susceptible to wolf attacks, in which a wolf sample matches many enrolled user templates. In this work, we demonstrated that wolf generic faces, which we call master faces, can also compromise face recognition systems and that the master face concept can be generalized in some cases. Motivated by recent similar work in the fingerprint domain, we generated highquality master faces by using the stateoftheart face generator StyleGAN in a process called latent variable evolution. Experiments demonstrated that even attackers with limited resources using only pretrained models available on the Internet can initiate master face attacks. The results, in addition to demonstrating performance from the attacker's point of view, can also be used to clarify and improve the performance of face recognition systems and harden face authentication systems.

Overparameterization and generalization error weighted trigonometric interpolation ; Motivated by surprisingly good generalization properties of learned deep neural networks in overparameterized scenarios and by the related double descent phenomenon, this paper analyzes the relation between smoothness and low generalization error in an overparameterized linear learning problem. We study a random Fourier series model, where the task is to estimate the unknown Fourier coefficients from equidistant samples. We derive exact expressions for the generalization error of both plain and weighted least squares estimators. We show precisely how a bias towards smooth interpolants, in the form of weighted trigonometric interpolation, can lead to smaller generalization error in the overparameterized regime compared to the underparameterized regime. This provides insight into the power of overparameterization, which is common in modern machine learning.

DeshuffleGAN A SelfSupervised GAN to Improve Structure Learning ; Generative Adversarial Networks GANs triggered an increased interest in problem of image generation due to their improved output image quality and versatility for expansion towards new methods. Numerous GANbased works attempt to improve generation by architectural and lossbased extensions. We argue that one of the crucial points to improve the GAN performance in terms of realism and similarity to the original data distribution is to be able to provide the model with a capability to learn the spatial structure in data. To that end, we propose the DeshuffleGAN to enhance the learning of the discriminator and the generator, via a selfsupervision approach. Specifically, we introduce a deshuffling task that solves a puzzle of randomly shuffled image tiles, which in turn helps the DeshuffleGAN learn to increase its expressive capacity for spatial structure and realistic appearance. We provide experimental evidence for the performance improvement in generated images, compared to the baseline methods, which is consistently observed over two different datasets.

Fukaya categories of blowups ; We compute the Fukaya category of the symplectic blowup of a compact rational symplectic manifold at a point in the following sense Suppose a collection of Lagrangian branes satisfy Abouzaid's criterion for splitgeneration of a bulkdeformed Fukaya category of cleanlyintersecting Lagrangian branes. We show that for a small blowup parameter, their inverse images in the blowup together with a collection of branes near the exceptional locus splitgenerate the Fukaya category of the blowup. This categorifies a result on quantum cohomology by Bayer and is an example of a more general conjectural description of the behavior of the Fukaya category under transitions occuring in the minimal model program, namely that mmp transitions generate additional summands.

Delocalized SPM rogue waves in normal dispersion cascaded supercontinuum generation ; In the numerical modelling of cascaded midinfrared IR supercontinuum generation SCG we have studied how an ensemble of spectrally and temporally distributed solitons from the longwavelength part of an SC evolves and interacts when coupled into the normal dispersion regime of a highly nonlinear chalcogenide fiber. This has revealed a novel fundamental phenomenon the generation of a temporally and spectrally delocalized high energy rogue wave in the normal dispersion regime in the form of a strongly selfphasemodulation SPM broadened pulse. Along the local SPM shape the rogue wave is localized both temporally and spectrally. We demonstrate that this novel form of rogue wave is generated by interpulse Raman amplification between the SPM lobes of the many pulses causing the initially most delayed pulse to swallow the energy of all the other pulses. We further demonstrate that this novel type of rogue wave generation is a key effect in efficient longwavelength midIR SCG based on the cascading of SC spectra and demonstrate how the midIR SC spectrum can be shaped by manipulating the rogue wave.

Complementary Boundary Generator with ScaleInvariant Relation Modeling for Temporal Action Localization Submission to ActivityNet Challenge 2020 ; This technical report presents an overview of our solution used in the submission to ActivityNet Challenge 2020 Task 1 textbftemporal action localizationdetection. Temporal action localization requires to not only precisely locate the temporal boundaries of action instances, but also accurately classify the untrimmed videos into specific categories. In this paper, we decouple the temporal action localization task into two stages i.e. proposal generation and classification and enrich the proposal diversity through exhaustively exploring the influences of multiple components from different but complementary perspectives. Specifically, in order to generate highquality proposals, we consider several factors including the video feature encoder, the proposal generator, the proposalproposal relations, the scale imbalance, and ensemble strategy. Finally, in order to obtain accurate detections, we need to further train an optimal video classifier to recognize the generated proposals. Our proposed scheme achieves the stateoftheart performance on the temporal action localization task with textbf42.26 average mAP on the challenge testing set.

DeepSVG A Hierarchical Generative Network for Vector Graphics Animation ; Scalable Vector Graphics SVG are ubiquitous in modern 2D interfaces due to their ability to scale to different resolutions. However, despite the success of deep learningbased models applied to rasterized images, the problem of vector graphics representation learning and generation remains largely unexplored. In this work, we propose a novel hierarchical generative network, called DeepSVG, for complex SVG icons generation and interpolation. Our architecture effectively disentangles highlevel shapes from the lowlevel commands that encode the shape itself. The network directly predicts a set of shapes in a nonautoregressive fashion. We introduce the task of complex SVG icons generation by releasing a new largescale dataset along with an opensource library for SVG manipulation. We demonstrate that our network learns to accurately reconstruct diverse vector graphics, and can serve as a powerful animation tool by performing interpolations and other latent space operations. Our code is available at httpsgithub.comalexandre01deepsvg.

Nonparallel Emotion Conversion using a DeepGenerative Hybrid Network and an Adversarial Pair Discriminator ; We introduce a novel method for emotion conversion in speech that does not require parallel training data. Our approach loosely relies on a cycleGAN schema to minimize the reconstruction error from converting back and forth between emotion pairs. However, unlike the conventional cycleGAN, our discriminator classifies whether a pair of input real and generated samples corresponds to the desired emotion conversion e.g., A to B or to its inverse B to A. We will show that this setup, which we refer to as a variational cycleGAN VCGAN, is equivalent to minimizing the empirical KL divergence between the source features and their cyclic counterpart. In addition, our generator combines a trainable deep network with a fixed generative block to implement a smooth and invertible transformation on the input features, in our case, the fundamental frequency F0 contour. This hybrid architecture regularizes our adversarial training procedure. We use crowd sourcing to evaluate both the emotional saliency and the quality of synthesized speech. Finally, we show that our model generalizes to new speakers by modifying speech produced by Wavenet.

Harvesting Energy from Sun, Outer Space, and Soil ; While solar power systems have offered a wide variety of electricity generation approaches including photovoltaics, solar thermal power systems, and solar thermoelectric generators, the ability of generating electricity at both the daytime and nighttime with no necessity of energy storage still remains challenging. Here, we propose and verify a strategy of harvesting solar energy by solar heating during the daytime and harnessing the coldness of the outer space through radiative cooling to produce electricity at night using a commercial thermoelectric module. It enables electricity generation for 24 hours a day. We experimentally demonstrate a peak power density of 37 mWm2 at night and a peak value of 723 mWm2 during the daytime. A theoretical model that accurately predicts the performance of the device is developed and validated. The feature of 24hour electricity generation shows great potential energy applications of offgrid and batteryfree lighting and sensing.

EncoderDecoder Generative Adversarial Nets for Suffix Generation and Remaining Time Prediction of Business Process Models ; This paper proposes an encoderdecoder architecture grounded on Generative Adversarial Networks GANs, that generates a sequence of activities and their timestamps in an endtoend way. GANs work well with differentiable data such as images. However, a suffix is a sequence of categorical items. To this end, we use the GumbelSoftmax distribution to get a differentiable continuous approximation. The training works by putting one neural network against the other in a twoplayer game hence the adversarial nature, which leads to generating suffixes close to the ground truth. From the experimental evaluation it emerges that the approach is superior to the baselines in terms of the accuracy of the predicted suffixes and corresponding remaining times, despite using a naive feature encoding and only engineering features based on control flow and events completion time.

Meta Sequence Learning for Generating Adequate QuestionAnswer Pairs ; Creating multiplechoice questions to assess reading comprehension of a given article involves generating questionanswer pairs QAPs on the main points of the document. We present a learning scheme to generate adequate QAPs via metasequence representations of sentences. A meta sequence is a sequence of vectors comprising semantic and syntactic tags. In particular, we devise a scheme called MetaQA to learn meta sequences from training data to form pairs of a meta sequence for a declarative sentence MD and a corresponding interrogative sentence MIs. On a given declarative sentence, a trained MetaQA model converts it to a meta sequence, finds a matched MD, and uses the corresponding MIs and the input sentence to generate QAPs. We implement MetaQA for the English language using semanticrole labeling, partofspeech tagging, and namedentity recognition, and show that trained on a small dataset, MetaQA generates efficiently over the official SAT practice reading tests a large number of syntactically and semantically correct QAPs with over 97 accuracy.

Incorporating Behavioral Hypotheses for Query Generation ; Generative neural networks have been shown effective on query suggestion. Commonly posed as a conditional generation problem, the task aims to leverage earlier inputs from users in a search session to predict queries that they will likely issue at a later time. User inputs come in various forms such as querying and clicking, each of which can imply different semantic signals channeled through the corresponding behavioral patterns. This paper induces these behavioral biases as hypotheses for query generation, where a generic encoderdecoder Transformer framework is presented to aggregate arbitrary hypotheses of choice. Our experimental results show that the proposed approach leads to significant improvements on topk word error rate and Bert F1 Score compared to a recent BART model.

Generative Melody Composition with HumanintheLoop Bayesian Optimization ; Deep generative models allow even novice composers to generate various melodies by sampling latent vectors. However, finding the desired melody is challenging since the latent space is unintuitive and highdimensional. In this work, we present an interactive system that supports generative melody composition with humanintheloop Bayesian optimization BO. This system takes a mixedinitiative approach; the system generates candidate melodies to evaluate, and the user evaluates them and provides preferential feedback i.e., picking the best melody among the candidates to the system. This process is iteratively performed based on BO techniques until the user finds the desired melody. We conducted a pilot study using our prototype system, suggesting the potential of this approach.

Multipole decomposition of the general luminosity distance 'Hubble law' a new framework for observational cosmology ; We present the luminosity distance series expansion to third order in redshift for a general spacetime with no assumption on the metric tensor or the field equations prescribing it. It turns out that the coefficients of this general 'Hubble law' can be expressed in terms of a finite number of physically interpretable multipole coefficients. The multipole terms can be combined into effective direction dependent parameters replacing the Hubble constant, deceleration parameter, curvature parameter, and 'jerk' parameter of the FriedmannLemaitreRobertsonWalker FLRW class of metrics. Due to the finite number of multipole coefficients, the exact anisotropic Hubble law is given by 9, 25, 61 degrees of freedom in the mathcalOz, mathcalOz2, mathcalOz3 vicinity of the observer respectively, where z,redshift. This makes possible model independent determination of dynamical degrees of freedom of the cosmic neighbourhood of the observer and direct testing of the FLRW ansatz. We argue that the derived multipole representation of the general Hubble law provides a new framework with broad applications in observational cosmology.

Automorphic Forms and Fermion Masses ; We extend the framework of modular invariant supersymmetric theories to encompass invariance under more general discrete groups Gamma, that allow the presence of several moduli and make connection with the theory of automorphic forms. Moduli span a coset space GK, where G is a Lie group and K is a compact subgroup of G, modded out by Gamma. For a general choice of G, K, Gamma and a generic matter content, we explicitly construct a minimal Kahler potential and a general superpotential, for both rigid and local N1 supersymmetric theories. We also specialize our construction to the case GSp2g,R, KUg and GammaSp2g,Z, whose automorphic forms are Siegel modular forms. We show how our general theory can be consistently restricted to multidimensional regions of the moduli space enjoying residual symmetries. After choosing g2, we present several examples of models for lepton and quark masses where Yukawa couplings are Siegel modular forms of level 2.

Reflective Decoding Beyond Unidirectional Generation with OfftheShelf Language Models ; Publicly available, large pretrained LanguageModels LMs generate text with remarkable quality, but only sequentially from left to right. As a result, they are not immediately applicable to generation tasks that break the unidirectional assumption, such as paraphrasing or textinfilling, necessitating taskspecific supervision. In this paper, we present Reflective Decoding, a novel unsupervised algorithm that allows for direct application of unidirectional LMs to nonsequential tasks. Our 2step approach requires no supervision or even parallel corpora, only two offtheshelf pretrained LMs in opposite directions forward and backward. First, in the contextualization step, we use LMs to generate ensembles of past and future contexts which collectively capture the input e.g. the source sentence for paraphrasing. Second, in the reflection step, we condition on these context ensembles, generating outputs that are compatible with them. Comprehensive empirical results demonstrate that Reflective Decoding outperforms strong unsupervised baselines on both paraphrasing and abductive text infilling, significantly narrowing the gap between unsupervised and supervised methods. Reflective Decoding surpasses multiple supervised baselines on various metrics including human evaluation.

CRWalker TreeStructured Graph Reasoning and Dialog Acts for Conversational Recommendation ; Growing interests have been attracted in Conversational Recommender Systems CRS, which explore user preference through conversational interactions in order to make appropriate recommendation. However, there is still a lack of ability in existing CRS to 1 traverse multiple reasoning paths over background knowledge to introduce relevant items and attributes, and 2 arrange selected entities appropriately under current system intents to control response generation. To address these issues, we propose CRWalker in this paper, a model that performs treestructured reasoning on a knowledge graph, and generates informative dialog acts to guide language generation. The unique scheme of treestructured reasoning views the traversed entity at each hop as part of dialog acts to facilitate language generation, which links how entities are selected and expressed. Automatic and human evaluations show that CRWalker can arrive at more accurate recommendation, and generate more informative and engaging responses.

Frequency Response Study on the ERCOT under High Photovoltaic PV Penetration Conditions ; Solar photovoltaic PV generation is growing rapidly around the world. However, PV generation, based on inverter, is fundamentally different from conventional synchronous generators. It is of vital importance to understand the impacts of increased penetration of PV generation on power system dynamic performance. This paper investigates frequency response of the Electric Reliability Council of Texas ERCOT with high PV penetration in the future year. In this work, a realistic baseline dynamic model is validated using synchrophasor measurements. Then, dynamic simulation is performed to evaluate the impacts of high PV generation on frequency response.

MetaLearning for Domain Generalization in Semantic Parsing ; The importance of building semantic parsers which can be applied to new domains and generate programs unseen at training has long been acknowledged, and datasets testing outofdomain performance are becoming increasingly available. However, little or no attention has been devoted to learning algorithms or objectives which promote domain generalization, with virtually all existing approaches relying on standard supervised learning. In this work, we use a metalearning framework which targets zeroshot domain generalization for semantic parsing. We apply a modelagnostic training algorithm that simulates zeroshot parsing by constructing virtual train and test sets from disjoint domains. The learning objective capitalizes on the intuition that gradient steps that improve sourcedomain performance should also improve targetdomain performance, thus encouraging a parser to generalize to unseen target domains. Experimental results on the English Spider and Chinese Spider datasets show that the metalearning objective significantly boosts the performance of a baseline parser.

Experimental investigation of tsunami waves generated by granular collapse into water ; The generation of a tsunami wave by an aerial landslide is investigated through model laboratory experiments. We examine the collapse of an initially dry column of grains into a shallow water layer and the subsequent generation of waves. The experiments show that the collective entry of the granular material into water governs the wave generation process. We observe that the amplitude of the wave relative to the water height scales linearly with the Froude number based on the horizontal velocity of the moving granular front relative to the wave velocity. For all the different parameters considered here, the aspect ratio and the volume of the column, the diameter and density of the grains, and the height of the water, the granular collapse acts like a moving piston displacing the water. We also highlight that the density of the falling grains has a negligible influence on the wave amplitude, which suggests that the volume of grains entering the water is the relevant parameter in the wave generation.

Constrained Abstractive Summarization Preserving Factual Consistency with Constrained Generation ; Despite significant progress, stateoftheart abstractive summarization methods are still prone to hallucinate content inconsistent with the source document. In this paper, we propose Constrained Abstractive Summarization CAS, a general setup that preserves the factual consistency of abstractive summarization by specifying tokens as constraints that must be present in the summary. We adopt lexically constrained decoding, a technique generally applicable to autoregressive generative models, to fulfill CAS and conduct experiments in two scenarios 1 automatic summarization without human involvement, where keyphrases are extracted from the source document and used as constraints; 2 humanguided interactive summarization, where human feedback in the form of manual constraints are used to guide summary generation. Automatic and human evaluations on two benchmark datasets demonstrate that CAS improves both lexical overlap ROUGE and factual consistency of abstractive summarization. In particular, we observe up to 13.8 ROUGE2 gains when only one manual constraint is used in interactive summarization.

GENs Generative Encoding Networks ; Mapping data from andor onto a known family of distributions has become an important topic in machine learning and data analysis. Deep generative models e.g., generative adversarial networks have been used effectively to match known and unknown distributions. Nonetheless, when the form of the target distribution is known, analytical methods are advantageous in providing robust results with provable properties. In this paper, we propose and analyze the use of nonparametric density methods to estimate the JensenShannon divergence for matching unknown data distributions to known target distributions, such Gaussian or mixtures of Gaussians, in latent spaces. This analytical method has several advantages better behavior when training sample quantity is low, provable convergence properties, and relatively few parameters, which can be derived analytically. Using the proposed method, we enforce the latent representation of an autoencoder to match a target distribution in a learning framework that we call a em generative encoding network. Here, we present the numerical methods; derive the expected distribution of the data in the latent space; evaluate the properties of the latent space, sample reconstruction, and generated samples; show the advantages over the adversarial counterpart; and demonstrate the application of the method in real world.

Exploring QuestionSpecific Rewards for Generating Deep Questions ; Recent question generation QG approaches often utilize the sequencetosequence framework Seq2Seq to optimize the loglikelihood of groundtruth questions using teacher forcing. However, this training objective is inconsistent with actual question quality, which is often reflected by certain global properties such as whether the question can be answered by the document. As such, we directly optimize for QGspecific objectives via reinforcement learning to improve question quality. We design three different rewards that target to improve the fluency, relevance, and answerability of generated questions. We conduct both automatic and human evaluations in addition to a thorough analysis to explore the effect of each QGspecific reward. We find that optimizing questionspecific rewards generally leads to better performance in automatic evaluation metrics. However, only the rewards that correlate well with human judgement e.g., relevance lead to real improvement in question quality. Optimizing for the others, especially answerability, introduces incorrect bias to the model, resulting in poor question quality. Our code is publicly available at httpsgithub.comYuxiXieRLforQuestionGeneration.

Generating Synthetic Data for TaskOriented Semantic Parsing with Hierarchical Representations ; Modern conversational AI systems support natural language understanding for a wide variety of capabilities. While a majority of these tasks can be accomplished using a simple and flat representation of intents and slots, more sophisticated capabilities require complex hierarchical representations supported by semantic parsing. Stateoftheart semantic parsers are trained using supervised learning with data labeled according to a hierarchical schema which might be costly to obtain or not readily available for a new domain. In this work, we explore the possibility of generating synthetic data for neural semantic parsing using a pretrained denoising sequencetosequence model i.e., BART. Specifically, we first extract masked templates from the existing labeled utterances, and then finetune BART to generate synthetic utterances conditioning on the extracted templates. Finally, we use an auxiliary parser AP to filter the generated utterances. The AP guarantees the quality of the generated data. We show the potential of our approach when evaluating on the Facebook TOP dataset for navigation domain.

Conditioned Text Generation with Transfer for ClosedDomain Dialogue Systems ; Scarcity of training data for taskoriented dialogue systems is a well known problem that is usually tackled with costly and timeconsuming manual data annotation. An alternative solution is to rely on automatic text generation which, although less accurate than human supervision, has the advantage of being cheap and fast. Our contribution is twofold. First we show how to optimally train and control the generation of intentspecific sentences using a conditional variational autoencoder. Then we introduce a new protocol called query transfer that allows to leverage a large unlabelled dataset, possibly containing irrelevant queries, to extract relevant information. Comparison with two different baselines shows that this method, in the appropriate regime, consistently improves the diversity of the generated queries without compromising their quality. We also demonstrate the effectiveness of our generation method as a data augmentation technique for language modelling tasks.

An Efficient Scheme for the Generation of Ordered Trees in Constant Amortized Time ; Trees are useful entities allowing to model data structures and hierarchical relationships in networked decision systems ubiquitously. An ordered tree is a rooted tree where the order of the subtrees children of a node is significant. In combinatorial optimization, generating ordered trees is relevant to evaluate candidate combinatorial objects. In this paper, we present an algebraic scheme to generate ordered trees with n vertices with utmost efficiency; whereby our approach uses mathcalOn space and mathcalO1 time in average per tree. Our computational studies have shown the feasibility and efficiency to generate ordered trees in constant time in average, in about one tenth of a millisecond per ordered tree. Due to the 11 bijective nature to other combinatorial classes, our approach is favorable to study the generation of binary trees with n external nodes, trees with n nodes, legal sequences of n pairs of parentheses, triangulated ngons, gambler's sequences and lattice paths. We believe our scheme may find its use in devising algorithms for planning and combinatorial optimization involving Catalan numbers.

Encoding large scale cosmological structure with Generative Adversarial Networks ; Recently a type of neural networks called Generative Adversarial Networks GANs has been proposed as a solution for fast generation of simulationlike datasets, in an attempt to bypass heavy computations and expensive cosmological simulations to run in terms of time and computing power. In the present work, we build and train a GAN to look further into the strengths and limitations of such an approach. We then propose a novel method in which we make use of a trained GAN to construct a simple autoencoder AE as a first step towards building a predictive model. Both the GAN and AE are trained on images issued from two types of Nbody simulations, namely 2D and 3D simulations. We find that the GAN successfully generates new images that are statistically consistent with the images it was trained on. We then show that the AE manages to efficiently extract information from simulation images, satisfyingly inferring the latent encoding of the GAN to generate an image with similar large scale structures.

PCGAIN Pseudolabel Conditional Generative Adversarial Imputation Networks for Incomplete Data ; Datasets with missing values are very common in real world applications. GAIN, a recently proposed deep generative model for missing data imputation, has been proved to outperform many stateoftheart methods. But GAIN only uses a reconstruction loss in the generator to minimize the imputation error of the nonmissing part, ignoring the potential category information which can reflect the relationship between samples. In this paper, we propose a novel unsupervised missing data imputation method named PCGAIN, which utilizes potential category information to further enhance the imputation power. Specifically, we first propose a pretraining procedure to learn potential category information contained in a subset of lowmissingrate data. Then an auxiliary classifier is determined using the synthetic pseudolabels. Further, this classifier is incorporated into the generative adversarial framework to help the generator to yield higher quality imputation results. The proposed method can improve the imputation quality of GAIN significantly. Experimental results on various benchmark datasets show that our method is also superior to other baseline approaches. Our code is available at urlhttpsgithub.comWYuFengpcgain.

TFGAN Time and Frequency Domain Based Generative Adversarial Network for Highfidelity Speech Synthesis ; Recently, GAN based speech synthesis methods, such as MelGAN, have become very popular. Compared to conventional autoregressive based methods, parallel structures based generators make waveform generation process fast and stable. However, the quality of generated speech by autoregressive based neural vocoders, such as WaveRNN, is still higher than GAN. To address this issue, we propose a novel vocoder model TFGAN, which is adversarially learned both in time and frequency domain. On one hand, we propose to discriminate groundtruth waveform from synthetic one in frequency domain for offering more consistency guarantees instead of only in time domain. On the other hand, in contrast to the conventionally frequencydomain STFT loss approach or feature map loss by discriminator to learn waveform, we propose a set of timedomain loss that encourage the generator to capture the waveform directly. TFGAN has nearly same synthesis speed as MelGAN, but the fidelity is significantly improved by our novel learning method. In our experiments, TFGAN shows the ability to achieve comparable mean opinion score MOS than autoregressive vocoder under speech synthesis context.

Decay A Monte Carlo library for the decay of a particle with ROOT compatibility ; Recently, there is a need for a generalpurpose event generator of decays of an elementary particle or a hadron to a state of higher multiplicity N 2 that is simple to use and universal. We present the structure of such a library to produce generators that generate kinematics of decay processes and can be used to integrate any matrix element squared over phase space of this decay. Some test examples are presented, and results are compared with results known from the literature. As one of examples we consider the Standard Model Higgs boson decay into four leptons. The generators discussed here are compatible with the ROOT interface.

Demonstrationefficient Inverse Reinforcement Learning in Procedurally Generated Environments ; Deep Reinforcement Learning achieves very good results in domains where reward functions can be manually engineered. At the same time, there is growing interest within the community in using games based on Procedurally Content Generation PCG as benchmark environments since this type of environment is perfect for studying overfitting and generalization of agents under domain shift. Inverse Reinforcement Learning IRL can instead extrapolate reward functions from expert demonstrations, with good results even on highdimensional problems, however there are no examples of applying these techniques to procedurallygenerated environments. This is mostly due to the number of demonstrations needed to find a good reward model. We propose a technique based on Adversarial Inverse Reinforcement Learning which can significantly decrease the need for expert demonstrations in PCG games. Through the use of an environment with a limited set of initial seed levels, plus some modifications to stabilize training, we show that our approach, DEAIRL, is demonstrationefficient and still able to extrapolate reward functions which generalize to the fully procedural domain. We demonstrate the effectiveness of our technique on two procedural environments, MiniGrid and DeepCrawl, for a variety of tasks.

ScalarConnection Gravity and Spontaneous Scalarization ; Scalartensor theories of gravity are known to allow significant deviations from general relativity through various astrophysical phenomena. In this paper, we formulate a scalarconnection gravity by setting up scalars and connection configurations instead of metric. Since the matter sector is not straightforward to conceive without a metric, we invoke cosmological fluids in terms of their oneform velocity in the volume element of the invariant action. This leads to gravitational equations with a perfect fluid source and a generated metric, which are expected to produce reasonable deviations from general relativity in the strongfield regime. As a relevant application, we study spontaneous scalarization mechanism and show that the DamourEspositoFarese model arises in a certain class of scalarconnection gravity. Furthermore, we investigate a general study in which the present framework becomes distinguishable from the famed scalartensor theories.

The StyleContent Duality of Attractiveness Learning to Write EyeCatching Headlines via Disentanglement ; Eyecatching headlines function as the first device to trigger more clicks, bringing reciprocal effect between producers and viewers. Producers can obtain more traffic and profits, and readers can have access to outstanding articles. When generating attractive headlines, it is important to not only capture the attractive content but also follow an eyecatching written style. In this paper, we propose a Disentanglementbased Attractive Headline Generator DAHG that generates headline which captures the attractive content following the attractive style. Concretely, we first devise a disentanglement module to divide the style and content of an attractive prototype headline into latent spaces, with two auxiliary constraints to ensure the two spaces are indeed disentangled. The latent content information is then used to further polish the document representation and help capture the salient part. Finally, the generator takes the polished document as input to generate headline under the guidance of the attractive style. Extensive experiments on the public Kuaibao dataset show that DAHG achieves stateoftheart performance. Human evaluation also demonstrates that DAHG triggers 22 more clicks than existing models.

A PACBayesian Approach to Generalization Bounds for Graph Neural Networks ; In this paper, we derive generalization bounds for the two primary classes of graph neural networks GNNs, namely graph convolutional networks GCNs and message passing GNNs MPGNNs, via a PACBayesian approach. Our result reveals that the maximum node degree and spectral norm of the weights govern the generalization bounds of both models. We also show that our bound for GCNs is a natural generalization of the results developed in arXiv1707.09564v2 cs.LG for fullyconnected and convolutional neural networks. For message passing GNNs, our PACBayes bound improves over the Rademacher complexity based bound in arXiv2002.06157v1 cs.LG, showing a tighter dependency on the maximum node degree and the maximum hidden dimension. The key ingredients of our proofs are a perturbation analysis of GNNs and the generalization of PACBayes analysis to nonhomogeneous GNNs. We perform an empirical study on several realworld graph datasets and verify that our PACBayes bound is tighter than others.

A Note on Generalized Algebraic Theories and Categories with Families ; We give a new syntax independent definition of the notion of a generalized algebraic theory as an initial object in a category of categories with families cwfs with extra structure. To this end we define inductively how to build a valid signature Sigma for a generalized algebraic theory and the associated category of cwfs with a Sigmastructure and cwfmorphisms that preserve this structure on the nose. Our definition refers to uniform families of contexts, types, and terms, a purely semantic notion. Furthermore, we show how to syntactically construct initial cwfs with Sigmastructures. This result can be viewed as a generalization of Birkhoff's completeness theorem for equational logic. It is obtained by extending Castellan, Clairambault, and Dybjer's construction of an initial cwf. We provide examples of generalized algebraic theories for monoids, categories, categories with families, and categories with families with extra structure for some type formers of dependent type theory. The models of these are internal monoids, internal categories, and internal categories with families with extra structure in a category with families.

Contributions of the Cartan generators in potentials between static sources ; We investigate the contributions of the Cartan generators in the static potentials for various representations in the framework of the domain model of center vortices for SU3 gauge theory. Using the center domains with the cores corresponding to only one Cartan generator H8, already given as a particular proposal, leads to some concavities in the potentials for higher representations. Furthermore, the string tension of the fundamental representation is the same at Casimir scaling and Nality regimes. We add the contribution of the other Cartan generator H3 to the potentials and therefore these shortcomings can be eliminated. However, we discuss that at intermediate range of distances the potentials induced by only H8 agree with the Casimir scaling better than those corresponding to both Cartan generators.

GuidedStyle Attribute Knowledge Guided Style Manipulation for Semantic Face Editing ; Although significant progress has been made in synthesizing highquality and visually realistic face images by unconditional Generative Adversarial Networks GANs, there still lacks of control over the generation process in order to achieve semantic face editing. In addition, it remains very challenging to maintain other face information untouched while editing the target attributes. In this paper, we propose a novel learning framework, called GuidedStyle, to achieve semantic face editing on StyleGAN by guiding the image generation process with a knowledge network. Furthermore, we allow an attention mechanism in StyleGAN generator to adaptively select a single layer for style manipulation. As a result, our method is able to perform disentangled and controllable edits along various attributes, including smiling, eyeglasses, gender, mustache and hair color. Both qualitative and quantitative results demonstrate the superiority of our method over other competing methods for semantic face editing. Moreover, we show that our model can be also applied to different types of real and artistic face editing, demonstrating strong generalization ability.

On Generating Extended Summaries of Long Documents ; Prior work in document summarization has mainly focused on generating short summaries of a document. While this type of summary helps get a highlevel view of a given document, it is desirable in some cases to know more detailed information about its salient points that can't fit in a short summary. This is typically the case for longer documents such as a research paper, legal document, or a book. In this paper, we present a new method for generating extended summaries of long papers. Our method exploits hierarchical structure of the documents and incorporates it into an extractive summarization model through a multitask learning approach. We then present our results on three long summarization datasets, arXivLong, PubMedLong, and Longsumm. Our method outperforms or matches the performance of strong baselines. Furthermore, we perform a comprehensive analysis over the generated results, shedding insights on future research for longform summary generation task. Our analysis shows that our multitasking approach can adjust extraction probability distribution to the favor of summaryworthy sentences across diverse sections. Our datasets, and codes are publicly available at httpsgithub.comGeorgetownIRLabExtendedSumm

Narration Generation for Cartoon Videos ; Research on text generation from multimodal inputs has largely focused on static images, and less on video data. In this paper, we propose a new task, narration generation, that is complementing videos with narration texts that are to be interjected in several places. The narrations are part of the video and contribute to the storyline unfolding in it. Moreover, they are contextinformed, since they include information appropriate for the timeframe of video they cover, and also, do not need to include every detail shown in input scenes, as a caption would. We collect a new dataset from the animated television series Peppa Pig. Furthermore, we formalize the task of narration generation as including two separate tasks, timing and content generation, and present a set of models on the new task.

Information theoretic results for stationary time series and the Gaussiangeneralized von Mises time series ; This chapter presents some novel information theoretic results for the analysis of stationary time series in the frequency domain. In particular, the spectral distribution that corresponds to the most uncertain or unpredictable time series with some values of the autocovariance function fixed, is the generalized von Mises spectral distribution. It is thus a maximum entropy spectral distribution and the corresponding stationary time series is called the generalized von Mises time series. The generalized von Mises distribution is used in directional statistics for modelling planar directions that follow a multimodal distribution. Furthermore, the Gaussiangeneralized von Mises times series is presented as the stationary time series that maximizes entropies in frequency and time domains, respectively referred to as spectral and temporal entropies. Parameter estimation and some computational aspects with this time series are briefly analyzed.

Explanation as a Defense of Recommendation ; Textual explanations have proved to help improve user satisfaction on machinemade recommendations. However, current mainstream solutions loosely connect the learning of explanation with the learning of recommendation for example, they are often separately modeled as rating prediction and content generation tasks. In this work, we propose to strengthen their connection by enforcing the idea of sentiment alignment between a recommendation and its corresponding explanation. At training time, the two learning tasks are joined by a latent sentiment vector, which is encoded by the recommendation module and used to make word choices for explanation generation. At both training and inference time, the explanation module is required to generate explanation text that matches sentiment predicted by the recommendation module. Extensive experiments demonstrate our solution outperforms a rich set of baselines in both recommendation and explanation tasks, especially on the improved quality of its generated explanations. More importantly, our user studies confirm our generated explanations help users better recognize the differences between recommended items and understand why an item is recommended.

LAIF AI, Deep Learning for Germany Suetterlin Letter Recognition and Generation ; One of the successful early implementation of deep learning AI technology was on letter recognition. With the recent breakthrough of artificial intelligence AI brings more solid technology for complex problems like handwritten letter recognition and even automatic generation of them. In this research, we proposed deep learning framework called Ludwig AI FrameworkLAIF for Germany Suetterlin letter recognition and generation. To recognize Suetterlin letter, we proposed deep convolutional neural network. Since lack of big amount of data to train for the deep models and huge cost to label existing hard copy of handwritten letters, we also introduce the methodology with deep generative adversarial network to generate handwritten letters as synthetic data. Main source code is in httpsgithub.comenkhtogtokhLAIF repository.

A Robust Adversarial NetworkBased EndtoEnd Communications System With Strong Generalization Ability Against Adversarial Attacks ; We propose a novel defensive mechanism based on a generative adversarial network GAN framework to defend against adversarial attacks in endtoend communications systems. Specifically, we utilize a generative network to model a powerful adversary and enable the endtoend communications system to combat the generative attack network via a minimax game. We show that the proposed system not only works well against whitebox and blackbox adversarial attacks but also possesses excellent generalization capabilities to maintain good performance under no attacks. We also show that our GANbased endtoend system outperforms the conventional communications system and the endtoend communications system withwithout adversarial training.

Distributed Optimal Load Frequency Control with Stochastic Wind Power Generation ; Motivated by the inadequacy of conventional control methods for power networks with a large share of renewable generation, in this paper we study the stochastic passivity property of wind turbines based on the Doubly Fed Induction Generator DFIG. Differently from the majority of the results in the literature, where renewable generation is ignored or assumed to be constant, we model wind power generation as a stochastic process, where wind speed is described by a class of stochastic differential equations. Then, we design a distributed control scheme that achieves load frequency control and economic dispatch, ensuring the stochastic stability of the controlled network.

Improving Global Adversarial Robustness Generalization With Adversarially Trained GAN ; Convolutional neural networks CNNs have achieved beyond humanlevel accuracy in the image classification task and are widely deployed in realworld environments. However, CNNs show vulnerability to adversarial perturbations that are welldesigned noises aiming to mislead the classification models. In order to defend against the adversarial perturbations, adversarially trained GAN ATGAN is proposed to improve the adversarial robustness generalization of the stateoftheart CNNs trained by adversarial training. ATGAN incorporates adversarial training into standard GAN training procedure to remove obfuscated gradients which can lead to a false sense in defending against the adversarial perturbations and are commonly observed in existing GANsbased adversarial defense methods. Moreover, ATGAN adopts the imagetoimage generator as data augmentation to increase the sample complexity needed for adversarial robustness generalization in adversarial training. Experimental results in MNIST SVHN and CIFAR10 datasets show that the proposed method doesn't rely on obfuscated gradients and achieves better global adversarial robustness generalization performance than the adversarially trained stateoftheart CNNs.

Conversational Answer Generation and Factuality for Reading Comprehension QuestionAnswering ; Question answering QA is an important use case on voice assistants. A popular approach to QA is extractive reading comprehension RC which finds an answer span in a text passage. However, extractive answers are often unnatural in a conversational context which results in suboptimal user experience. In this work, we investigate conversational answer generation for QA. We propose AnswerBART, an endtoend generative RC model which combines answer generation from multiple passages with passage ranking and answerability. Moreover, a hurdle in applying generative RC are hallucinations where the answer is factually inconsistent with the passage text. We leverage recent work from summarization to evaluate factuality. Experiments show that AnswerBART significantly improves over previous best published results on MS MARCO 2.1 NLGEN by 2.5 ROUGEL and NarrativeQA by 9.4 ROUGEL.

Constrained Text Generation with Global Guidance Case Study on CommonGen ; This paper studies constrained text generation, which is to generate sentences under certain preconditions. We focus on CommonGen, the task of generating text based on a set of concepts, as a representative task of constrained text generation. Traditional methods mainly rely on supervised training to maximize the likelihood of target sentences.However, global constraints such as common sense and coverage cannot be incorporated into the likelihood objective of the autoregressive decoding process. In this paper, we consider using reinforcement learning to address the limitation, measuring global constraints including fluency, common sense and concept coverage with a comprehensive score, which serves as the reward for reinforcement learning. Besides, we design a guided decoding method at the word, fragment and sentence levels. Experiments demonstrate that our method significantly increases the concept coverage and outperforms existing models in various automatic evaluations.

Empirical Analysis of Capacity Investment Solution in Distribution Grids ; This paper presents an analysis of the stability and quality of the distributed generation planning problem's investment solution. The entry of distributed generators power based on nonconventional energy sources has been extensively promoted in distribution grids. In this paper, a twostage stochastic programming model is used to find the optimal distributed generators' installed capacities. We emphasize the design of scenarios to represent the stochasticity of power production on renewable sources. In the scenario generation, a method is proposed based on the clustering of real measurements of meteorological variables. We measure the quality and stability of the investment solution as a function of the number of scenarios. The results show that a reduced selection of scenarios can give an inadequate solution to distributed generators' investment strategy.

Fully Convolutional Scene Graph Generation ; This paper presents a fully convolutional scene graph generation FCSGG model that detects objects and relations simultaneously. Most of the scene graph generation frameworks use a pretrained twostage object detector, like Faster RCNN, and build scene graphs using bounding box features. Such pipeline usually has a large number of parameters and low inference speed. Unlike these approaches, FCSGG is a conceptually elegant and efficient bottomup approach that encodes objects as bounding box center points, and relationships as 2D vector fields which are named as Relation Affinity Fields RAFs. RAFs encode both semantic and spatial features, and explicitly represent the relationship between a pair of objects by the integral on a subregion that points from subject to object. FCSGG only utilizes visual features and still generates strong results for scene graph generation. Comprehensive experiments on the Visual Genome dataset demonstrate the efficacy, efficiency, and generalizability of the proposed method. FCSGG achieves highly competitive results on recall and zeroshot recall with significantly reduced inference time.

On the Arithmetic Fundamental Lemma conjecture over a general padic field ; We prove the arithmetic fundamental lemma conjecture over a general padic field with odd residue cardinality qgeq dim V. Our strategy is similar to the one used by the second author during his proof of the AFL over mathbbQp arXiv1909.02697, but only requires the modularity of divisor generating series on the Shimura variety as opposed to its integral model. The resulting increase in flexibility allows us to work over an arbitrary base field. To carry out the strategy, we also generalize results of Howard arXiv1303.0545 on CMcycle intersection and of EhlenSankaran arXiv1607.06545 on Green function comparison from mathbbQ to general totally real base fields.

Open Domain Generalization with DomainAugmented MetaLearning ; Leveraging datasets available to learn a model with high generalization ability to unseen domains is important for computer vision, especially when the unseen domain's annotated data are unavailable. We study a novel and practical problem of Open Domain Generalization OpenDG, which learns from different source domains to achieve high performance on an unknown target domain, where the distributions and label sets of each individual source domain and the target domain can be different. The problem can be generally applied to diverse source domains and widely applicable to realworld applications. We propose a DomainAugmented MetaLearning framework to learn opendomain generalizable representations. We augment domains on both featurelevel by a new Dirichlet mixup and labellevel by distilled softlabeling, which complements each domain with missing classes and other domain knowledge. We conduct metalearning over domains by designing new metalearning tasks and losses to preserve domain unique knowledge and generalize knowledge across domains simultaneously. Experiment results on various multidomain datasets demonstrate that the proposed DomainAugmented MetaLearning DAML outperforms prior methods for unseen domain recognition.

Effective field theories and inflationary magnetogenesis ; The effective approach is applied to the analysis of inflationary magnetogenesis. Rather than assuming a particular underlying description, all the generally covariant terms potentially appearing with four spacetime derivatives in the effective action have been included and weighted by inflatondependent couplings. The higher derivatives are suppressed by the negative powers of a typical mass scale whose specific values ultimately depend on the tensor to scalar ratio. During a quaside Sitter stage the corresponding corrections always lead to an asymmetry between the hypermagnetic and the hyperelectric susceptibilities. After presenting a general method for the estimate of the gauge power spectra, the obtained results are illustrated for generic models and also in the case of some nongeneric scenarios where either the inflaton has some extra symmetry or the higherorder terms are potentially dominant.

Polarization dynamics of ultrafast solitons ; We study the polarization dynamics of ultrafast solitons in modelocked fiber lasers. We find that when a stable soliton is generated, it's stateofpolarization shifts toward a stable state and when the soliton is generated with excess power levels it experiences relaxation oscillations in its intensity and timing. On the other hand, when a soliton is generated in an unstable stateofpolarization, it either decays in intensity until it disappears, or its temporal width decreases until it explodes into several solitons and then it disappears. We also found that when two solitons are simultaneously generated close to each other, they attract each other until they collide and merge into a single soliton. Although, these two solitons are generated with different statesofpolarization, they shift their stateofpolarization closer to each other until the polarization coincides when they collide. We support our findings by numerical calculations of a nonLagrangian approach by simulating the GinzburgLandau equation governing the dynamics of solitons in a laser cavity. Our model also predicts the relaxation oscillations of stable solitons and the two types of unstable solitons observed in the experimental measurements.

AdaptiFont Increasing Individuals' Reading Speed with a Generative Font Model and Bayesian Optimization ; Digital text has become one of the primary ways of exchanging knowledge, but text needs to be rendered to a screen to be read. We present AdaptiFont, a humanintheloop system that is aimed at interactively increasing readability of text displayed on a monitor. To this end, we first learn a generative font space with nonnegative matrix factorization from a set of classic fonts. In this space we generate new truetypefonts through active learning, render texts with the new font, and measure individual users' reading speed. Bayesian optimization sequentially generates new fonts on the fly to progressively increase individuals' reading speed. The results of a user study show that this adaptive font generation system finds regions in the font space corresponding to high reading speeds, that these fonts significantly increase participants' reading speed, and that the found fonts are significantly different across individual readers.

Improved gravitationalwave constraints on higherorder curvature theories of gravity ; Gravitational wave observations of compact binaries allow us to test general relativity and modifications thereof in the strong and highlydynamical field regime of gravity. Here we confront two extensions to general relativity, dynamical ChernSimons and EinsteindilatonGaussBonnet theories, against the gravitational wave sources from the GWTC1 and GWTC2 catalogs by the LIGOVirgo Collaboration. By stacking the posterior of individual events, we strengthen the constraint on the square root of the coupling parameter in EinsteindilatonGaussBonnet gravity to sqrtalpharm tiny EdGB 1.7 km, but we are unable to place meaningful constraints on dynamical ChernSimons gravity. Importantly, we also show that our bounds are robust to i the choice of generalrelativity base waveform model, upon which we add modifications, ii unknown higher postNewtonian order terms in the modifications to general relativity, iii the smallcoupling approximation, and iv uncertainties on the nature of the constituent compact objects.

Invariant polynomials and machine learning ; We present an application of invariant polynomials in machine learning. Using the methods developed in previous work, we obtain two types of generators of the Lorentz and permutationinvariant polynomials in particle momenta; minimal algebra generators and Hironaka decompositions. We discuss and prove some approximation theorems to make use of these invariant generators in machine learning algorithms in general and in neural networks specifically. By implementing these generators in neural networks applied to regression tasks, we test the improvements in performance under a wide range of hyperparameter choices and find a reduction of the loss on training data and a significant reduction of the loss on validation data. For a different approach on quantifying the performance of these neural networks, we treat the problem from a Bayesian inference perspective and employ nested sampling techniques to perform model comparison. Beyond a certain network size, we find that networks utilising Hironaka decompositions perform the best.

Ensembling with Deep Generative Views ; Recent generative models can synthesize views of artificial images that mimic realworld variations, such as changes in color or pose, simply by learning from unlabeled image collections. Here, we investigate whether such views can be applied to real images to benefit downstream analysis tasks such as image classification. Using a pretrained generator, we first find the latent code corresponding to a given real input image. Applying perturbations to the code creates natural variations of the image, which can then be ensembled together at testtime. We use StyleGAN2 as the source of generative augmentations and investigate this setup on classification tasks involving facial attributes, cat faces, and cars. Critically, we find that several design decisions are required towards making this process work; the perturbation procedure, weighting between the augmentations and original image, and training the classifier on synthesized images can all impact the result. Currently, we find that while testtime ensembling with GANbased augmentations can offer some small improvements, the remaining bottlenecks are the efficiency and accuracy of the GAN reconstructions, coupled with classifier sensitivities to artifacts in GANgenerated images.

Sustainability of Collusion and Market Transparency in a Sequential Search Market a Generalization ; The present work generalizes the analytical results of Petrikaite 2016 to a market where more than two firms interact. As a consequence, for a generic number of firms in the oligopoly model described by Janssen et al 2005, the relationship between the critical discount factor which sustains the monopoly collusive allocation and the share of perfectly informed buyers is nonmonotonic, reaching a unique internal point of minimum. The first section locates the work within the proper economic framework. The second section hosts the analytical computations and the mathematical reasoning needed to derive the desired generalization, which mainly relies on the Leibniz rule for the differentiation under the integral sign and the Bounded Convergence Theorem.

Oneshot Compositional Data Generation for Low Resource Handwritten Text Recognition ; Low resource Handwritten Text Recognition HTR is a hard problem due to the scarce annotated data and the very limited linguistic information dictionaries and language models. For example, in the case of historical ciphered manuscripts, which are usually written with invented alphabets to hide the message contents. Thus, in this paper we address this problem through a data generation technique based on Bayesian Program Learning BPL. Contrary to traditional generation approaches, which require a huge amount of annotated images, our method is able to generate humanlike handwriting using only one sample of each symbol in the alphabet. After generating symbols, we create synthetic lines to train stateoftheart HTR architectures in a segmentation free fashion. Quantitative and qualitative analyses were carried out and confirm the effectiveness of the proposed method.

Calibrating random number generator tests ; Currently, statistical tests for random number generators RNGs are widely used in practice, and some of them are even included in information security standards. But despite the popularity of RNGs, consistent tests are known only for stationary ergodic deviations of randomness a test is consistent if it detects any deviations from a given class when the sample size goes to infty . However, the model of a stationary ergodic source is too narrow for some RNGs, in particular, for generators based on physical effects. In this article, we propose computable consistent tests for some classes of deviations more general than stationary ergodic and describe some general properties of statistical tests. The proposed approach and the resulting test are based on the ideas and methods of information theory.

A Fourierbased Framework for Domain Generalization ; Modern deep neural networks suffer from performance degradation when evaluated on testing data under different distributions from training data. Domain generalization aims at tackling this problem by learning transferable knowledge from multiple source domains in order to generalize to unseen target domains. This paper introduces a novel Fourierbased perspective for domain generalization. The main assumption is that the Fourier phase information contains highlevel semantics and is not easily affected by domain shifts. To force the model to capture phase information, we develop a novel Fourierbased data augmentation strategy called amplitude mix which linearly interpolates between the amplitude spectrums of two images. A dualformed consistency loss called coteacher regularization is further introduced between the predictions induced from original and augmented images. Extensive experiments on three benchmarks have demonstrated that the proposed method is able to achieve stateofthearts performance for domain generalization.

Deep Descriptive Clustering ; Recent work on explainable clustering allows describing clusters when the features are interpretable. However, much modern machine learning focuses on complex data such as images, text, and graphs where deep learning is used but the raw features of data are not interpretable. This paper explores a novel setting for performing clustering on complex data while simultaneously generating explanations using interpretable tags. We propose deep descriptive clustering that performs subsymbolic representation learning on complex data while generating explanations based on symbolic data. We form good clusters by maximizing the mutual information between empirical distribution on the inputs and the induced clustering labels for clustering objectives. We generate explanations by solving an integer linear programming that generates concise and orthogonal descriptions for each cluster. Finally, we allow the explanation to inform better clustering by proposing a novel pairwise loss with selfgenerated constraints to maximize the clustering and explanation module's consistency. Experimental results on public data demonstrate that our model outperforms competitive baselines in clustering performance while offering highquality clusterlevel explanations.

Focus Attention Promoting Faithfulness and Diversity in Summarization ; Professional summaries are written with documentlevel information, such as the theme of the document, in mind. This is in contrast with most seq2seq decoders which simultaneously learn to focus on salient content, while deciding what to generate, at each decoding step. With the motivation to narrow this gap, we introduce Focus Attention Mechanism, a simple yet effective method to encourage decoders to proactively generate tokens that are similar or topical to the input document. Further, we propose a Focus Sampling method to enable generation of diverse summaries, an area currently understudied in summarization. When evaluated on the BBC extreme summarization task, two stateoftheart models augmented with Focus Attention generate summaries that are closer to the target and more faithful to their input documents, outperforming their vanilla counterparts on rouge and multiple faithfulness measures. We also empirically demonstrate that Focus Sampling is more effective in generating diverse and faithful summaries than topk or nucleus samplingbased decoding methods.

Data Augmentation for Text Generation Without Any Augmented Data ; Data augmentation is an effective way to improve the performance of many neural text generation models. However, current data augmentation methods need to define or choose proper data mapping functions that map the original samples into the augmented samples. In this work, we derive an objective to formulate the problem of data augmentation on text generation tasks without any use of augmented data constructed by specific mapping functions. Our proposed objective can be efficiently optimized and applied to popular loss functions on text generation tasks with a convergence rate guarantee. Experiments on five datasets of two text generation tasks show that our approach can approximate or even surpass popular data augmentation methods.

LyricJam A system for generating lyrics for live instrumental music ; We describe a realtime system that receives a live audio stream from a jam session and generates lyric lines that are congruent with the live music being played. Two novel approaches are proposed to align the learned latent spaces of audio and text representations that allow the system to generate novel lyric lines matching live instrumental music. One approach is based on adversarial alignment of latent representations of audio and lyrics, while the other approach learns to transfer the topology from the music latent space to the lyric latent space. A user study with music artists using the system showed that the system was useful not only in lyric composition, but also encouraged the artists to improvise and find new musical expressions. Another user study demonstrated that users preferred the lines generated using the proposed methods to the lines generated by a baseline model.

ImGAGNImbalanced Network Embedding via Generative Adversarial Graph Networks ; Imbalanced classification on graphs is ubiquitous yet challenging in many realworld applications, such as fraudulent node detection. Recently, graph neural networks GNNs have shown promising performance on many network analysis tasks. However, most existing GNNs have almost exclusively focused on the balanced networks, and would get unappealing performance on the imbalanced networks. To bridge this gap, in this paper, we present a generative adversarial graph network model, called ImGAGN to address the imbalanced classification problem on graphs. It introduces a novel generator for graph structure data, named GraphGenerator, which can simulate both the minority class nodes' attribute distribution and network topological structure distribution by generating a set of synthetic minority nodes such that the number of nodes in different classes can be balanced. Then a graph convolutional network GCN discriminator is trained to discriminate between real nodes and fake i.e., generated nodes, and also between minority nodes and majority nodes on the synthetic balanced network. To validate the effectiveness of the proposed method, extensive experiments are conducted on four realworld imbalanced network datasets. Experimental results demonstrate that the proposed method ImGAGN outperforms stateoftheart algorithms for semisupervised imbalanced node classification task.

Generative Adversarial Networks A Survey Towards Private and Secure Applications ; Generative Adversarial Networks GAN have promoted a variety of applications in computer vision, natural language processing, etc. due to its generative model's compelling ability to generate realistic examples plausibly drawn from an existing distribution of samples. GAN not only provides impressive performance on data generationbased tasks but also stimulates fertilization for privacy and security oriented research because of its game theoretic optimization strategy. Unfortunately, there are no comprehensive surveys on GAN in privacy and security, which motivates this survey paper to summarize those stateoftheart works systematically. The existing works are classified into proper categories based on privacy and security functions, and this survey paper conducts a comprehensive analysis of their advantages and drawbacks. Considering that GAN in privacy and security is still at a very initial stage and has imposed unique challenges that are yet to be well addressed, this paper also sheds light on some potential privacy and security applications with GAN and elaborates on some future research directions.

A SelfSupervised Framework for Function Learning and Extrapolation ; Understanding how agents learn to generalize and, in particular, to extrapolate in highdimensional, naturalistic environments remains a challenge for both machine learning and the study of biological agents. One approach to this has been the use of function learning paradigms, which allow peoples' empirical patterns of generalization for smooth scalar functions to be described precisely. However, to date, such work has not succeeded in identifying mechanisms that acquire the kinds of general purpose representations over which function learning can operate to exhibit the patterns of generalization observed in human empirical studies. Here, we present a framework for how a learner may acquire such representations, that then support generalization and extrapolation in particular in a fewshot fashion. Taking inspiration from a classic theory of visual processing, we construct a selfsupervised encoder that implements the basic inductive bias of invariance under topological distortions. We show the resulting representations outperform those from other models for unsupervised time series learning in several downstream function learning tasks, including extrapolation.

Generating Thermal Human Faces for Physiological Assessment Using Thermal Sensor Auxiliary Labels ; Thermal images reveal medically important physiological information about human stress, signs of inflammation, and emotional mood that cannot be seen on visible images. Providing a method to generate thermal faces from visible images would be highly valuable for the telemedicine community in order to show this medical information. To the best of our knowledge, there are limited works on visibletothermal VT face translation, and many current works go the opposite direction to generate visible faces from thermal surveillance images TV for law enforcement applications. As a result, we introduce favtGAN, a VT GAN which uses the pix2pix image translation model with an auxiliary sensor label prediction network for generating thermal faces from visible images. Since most TV methods are trained on only one data source drawn from one thermal sensor, we combine datasets from faces and cityscapes. These combined data are captured from similar sensors in order to bootstrap the training and transfer learning task, especially valuable because visiblethermal face datasets are limited. Experiments on these combined datasets show that favtGAN demonstrates an increase in SSIM and PSNR scores of generated thermal faces, compared to training on a single face dataset alone.

Learning to Disentangle GAN Fingerprint for Fake Image Attribution ; Rapid pace of generative models has brought about new threats to visual forensics such as malicious personation and digital copyright infringement, which promotes works on fake image attribution. Existing works on fake image attribution mainly rely on a direct classification framework. Without additional supervision, the extracted features could include many contentrelevant components and generalize poorly. Meanwhile, how to obtain an interpretable GAN fingerprint to explain the decision remains an open question. Adopting a multitask framework, we propose a GAN Fingerprint Disentangling Network GFDNet to simultaneously disentangle the fingerprint from GANgenerated images and produce a contentirrelevant representation for fake image attribution. A series of constraints are provided to guarantee the stability and discriminability of the fingerprint, which in turn helps contentirrelevant feature extraction. Further, we perform comprehensive analysis on GAN fingerprint, providing some clues about the properties of GAN fingerprint and which factors dominate the fingerprint in GAN architecture. Experiments show that our GFDNet achieves superior fake image attribution performance in both closedworld and openworld testing. We also apply our method in binary fake image detection and exhibit a significant generalization ability on unseen generators.

Random walks and Laplacians on hypergraphs When do they match ; We develop a general theory of random walks on hypergraphs which includes, as special cases, the different models that are found in literature. In particular, we introduce and analyze general random walk Laplacians for hypergraphs, and we compare them to hypergraph normalized Laplacians that are not necessarily related to random walks, but which are motivated by biological and chemical networks. We show that, although these two classes of Laplacians coincide in the case of graphs, they appear to have important conceptual differences in the general case. We study the spectral properties of both classes, as well as their applications to Coupled Hypergraph Maps discretetime dynamical systems that generalize the wellknown Coupled Map Lattices on graphs. Our results also show why for some hypergraph Laplacian variants one expects more classical results from weighted graphs to generalize directly, while these results must fail for other hypergraph Laplacians.

Multiple Organ Failure Prediction with ClassifierGuided Generative Adversarial Imputation Networks ; Multiple organ failure MOF is a severe syndrome with a high mortality rate among Intensive Care Unit ICU patients. Early and precise detection is critical for clinicians to make timely decisions. An essential challenge in applying machine learning models to electronic health records EHRs is the pervasiveness of missing values. Most existing imputation methods are involved in the data preprocessing phase, failing to capture the relationship between data and outcome for downstream predictions. In this paper, we propose classifierguided generative adversarial imputation networks ClassifierGAIN for MOF prediction to bridge this gap, by incorporating both observed data and label information. Specifically, the classifier takes imputed values from the generatorimputer to predict task outcomes and provides additional supervision signals to the generator by joint training. The classifierguide generator imputes missing values with labelawareness during training, improving the classifier's performance during inference. We conduct extensive experiments showing that our approach consistently outperforms classical and stateofart neural baselines across a range of missing data scenarios and evaluation metrics.

AliasFree Generative Adversarial Networks ; We observe that despite their hierarchical convolutional nature, the synthesis process of typical generative adversarial networks depends on absolute pixel coordinates in an unhealthy manner. This manifests itself as, e.g., detail appearing to be glued to image coordinates instead of the surfaces of depicted objects. We trace the root cause to careless signal processing that causes aliasing in the generator network. Interpreting all signals in the network as continuous, we derive generally applicable, small architectural changes that guarantee that unwanted information cannot leak into the hierarchical synthesis process. The resulting networks match the FID of StyleGAN2 but differ dramatically in their internal representations, and they are fully equivariant to translation and rotation even at subpixel scales. Our results pave the way for generative models better suited for video and animation.

Dungeon and Platformer Level Blending and Generation using Conditional VAEs ; Variational autoencoders VAEs have been used in prior works for generating and blending levels from different games. To add controllability to these models, conditional VAEs CVAEs were recently shown capable of generating output that can be modified using labels specifying desired content, albeit working with segments of levels and platformers exclusively. We expand these works by using CVAEs for generating whole platformer and dungeon levels, and blending levels across these genres. We show that CVAEs can reliably control door placement in dungeons and progression direction in platformer levels. Thus, by using appropriate labels, our approach can generate whole dungeons and platformer levels of interconnected rooms and segments respectively as well as levels that blend dungeons and platformers. We demonstrate our approach using The Legend of Zelda, Metroid, Mega Man and Lode Runner.

Towards a full general relativistic approach to galaxies ; We analyze the dynamics of a single spiral galaxy from a general relativistic viewpoint. We employ the known family of stationary axiallysymmetric solutions to Einstein gravity coupled with dust in order to model the halo external to the bulge. In particular, we generalize the known results of Balasin and Grumiller, relaxing the condition of corotation, thus including non corotating dust. This further highlights the discrepancy between Newtonian theory of gravity and general relativity at low velocities and energy densities. We investigate the role of dragging in simulating dark matter effects. In particular, we show that non corotance further reduce the amount of energy density required to explain the rotation curves for spiral galaxies.

Learning to Sample Replacements for ELECTRA PreTraining ; ELECTRA pretrains a discriminator to detect replaced tokens, where the replacements are sampled from a generator trained with masked language modeling. Despite the compelling performance, ELECTRA suffers from the following two issues. First, there is no direct feedback loop from discriminator to generator, which renders replacement sampling inefficient. Second, the generator's prediction tends to be overconfident along with training, making replacements biased to correct tokens. In this paper, we propose two methods to improve replacement sampling for ELECTRA pretraining. Specifically, we augment sampling with a hardness prediction mechanism, so that the generator can encourage the discriminator to learn what it has not acquired. We also prove that efficient sampling reduces the training variance of the discriminator. Moreover, we propose to use a focal loss for the generator in order to relieve oversampling of correct tokens as replacements. Experimental results show that our method improves ELECTRA pretraining on various downstream tasks.

An evaluation of template and MLbased generation of userreadable text from a knowledge graph ; Typical userfriendly renderings of knowledge graphs are visualisations and natural language text. Within the latter HCI solution approach, datadriven natural language generation systems receive increased attention, but they are often outperformed by templatebased systems due to suffering from errors such as content dropping, hallucination, or repetition. It is unknown which of those errors are associated significantly with low quality judgements by humans who the text is aimed for, which hampers addressing errors based on their impact on improving human evaluations. We assessed their possible association with an experiment availing of expert and crowdsourced evaluations of human authored text, template generated text, and sequencetosequence model generated text. The results showed that there was no significant association between human authored texts with errors and the low human judgements of naturalness and quality. There was also no significant association between machine learning generated texts with dropped or hallucinated slots and the low human judgements of naturalness and quality. Thus, both approaches appear to be viable options for designing a natural language interface for knowledge graphs.

Progressive OpenDomain Response Generation with Multiple Controllable Attributes ; It is desirable to include more controllable attributes to enhance the diversity of generated responses in opendomain dialogue systems. However, existing methods can generate responses with only one controllable attribute or lack a flexible way to generate them with multiple controllable attributes. In this paper, we propose a Progressively trained Hierarchical EncoderDecoder PHED to tackle this task. More specifically, PHED deploys Conditional Variational AutoEncoder CVAE on Transformer to include one aspect of attributes at one stage. A vital characteristic of the CVAE is to separate the latent variables at each stage into two types a global variable capturing the common semantic features and a specific variable absorbing the attribute information at that stage. PHED then couples the CVAE latent variables with the Transformer encoder and is trained by minimizing a newly derived ELBO and controlled losses to produce the next stage's input and produce responses as required. Finally, we conduct extensive evaluations to show that PHED significantly outperforms the stateoftheart neural generation models and produces more diverse responses as expected.

Boggart Towards GeneralPurpose Acceleration of Retrospective Video Analytics ; Commercial retrospective video analytics platforms have increasingly adopted general interfaces to support the custom queries and convolutional neural networks CNNs that different applications require. However, existing optimizations were designed for settings where CNNs were platform not user determined, and fail to meet at least one of the following key platform goals when that condition is violated reliable accuracy, low latency, and minimal wasted work. We present Boggart, a system that simultaneously meets all three goals while supporting the generality that today's platforms seek. Prior to queries being issued, Boggart carefully employs traditional computer vision algorithms to generate indices that are imprecise, but are fundamentally comprehensive across different CNNsqueries. For each issued query, Boggart employs new techniques to quickly characterize the imprecision of its index, and sparingly run CNNs and propagate the results to other frames in a way that bounds accuracy drops. Our results highlight that Boggart's improved generality comes at low cost, with speedups that match and most often, exceed prior, modelspecific approaches.

HelpViz Automatic Generation of Contextual Visual MobileTutorials from TextBased Instructions ; We present HelpViz, a tool for generating contextual visual mobile tutorials from textbased instructions that are abundant on the web. HelpViz transforms text instructions to graphical tutorials in batch, by extracting a sequence of actions from each text instruction through an instruction parsing model, and executing the extracted actions on a simulation infrastructure that manages an array of Android emulators. The automatic execution of each instruction produces a set of graphical and structural assets, including images, videos, and metadata such as clicked elements for each step. HelpViz then synthesizes a tutorial by combining parsed text instructions with the generated assets, and contextualizes the tutorial to user interaction by tracking the user's progress and highlighting the next step. Our experiments with HelpViz indicate that our pipeline improved tutorial execution robustness and that participants preferred tutorials generated by HelpViz over textbased instructions. HelpViz promises a costeffective approach for generating contextual visual tutorials for mobile interaction at scale.

Relativistic Generalized Uncertainty Principle and Its Implications ; The fundamental physical description of the Universe is based on two theories Quantum Mechanics and General Relativity. Unified theory of Quantum Gravity QG is an open problem. Quantum Gravity Phenomenology QGP studies QG effects in lowenergy systems. The basis of one such phenomenological model is the Generalized Uncertainty Principle GUP, which is a modified Heisenberg uncertainty relation and predicts a deformed positionmomentum commutator. Relativistic Generalized Uncertainty Principle RGUP is proposed in this thesis, which gives a Loerentz invariant minimum length and resolves the composition law problem. RGUP modified KleinGordon, Schrodinger and Dirac equations with QG corrections to several systems are presented. The Lagrangians of Quantum Electrodynamics for the gauge, scalar, and spinor fields are obtained. The RGUP corrections to scattering amplitudes are then calculated. The results are applied to high energy scattering experiments providing much needed window for testing minimum length and QG theories in the laboratory.

MISS GAN A MultiIlluStrator Style Generative Adversarial Network for image to illustration translation ; Unsupervised style transfer that supports diverse input styles using only one trained generator is a challenging and interesting task in computer vision. This paper proposes a MultiIlluStrator Style Generative Adversarial Network MISS GAN that is a multistyle framework for unsupervised imagetoillustration translation, which can generate styled yet content preserving images. The illustrations dataset is a challenging one since it is comprised of illustrations of seven different illustrators, hence contains diverse styles. Existing methods require to train several generators as the number of illustrators to handle the different illustrators' styles, which limits their practical usage, or require to train an image specific network, which ignores the style information provided in other images of the illustrator. MISS GAN is both input image specific and uses the information of other images using only one trained model.

MVTON Memorybased Video Virtual Tryon network ; With the development of Generative Adversarial Network, imagebased virtual tryon methods have made great progress. However, limited work has explored the task of videobased virtual tryon while it is important in realworld applications. Most existing videobased virtual tryon methods usually require clothing templates and they can only generate blurred and lowresolution results. To address these challenges, we propose a Memorybased Video virtual TryOn Network MVTON, which seamlessly transfers desired clothes to a target person without using any clothing templates and generates highresolution realistic videos. Specifically, MVTON consists of two modules 1 a tryon module that transfers the desired clothes from model images to frame images by pose alignment and regionwise replacing of pixels; 2 a memory refinement module that learns to embed the existing generated frames into the latent space as external memory for the following frame generation. Experimental results show the effectiveness of our method in the video virtual tryon task and its superiority over other existing methods.

The physics of nonideal general relativistic magnetohydrodynamics ; We consider a framework for nonideal magnetohydrodynamics in general relativity, paying particular attention to the physics involved. The discussion highlights the connection between the microphysics associated with a given equation of state and the global dynamics from the point of view of numerical simulations, and includes a careful consideration of the assumptions that lead to ideal and resistive magnetohydrodynamics. We pay particular attention to the issue of local charge neutrality, which tends to be assumed but appears to be more involved than is generally appreciated. While we do not resolve all the involved issues, we highlight how some of the assumptions and simplifications may be tested by simulations. The final formulation is consistent, both logically and physically, preparing the ground for a new generation of models of relevant astrophysical scenarios.

Graphto3D EndtoEnd Generation and Manipulation of 3D Scenes Using Scene Graphs ; Controllable scene synthesis consists of generating 3D information that satisfy underlying specifications. Thereby, these specifications should be abstract, i.e. allowing easy user interaction, whilst providing enough interface for detailed control. Scene graphs are representations of a scene, composed of objects nodes and interobject relationships edges, proven to be particularly suited for this task, as they allow for semantic control on the generated content. Previous works tackling this task often rely on synthetic data, and retrieve object meshes, which naturally limits the generation capabilities. To circumvent this issue, we instead propose the first work that directly generates shapes from a scene graph in an endtoend manner. In addition, we show that the same model supports scene modification, using the respective scene graph as interface. Leveraging Graph Convolutional Networks GCN we train a variational AutoEncoder on top of the object and edge categories, as well as 3D shapes and scene layouts, allowing latter sampling of new scenes and shapes.

ARAPReg An AsRigidAs Possible Regularization Loss for Learning Deformable Shape Generators ; This paper introduces an unsupervised loss for training parametric deformation shape generators. The key idea is to enforce the preservation of local rigidity among the generated shapes. Our approach builds on an approximation of the asrigidas possible or ARAP deformation energy. We show how to develop the unsupervised loss via a spectral decomposition of the Hessian of the ARAP energy. Our loss nicely decouples pose and shape variations through a robust norm. The loss admits simple closedform expressions. It is easy to train and can be plugged into any standard generation models, e.g., variational autoencoder VAE and autodecoder AD. Experimental results show that our approach outperforms existing shape generation approaches considerably on public benchmark datasets of various shape categories such as human, animal and bone.

Image Inpainting via Conditional Texture and Structure Dual Generation ; Deep generative approaches have recently made considerable progress in image inpainting by introducing structure priors. Due to the lack of proper interaction with image texture during structure reconstruction, however, current solutions are incompetent in handling the cases with large corruptions, and they generally suffer from distorted results. In this paper, we propose a novel twostream network for image inpainting, which models the structureconstrained texture synthesis and textureguided structure reconstruction in a coupled manner so that they better leverage each other for more plausible generation. Furthermore, to enhance the global consistency, a Bidirectional Gated Feature Fusion BiGFF module is designed to exchange and combine the structure and texture information and a Contextual Feature Aggregation CFA module is developed to refine the generated contents by region affinity learning and multiscale feature aggregation. Qualitative and quantitative experiments on the CelebA, Paris StreetView and Places2 datasets demonstrate the superiority of the proposed method. Our code is available at httpsgithub.comXiefanGuoCTSDG.

Heredityaware Child Face Image Generation with Latent Space Disentanglement ; Generative adversarial networks have been widely used in image synthesis in recent years and the quality of the generated image has been greatly improved. However, the flexibility to control and decouple facial attributes e.g., eyes, nose, mouth is still limited. In this paper, we propose a novel approach, called ChildGAN, to generate a child's image according to the images of parents with heredity prior. The main idea is to disentangle the latent space of a pretrained generation model and precisely control the face attributes of child images with clear semantics. We use distances between face landmarks as pseudo labels to figure out the most influential semantic vectors of the corresponding face attributes by calculating the gradient of latent vectors to pseudo labels. Furthermore, we disentangle the semantic vectors by weighting irrelevant features and orthogonalizing them with Schmidt Orthogonalization. Finally, we fuse the latent vector of the parents by leveraging the disentangled semantic vectors under the guidance of biological genetic laws. Extensive experiments demonstrate that our approach outperforms the existing methods with encouraging results.

Multivariate Generalized Hermite Subdivision Schemes ; Due to properties such as interpolation, smoothness, and spline connections, Hermite subdivision schemes employ fast iterative algorithms for geometrically modeling curvessurfaces in CAGD and for building Hermite wavelets in numerical PDEs. In this paper we introduce a notion of generalized Hermite dyadic subdivision schemes and then we characterize their convergence, smoothness and underlying matrix masks with or without interpolation properties. We also introduce the notion of linearphase moments for achieving the polynomialinterpolation property. For any given positive integer m, we constructively prove that there always exist convergent smooth generalized Hermite subdivision schemes with linearphase moments such that their basis vector functions are spline functions in Cm and have linearly independent integer shifts. As byproducts, our results resolve convergence, smoothness and existence of Lagrange, Hermite, or Birkhoff subdivision schemes. Even in dimension one our results significantly generalize and extend many known results on extensively studied univariate Hermite subdivision schemes. To illustrate the theoretical results in this paper, we provide examples of convergent generalized Hermite subdivision schemes with symmetric matrix masks having short support and smooth basis vector functions with or without interpolation property.

Formal Query Building with Query Structure Prediction for Complex Question Answering over Knowledge Base ; Formal query building is an important part of complex question answering over knowledge bases. It aims to build correct executable queries for questions. Recent methods try to rank candidate queries generated by a statetransition strategy. However, this candidate generation strategy ignores the structure of queries, resulting in a considerable number of noisy queries. In this paper, we propose a new formal query building approach that consists of two stages. In the first stage, we predict the query structure of the question and leverage the structure to constrain the generation of the candidate queries. We propose a novel graph generation framework to handle the structure prediction task and design an encoderdecoder model to predict the argument of the predetermined operation in each generative step. In the second stage, we follow the previous methods to rank the candidate queries. The experimental results show that our formal query building approach outperforms existing methods on complex questions while staying competitive on simple questions.

Classconditioned Domain Generalization via Wasserstein Distributional Robust Optimization ; Given multiple source domains, domain generalization aims at learning a universal model that performs well on any unseen but related target domain. In this work, we focus on the domain generalization scenario where domain shifts occur among classconditional distributions of different domains. Existing approaches are not sufficiently robust when the variation of conditional distributions given the same class is large. In this work, we extend the concept of distributional robust optimization to solve the classconditional domain generalization problem. Our approach optimizes the worstcase performance of a classifier over classconditional distributions within a Wasserstein ball centered around the barycenter of the source conditional distributions. We also propose an iterative algorithm for learning the optimal radius of the Wasserstein balls automatically. Experiments show that the proposed framework has better performance on unseen target domain than approaches without domain generalization.

HypoGen Hyperbole Generation with Commonsense and Counterfactual Knowledge ; A hyperbole is an intentional and creative exaggeration not to be taken literally. Despite its ubiquity in daily life, the computational explorations of hyperboles are scarce. In this paper, we tackle the underexplored and challenging task sentencelevel hyperbole generation. We start with a representative syntactic pattern for intensification and systematically study the semantic commonsense and counterfactual relationships between each component in such hyperboles. Next, we leverage the COMeT and reverse COMeT models to do commonsense and counterfactual inference. We then generate multiple hyperbole candidates based on our findings from the pattern, and train neural classifiers to rank and select highquality hyperboles. Automatic and human evaluations show that our generation method is able to generate hyperboles creatively with high success rate and intensity scores.