title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Wasserstein $F$-tests and Confidence Bands for the Fr\`echet Regression
of Density Response Curves | Data consisting of samples of probability density functions are increasingly
prevalent, necessitating the development of methodologies for their analysis
that respect the inherent nonlinearities associated with densities. In many
applications, density curves appear as functional response objects in a
regression model with vector predictors. For such models, inference is key to
understand the importance of density-predictor relationships, and the
uncertainty associated with the estimated conditional mean densities, defined
as conditional Fr\'echet means under a suitable metric. Using the Wasserstein
geometry of optimal transport, we consider the Fr\'echet regression of density
curve responses and develop tests for global and partial effects, as well as
simultaneous confidence bands for estimated conditional mean densities. The
asymptotic behavior of these objects is based on underlying functional central
limit theorems within Wasserstein space, and we demonstrate that they are
asymptotically of the correct size and coverage, with uniformly strong
consistency of the proposed tests under sequences of contiguous alternatives.
The accuracy of these methods, including nominal size, power, and coverage, is
assessed through simulations, and their utility is illustrated through a
regression analysis of post-intracerebral hemorrhage hematoma densities and
their associations with a set of clinical and radiological covariates.
| stat |
Equivalence tests for binary efficacy-toxicity responses | Clinical trials often aim to compare a new drug with a reference treatment in
terms of efficacy and/or toxicity depending on covariates such as, for example,
the dose level of the drug. Equivalence of these treatments can be claimed if
the difference in average outcome is below a certain threshold over the
covariate range. In this paper we assume that the efficacy and toxicity of the
treatments are measured as binary outcome variables and we address two
problems. First, we develop a new test procedure for the assessment of
equivalence of two treatments over the entire covariate range for a single
binary endpoint. Our approach is based on a parametric bootstrap, which
generates data under the constraint that the distance between the curves is
equal to the pre-specified equivalence threshold. Second, we address
equivalence for bivariate binary (correlated) outcomes by extending the
previous approach for a univariate response. For this purpose we use a
2-dimensional Gumbel model for binary efficacy-toxicity responses. We
investigate the operating characteristics of the proposed approaches by means
of a simulation study and present a case study as an illustration.
| stat |
Implicit differentiation for fast hyperparameter selection in non-smooth
convex learning | Finding the optimal hyperparameters of a model can be cast as a bilevel
optimization problem, typically solved using zero-order techniques. In this
work we study first-order methods when the inner optimization problem is convex
but non-smooth. We show that the forward-mode differentiation of proximal
gradient descent and proximal coordinate descent yield sequences of Jacobians
converging toward the exact Jacobian. Using implicit differentiation, we show
it is possible to leverage the non-smoothness of the inner problem to speed up
the computation. Finally, we provide a bound on the error made on the
hypergradient when the inner optimization problem is solved approximately.
Results on regression and classification problems reveal computational benefits
for hyperparameter optimization, especially when multiple hyperparameters are
required.
| stat |
On Heavy-user Bias in A/B Testing | On-line experimentation (also known as A/B testing) has become an integral
part of software development. To timely incorporate user feedback and
continuously improve products, many software companies have adopted the culture
of agile deployment, requiring online experiments to be conducted and concluded
on limited sets of users for a short period. While conceptually efficient, the
result observed during the experiment duration can deviate from what is seen
after the feature deployment, which makes the A/B test result biased. In this
paper, we provide theoretical analysis to show that heavy-users can contribute
significantly to the bias, and propose a re-sampling estimator for bias
adjustment.
| stat |
Low-pass filtering as Bayesian inference | We propose a Bayesian nonparametric method for low-pass filtering that can
naturally handle unevenly-sampled and noise-corrupted observations. The
proposed model is constructed as a latent-factor model for time series, where
the latent factors are Gaussian processes with non-overlapping spectra. With
this construction, the low-pass version of the time series can be identified as
the low-frequency latent component, and therefore it can be found by means of
Bayesian inference. We show that the model admits exact training and can be
implemented with minimal numerical approximations. Finally, the proposed model
is validated against standard linear filters on synthetic and real-world time
series.
| stat |
Calibrating Multivariate L\'evy Processes with Neural Networks | Calibrating a L\'evy process usually requires characterizing its jump
distribution. Traditionally this problem can be solved with nonparametric
estimation using the empirical characteristic functions (ECF), assuming certain
regularity, and results to date are mostly in 1D. For multivariate L\'evy
processes and less smooth L\'evy densities, the problem becomes challenging as
ECFs decay slowly and have large uncertainty because of limited observations.
We solve this problem by approximating the L\'evy density with a parametrized
functional form; the characteristic function is then estimated using numerical
integration. In our benchmarks, we used deep neural networks and found that
they are robust and can capture sharp transitions in the L\'evy density. They
perform favorably compared to piecewise linear functions and radial basis
functions. The methods and techniques developed here apply to many other
problems that involve nonparametric estimation of functions embedded in a
system model.
| stat |
Modelling spine locations on dendrite trees using inhomogeneous Cox
point processes | Dendritic spines, which are small protrusions on the dendrites of a neuron,
are of interest in neuroscience as they are related to cognitive processes such
as learning and memory. We analyse the distribution of spine locations on six
different dendrite trees from mouse neurons using point process theory for
linear networks. Besides some possible small-scale repulsion, { we find that
two of the spine point pattern data sets may be described by inhomogeneous
Poisson process models}, while the other point pattern data sets exhibit
clustering between spines at a larger scale. To model this we propose an
inhomogeneous Cox process model constructed by thinning a Poisson process on a
linear network with retention probabilities determined by a spatially
correlated random field. For model checking we consider network analogues of
the empirical $F$-, $G$-, and $J$-functions originally introduced for
inhomogeneous point processes on a Euclidean space. The fitted Cox process
models seem to catch the clustering of spine locations between spines, but also
posses a large variance in the number of points for some of the data sets
causing large confidence regions for the empirical $F$- and $G$-functions.
| stat |
EnergyStar++: Towards more accurate and explanatory building energy
benchmarking | Building energy performance benchmarking has been adopted widely in the USA
and Canada through the Energy Star Portfolio Manager platform. Building
operations and energy management professionals have long used a simple 1-100
score to understand how their building compares to its peers. This single
number is easy to use, but is created by inaccurate linear regression (MLR)
models. This paper proposes a methodology that enhances the existing Energy
Star calculation method by increasing accuracy and providing additional model
output processing to help explain why a building is achieving a certain score.
We propose and test two new prediction models: multiple linear regression with
feature interactions (MLRi) and gradient boosted trees (GBT). Both models have
better average accuracy than the baseline Energy Star models. The third order
MLRi and GBT models achieve 4.9% and 24.9% increase in adjusted R2,
respectively, and 7.0% and 13.7% decrease in normalized root mean squared error
(NRMSE), respectively, on average than MLR models for six building types. Even
more importantly, a set of techniques is developed to help determine which
factors most influence the score using SHAP values. The SHAP force
visualization in particular offers an accessible overview of the aspects of the
building that influence the score that non-technical users can readily
interpret. This methodology is tested on the 2012 Commercial Building Energy
Consumption Survey (CBECS)(1,812 buildings) and public data sets from the
energy disclosure programs of New York City (11,131 buildings) and Seattle
(2,073 buildings).
| stat |
Non-convex weakly smooth Langevin Monte Carlo using regularization | Discretization of continuous-time diffusion processes is a widely recognized
method for sampling. However, the canonical Euler Maruyama discretization of
the Langevin diffusion process, referred as Langevin Monte Carlo (LMC), studied
mostly in the context of smooth (gradient Lipschitz) and strongly log-concave
densities, is a considerable hindrance for its deployment in many sciences,
including computational statistics and statistical learning. In this paper, we
establish several theoretical contributions to the literature on such sampling
methods for weakly smooth and non-convex densities. Particularly, we use
convexification of nonconvex domain \citep{ma2019sampling} in combination with
regularization to prove convergence in Kullback-Leibler (KL) divergence with
the number of iterations to reach $\epsilon-$ neighborhood of a target
distribution in only polynomial dependence on the dimension. We relax the
conditions of \citep{vempala2019rapid} and prove convergence guarantees under
isoperimetry, degenerated convex, and non strongly convex at infinity.
| stat |
Dream Distillation: A Data-Independent Model Compression Framework | Model compression is eminently suited for deploying deep learning on
IoT-devices. However, existing model compression techniques rely on access to
the original or some alternate dataset. In this paper, we address the model
compression problem when no real data is available, e.g., when data is private.
To this end, we propose Dream Distillation, a data-independent model
compression framework. Our experiments show that Dream Distillation can achieve
88.5% accuracy on the CIFAR-10 test set without actually training on the
original data!
| stat |
Conjoined Dirichlet Process | Biclustering is a class of techniques that simultaneously clusters the rows
and columns of a matrix to sort heterogeneous data into homogeneous blocks.
Although many algorithms have been proposed to find biclusters, existing
methods suffer from the pre-specification of the number of biclusters or place
constraints on the model structure. To address these issues, we develop a
novel, non-parametric probabilistic biclustering method based on Dirichlet
processes to identify biclusters with strong co-occurrence in both rows and
columns. The proposed method utilizes dual Dirichlet process mixture models to
learn row and column clusters, with the number of resulting clusters determined
by the data rather than pre-specified. Probabilistic biclusters are identified
by modeling the mutual dependence between the row and column clusters. We apply
our method to two different applications, text mining and gene expression
analysis, and demonstrate that our method improves bicluster extraction in many
settings compared to existing approaches.
| stat |
Linear-Cost Covariance Functions for Gaussian Random Fields | Gaussian random fields (GRF) are a fundamental stochastic model for
spatiotemporal data analysis. An essential ingredient of GRF is the covariance
function that characterizes the joint Gaussian distribution of the field.
Commonly used covariance functions give rise to fully dense and unstructured
covariance matrices, for which required calculations are notoriously expensive
to carry out for large data. In this work, we propose a construction of
covariance functions that result in matrices with a hierarchical structure.
Empowered by matrix algorithms that scale linearly with the matrix dimension,
the hierarchical structure is proved to be efficient for a variety of random
field computations, including sampling, kriging, and likelihood evaluation.
Specifically, with $n$ scattered sites, sampling and likelihood evaluation has
an $O(n)$ cost and kriging has an $O(\log n)$ cost after preprocessing,
particularly favorable for the kriging of an extremely large number of sites
(e.g., predicting on more sites than observed). We demonstrate comprehensive
numerical experiments to show the use of the constructed covariance functions
and their appealing computation time. Numerical examples on a laptop include
simulated data of size up to one million, as well as a climate data product
with over two million observations.
| stat |
Composite Estimation for Quantile Regression Kink Models with
Longitudinal Data | Kink model is developed to analyze the data where the regression function is
twostage linear but intersects at an unknown threshold. In quantile regression
with longitudinal data, previous work assumed that the unknown threshold
parameters or kink points are heterogeneous across different quantiles.
However, the location where kink effect happens tend to be the same across
different quantiles, especially in a region of neighboring quantile levels.
Ignoring such homogeneity information may lead to efficiency loss for
estimation. In view of this, we propose a composite estimator for the common
kink point by absorbing information from multiple quantiles. In addition, we
also develop a sup-likelihood-ratio test to check the kink effect at a given
quantile level. A test-inversion confidence interval for the common kink point
is also developed based on the quantile rank score test. The simulation study
shows that the proposed composite kink estimator is more competitive with the
least square estimator and the single quantile estimator. We illustrate the
practical value of this work through the analysis of a body mass index and
blood pressure data set.
| stat |
(f)RFCDE: Random Forests for Conditional Density Estimation and
Functional Data | Random forests is a common non-parametric regression technique which performs
well for mixed-type unordered data and irrelevant features, while being robust
to monotonic variable transformations. Standard random forests, however, do not
efficiently handle functional data and runs into a curse-of dimensionality when
presented with high-resolution curves and surfaces. Furthermore, in settings
with heteroskedasticity or multimodality, a regression point estimate with
standard errors do not fully capture the uncertainty in our predictions. A more
informative quantity is the conditional density p(y | x) which describes the
full extent of the uncertainty in the response y given covariates x. In this
paper we show how random forests can be efficiently leveraged for conditional
density estimation, functional covariates, and multiple responses without
increasing computational complexity. We provide open-source software for all
procedures with R and Python versions that call a common C++ library.
| stat |
Measuring the Algorithmic Convergence of Randomized Ensembles: The
Regression Setting | When randomized ensemble methods such as bagging and random forests are
implemented, a basic question arises: Is the ensemble large enough? In
particular, the practitioner desires a rigorous guarantee that a given ensemble
will perform nearly as well as an ideal infinite ensemble (trained on the same
data). The purpose of the current paper is to develop a bootstrap method for
solving this problem in the context of regression --- which complements our
companion paper in the context of classification (Lopes 2019). In contrast to
the classification setting, the current paper shows that theoretical guarantees
for the proposed bootstrap can be established under much weaker assumptions. In
addition, we illustrate the flexibility of the method by showing how it can be
adapted to measure algorithmic convergence for variable selection. Lastly, we
provide numerical results demonstrating that the method works well in a range
of situations.
| stat |
Measuring and assessing economic uncertainty | This paper evaluates the dynamic response of economic activity to shocks in
uncertainty as percieved by agents.The study focuses on the comparison between
the perception of economic uncertainty by manufacturers and consumers.Since
uncertainty is not directly observable, we approximate it using the geometric
discrepancy indicator of Claveria et al.(2019).This approach allows us
quantifying the proportion of disagreement in business and consumer
expectations of eleven European countries and the Euro Area.First, we compute
three independent indices of discrepancy corresponding to three dimensions of
uncertainty (economic, inflation and employment) and we average them to obtain
aggregate disagreement measures for businesses and for consumers.Next, we use a
bivariate Bayesian vector autoregressive framework to estimate the impulse
response functions to innovations in disagreement in every country.We find that
the effect on economic activity of shocks to the perception of uncertainty
differ markedly between manufacturers and consumers.On the one hand, shocks to
consumer discrepancy tend to be of greater magnitude and duration than those to
manufacturer discrepancy.On the other hand, innovations in disagreement between
the two collectives have an opposite effect on economic activity:shocks to
manufacturer discrepancy lead to a decrease in economic activity, as opposed to
shocks to consumer discrepancy.This finding is of particular relevance to
researchers when using cross-sectional dispersion of survey-based expectations,
since the effect on economic growth of shocks to disagreement depend on the
type of agent.
| stat |
Addressing the Mystery of Population Decline of the Rose-Crested Blue
Pipit in a Nature Preserve using Data Visualization | Two main methods for exploring patterns in data are data visualization and
machine learning. The former relies on humans for investigating the patterns
while the latter relies on machine learning algorithms. This paper tries to
find the patterns using only data visualization. It addresses the mystery of
population decline of a bird, named Rose-Crested Blue Pipit, in a hypothetical
nature preserve. Different visualization techniques are used and the reasons of
the problem are found and categorized. Finally, the solutions for preventing
the future similar problems are suggested. This paper can be useful for getting
introduced to some data visualization tools and techniques.
| stat |
Anomaly Detection of Mobility Data with Applications to COVID-19
Situational Awareness | This work introduces a live anomaly detection system for high frequency and
high-dimensional data collected at regional scale such as Origin Destination
Matrices of mobile positioning data. To take into account different granularity
in time and space of the data coming from different sources, the system is
designed to be simple, yet robust to the data diversity, with the aim of
detecting abrupt increase of mobility towards specific regions as well as
sudden drops of movements. The methodology is designed to help policymakers or
practitioners, and makes it possible to visualise anomalies as well as estimate
the effect of COVID-19 related containment or lifting measures in terms of
their impact on human mobility as well as spot potential new outbreaks related
to large gatherings.
| stat |
A Bayesian model of microbiome data for simultaneous identification of
covariate associations and prediction of phenotypic outcomes | One of the major research questions regarding human microbiome studies is the
feasibility of designing interventions that modulate the composition of the
microbiome to promote health and cure disease. This requires extensive
understanding of the modulating factors of the microbiome, such as dietary
intake, as well as the relation between microbial composition and phenotypic
outcomes, such as body mass index (BMI). Previous efforts have modeled these
data separately, employing two-step approaches that can produce biased
interpretations of the results. Here, we propose a Bayesian joint model that
simultaneously identifies clinical covariates associated with microbial
composition data and predicts a phenotypic response using information contained
in the compositional data. Using spike-and-slab priors, our approach can handle
high-dimensional compositional as well as clinical data. Additionally, we
accommodate the compositional structure of the data via balances and
overdispersion typically found in microbial samples. We apply our model to
understand the relations between dietary intake, microbial samples, and BMI. In
this analysis, we find numerous associations between microbial taxa and dietary
factors that may lead to a microbiome that is generally more hospitable to the
development of chronic diseases, such as obesity. Additionally, we demonstrate
on simulated data how our method outperforms two-step approaches and also
present a sensitivity analysis.
| stat |
Nonlinear Regression without i.i.d. Assumption | In this paper, we consider a class of nonlinear regression problems without
the assumption of being independent and identically distributed. We propose a
correspondent mini-max problem for nonlinear regression and give a numerical
algorithm. Such an algorithm can be applied in regression and machine learning
problems, and yield better results than traditional least square and machine
learning methods.
| stat |
Learning Weighted Submanifolds with Variational Autoencoders and
Riemannian Variational Autoencoders | Manifold-valued data naturally arises in medical imaging. In cognitive
neuroscience, for instance, brain connectomes base the analysis of coactivation
patterns between different brain regions on the analysis of the correlations of
their functional Magnetic Resonance Imaging (fMRI) time series - an object thus
constrained by construction to belong to the manifold of symmetric positive
definite matrices. One of the challenges that naturally arises consists of
finding a lower-dimensional subspace for representing such manifold-valued
data. Traditional techniques, like principal component analysis, are
ill-adapted to tackle non-Euclidean spaces and may fail to achieve a
lower-dimensional representation of the data - thus potentially pointing to the
absence of lower-dimensional representation of the data. However, these
techniques are restricted in that: (i) they do not leverage the assumption that
the connectomes belong on a pre-specified manifold, therefore discarding
information; (ii) they can only fit a linear subspace to the data. In this
paper, we are interested in variants to learn potentially highly curved
submanifolds of manifold-valued data. Motivated by the brain connectomes
example, we investigate a latent variable generative model, which has the added
benefit of providing us with uncertainty estimates - a crucial quantity in the
medical applications we are considering. While latent variable models have been
proposed to learn linear and nonlinear spaces for Euclidean data, or geodesic
subspaces for manifold data, no intrinsic latent variable model exists to learn
nongeodesic subspaces for manifold data. This paper fills this gap and
formulates a Riemannian variational autoencoder with an intrinsic generative
model of manifold-valued data. We evaluate its performances on synthetic and
real datasets by introducing the formalism of weighted Riemannian submanifolds.
| stat |
IMMIGRATE: A Margin-based Feature Selection Method with Interaction
Terms | Relief based algorithms have often been claimed to uncover feature
interactions. However, it is still unclear whether and how interaction terms
will be differentiated from marginal effects. In this paper, we propose
IMMIGRATE algorithm by including and training weights for interaction terms.
Besides applying the large margin principle, we focus on the robustness of the
contributors of margin and consider local and global information
simultaneously. Moreover, IMMIGRATE has been shown to enjoy attractive
properties, such as robustness and combination with Boosting. We evaluate our
proposed method on several tasks, which achieves state-of-the-art results
significantly.
| stat |
Navigated Weighting to Improve Inverse Probability Weighting for Missing
Data Problems and Causal Inference | The inverse probability weighting (IPW) is broadly utilized to address
missing data problems including causal inference but may suffer from large
variances and biases due to propensity score model misspecification. To solve
these problems, I propose an estimation method called the navigated weighting
(NAWT), which utilizes estimating equations suitable for a specific
pre-specified parameter of interest (e.g., the average treatment effects on the
treated). Since these pre-specified parameters determine the relative
importance of each unit as a function of propensity scores, the NAWT
prioritizes important units in the propensity score estimation to improve
efficiency and robustness to model misspecification. I investigate its
large-sample properties and demonstrate its finite sample improvements through
simulation studies and an empirical example. An R package nawtilus which
implements the NAWT is developed and available from the Comprehensive R Archive
Network (http://cran.r-project.org/package=nawtilus).
| stat |
Learning interaction kernels in mean-field equations of 1st-order
systems of interacting particles | We introduce a nonparametric algorithm to learn interaction kernels of
mean-field equations for 1st-order systems of interacting particles. The data
consist of discrete space-time observations of the solution. By least squares
with regularization, the algorithm learns the kernel on data-adaptive
hypothesis spaces efficiently. A key ingredient is a probabilistic error
functional derived from the likelihood of the mean-field equation's diffusion
process. The estimator converges, in a reproducing kernel Hilbert space and an
L2 space under an identifiability condition, at a rate optimal in the sense
that it equals the numerical integrator's order. We demonstrate our algorithm
on three typical examples: the opinion dynamics with a piecewise linear kernel,
the granular media model with a quadratic kernel, and the aggregation-diffusion
with a repulsive-attractive kernel.
| stat |
Modeling the Health Expenditure in Japan, 2011. A Healthy Life Years
Lost Methodology | The Healthy Life Years Lost Methodology (HLYL) is introduced to model and
estimate the Health Expenditure in Japan in 2011. The HLYL theory and
estimation methods are presented in our books in the Springer Series on
Demographic Methods and Population Analysis vol. 45 and 46 titled: Exploring
the Health State of a Population by Dynamic Modeling Methods and Demography and
Health Issues: Population Aging, Mortality and Data Analysis. Special
applications appear in Chapters of these books as in The Health-Mortality
Approach in Estimating the Healthy Life Years Lost Compared to the Global
Burden of Disease Studies and Applications in World, USA and Japan and in
Estimation of the Healthy Life Expectancy in Italy Through a Simple Model Based
on Mortality Rate by Skiadas and Arezzo. Here further to present the main part
of the methodology with more details and illustrations, we develop and extend a
life table important to estimate the healthy life years lost along with the
fitting to the health expenditure in the related case. The application results
are quite promising and important to support decision makers and health
agencies with a powerful tool to improve the health expenditure allocation and
the future predictions.
| stat |
Small Area Estimation with Linked Data | In Small Area Estimation data linkage can be used to combine values of the
variableof interest from a national survey with values of auxiliary variables
obtained from another source like a population register. Linkage errors can
induce bias when fitting regression models; moreover, they can create
non-representative outliers in the linked data in addition to the presence of
potential representative outliers. In this paper we adopt a secondary analyst's
point view, assuming limited information is available on the linkage process,
and we develop small area estimators based on linear mixed and linear
M-quantile models to accommodate linked data containing a mix of both types of
outliers. We illustrate the properties of these small area estimators, as well
as estimators of their mean squared error, by means of model-based and
design-based simulation experiments. These experiments show that the proposed
predictors can lead to more efficient estimators when there is linkage error.
Furthermore, the proposed mean-squared error estimation methods appear to
perform well.
| stat |
Quantifying uncertainty in spikes estimated from calcium imaging data | In recent years, a number of methods have been proposed to estimate the times
at which neurons spike on the basis of calcium imaging data. However,
quantifying the uncertainty associated with these estimated spikes remains an
open problem. We consider a simple and well-studied model for calcium imaging
data, which states that calcium decays exponentially in the absence of a spike,
and instantaneously increases when a spike occurs. We wish to test the null
hypothesis that the neuron did not spike -- i.e., that there was no increase in
calcium -- at a particular timepoint at which a spike was estimated. In this
setting, classical hypothesis tests lead to inflated Type I error, because the
spike was estimated on the same data. To address this problem, we propose a
selective inference approach to test the null hypothesis. We describe an
efficient algorithm to compute finite-sample p-values that control selective
Type I error, and confidence intervals with correct selective coverage, for
spikes estimated using a recent proposal from the literature. We apply our
proposal in simulation and on calcium imaging data from the spikefinder
challenge.
| stat |
VAR estimators using binary measurements | In this paper, two novel algorithms to estimate a Gaussian Vector
Autoregressive (VAR) model from 1-bit measurements are introduced. They are
based on the Yule-Walker scheme modified to account for quantisation. The
scalar case has been studied before. The main difficulty when going from the
scalar to the vector case is how to estimate the ratios of the variances of
pairwise components of the VAR model. The first method overcomes this
difficulty by requiring the quantisation to be non-symmetric: each component of
the VAR model output is replaced by a binary "zero" or a binary "one" depending
on whether its value is greater than a strictly positive threshold. Different
components of the VAR model can have different thresholds. As the choice of
these thresholds has a strong influence on the performance, this first method
is best suited for applications where the variance of each time series is
approximately known prior to choosing the corresponding threshold. The second
method relies instead on symmetric quantisations of not only each component of
the VAR model but also on the pairwise differences of the components. These
additional measurements are equivalent to a ranking of the instantaneous VAR
model output, from the smallest component to the largest component. This avoids
the need for choosing thresholds but requires additional hardware for
quantising the components in pairs. Numerical simulations show the efficiency
of both schemes.
| stat |
Robust subgaussian estimation with VC-dimension | Median-of-means (MOM) based procedures provide non-asymptotic and strong
deviation bounds even when data are heavy-tailed and/or corrupted. This work
proposes a new general way to bound the excess risk for MOM estimators. The
core technique is the use of VC-dimension (instead of Rademacher complexity) to
measure the statistical complexity. In particular, this allows to give the
first robust estimators for sparse estimation which achieves the so-called
subgaussian rate only assuming a finite second moment for the uncorrupted data.
By comparison, previous works using Rademacher complexities required a number
of finite moments that grows logarithmically with the dimension. With this
technique, we derive new robust sugaussian bounds for mean estimation in any
norm. We also derive a new robust estimator for covariance estimation that is
the first to achieve subgaussian bounds without $L_4-L_2$ norm equivalence.
| stat |
Adversarially Robust Classification based on GLRT | Machine learning models are vulnerable to adversarial attacks that can often
cause misclassification by introducing small but well designed perturbations.
In this paper, we explore, in the setting of classical composite hypothesis
testing, a defense strategy based on the generalized likelihood ratio test
(GLRT), which jointly estimates the class of interest and the adversarial
perturbation. We evaluate the GLRT approach for the special case of binary
hypothesis testing in white Gaussian noise under $\ell_{\infty}$ norm-bounded
adversarial perturbations, a setting for which a minimax strategy optimizing
for the worst-case attack is known. We show that the GLRT approach yields
performance competitive with that of the minimax approach under the worst-case
attack, and observe that it yields a better robustness-accuracy trade-off under
weaker attacks, depending on the values of signal components relative to the
attack budget. We also observe that the GLRT defense generalizes naturally to
more complex models for which optimal minimax classifiers are not known.
| stat |
Ensuring Reliable Monte Carlo Estimates of Network Properties | The literature in social network analysis has largely focused on methods and
models which require complete network data; however there exist many networks
which can only be studied via sampling methods due to the scale or complexity
of the network, access limitations, or the population of interest is hard to
reach. In such cases, the application of random walk-based Markov chain Monte
Carlo (MCMC) methods to estimate multiple network features is common. However,
the reliability of these estimates has been largely ignored. We consider and
further develop multivariate MCMC output analysis methods in the context of
network sampling to directly address the reliability of the multivariate
estimation. This approach yields principled, computationally efficient, and
broadly applicable methods for assessing the Monte Carlo estimation procedure.
In particular, with respect to two random-walk algorithms, a simple random walk
and a Metropolis-Hastings random walk, we construct and compare network
parameter estimates, effective sample sizes, coverage probabilities, and
stopping rules, all of which speaks to the estimation reliability.
| stat |
ABCDP: Approximate Bayesian Computation with Differential Privacy | We develop a novel approximate Bayesian computation (ABC) framework, ABCDP,
that produces differentially private (DP) and approximate posterior samples.
Our framework takes advantage of the Sparse Vector Technique (SVT), widely
studied in the differential privacy literature. SVT incurs the privacy cost
only when a condition (whether a quantity of interest is above/below a
threshold) is met. If the condition is met sparsely during the repeated
queries, SVT can drastically reduces the cumulative privacy loss, unlike the
usual case where every query incurs the privacy loss. In ABC, the quantity of
interest is the distance between observed and simulated data, and only when the
distance is below a threshold, we take the corresponding prior sample as a
posterior sample. Hence, applying SVT to ABC is an organic way to transform an
ABC algorithm to a privacy-preserving variant with minimal modification, but
yields the posterior samples with a high privacy level. We theoretically
analyze the interplay between the noise added for privacy and the accuracy of
the posterior samples.
| stat |
Data integration for high-resolution, continental-scale estimation of
air pollution concentrations | Air pollution constitutes the highest environmental risk factor in relation
to heath. In order to provide the evidence required for health impact analyses,
to inform policy and to develop potential mitigation strategies comprehensive
information is required on the state of air pollution. Information on air
pollution traditionally comes from ground monitoring (GM) networks but these
may not be able to provide sufficient coverage and may need to be supplemented
with information from other sources (e.g. chemical transport models; CTMs).
However, these may only be available on grids and may not capture micro-scale
features that may be important in assessing air quality in areas of high
population. We develop a model that allows calibration between multiple data
sources available at different levels of support by allowing the coefficients
of calibration equations to vary over space and time, enabling downscaling
where the data is sufficient to support it. The model is used to produce
high-resolution (1km $\times$ 1km) estimates of NO$_2$ and PM$_{2.5}$ across
Western Europe for 2010-2016. Concentrations of both pollutants are decreasing
during this period, however there remain large populations exposed to levels
exceeding the WHO Air Quality Guidelines and thus air pollution remains a
serious threat to health.
| stat |
Debiased distributed learning for sparse partial linear models in high
dimensions | Although various distributed machine learning schemes have been proposed
recently for pure linear models and fully nonparametric models, little
attention has been paid on distributed optimization for semi-paramemetric
models with multiple-level structures (e.g. sparsity, linearity and
nonlinearity). To address these issues, the current paper proposes a new
communication-efficient distributed learning algorithm for partially sparse
linear models with an increasing number of features. The proposed method is
based on the classical divide and conquer strategy for handing big data and
each sub-method defined on each subsample consists of a debiased estimation of
the double-regularized least squares approach. With the proposed method, we
theoretically prove that our global parametric estimator can achieve optimal
parametric rate in our semi-parametric model given an appropriate partition on
the total data. Specially, the choice of data partition relies on the
underlying smoothness of the nonparametric component, but it is adaptive to the
sparsity parameter. Even under the non-distributed setting, we develop a new
and easily-read proof for optimal estimation of the parametric error in high
dimensional partial linear model. Finally, several simulated experiments are
implemented to indicate comparable empirical performance of our debiased
technique under the distributed setting.
| stat |
Incomplete Reparameterizations and Equivalent Metrics | Reparameterizing a probabilisitic system is common advice for improving the
performance of a statistical algorithm like Markov chain Monte Carlo, even
though in theory such reparameterizations should leave the system, and the
performance of any algorithm, invariant. In this paper I show how the
reparameterizations common in practice are only incomplete reparameterizations
which result in different interactions between a target probabilistic system
and a given algorithm. I then consider how these changing interactions manifest
in the context of Markov chain Monte Carlo algorithms defined on Riemannian
manifolds. In particular I show how any incomplete reparameterization is
equivalent to modifying the metric geometry directly.
| stat |
Variational Inference with Numerical Derivatives: variance reduction
through coupling | The Black Box Variational Inference (Ranganath et al. (2014)) algorithm
provides a universal method for Variational Inference, but taking advantage of
special properties of the approximation family or of the target can improve the
convergence speed significantly. For example, if the approximation family is a
transformation family, such as a Gaussian, then switching to the
reparameterization gradient (Kingma and Welling (2014)) often yields a major
reduction in gradient variance. Ultimately, reducing the variance can reduce
the computational cost and yield better approximations.
We present a new method to extend the reparameterization trick to more
general exponential families including the Wishart, Gamma, and Student
distributions. Variational Inference with Numerical Derivatives (VIND)
approximates the gradient with numerical derivatives and reduces its variance
using a tight coupling of the approximation family. The resulting algorithm is
simple to implement and can profit from widely known couplings. Our experiments
confirm that VIND effectively decreases the gradient variance and therefore
improves the posterior approximation in relevant cases. It thus provides an
efficient yet simple Variational Inference method for computing non-Gaussian
approximations.
| stat |
Sparse Normal Means Estimation with Sublinear Communication | We consider the problem of sparse normal means estimation in a distributed
setting with communication constraints. We assume there are $M$ machines, each
holding a $d$-dimensional observation of a $K$-sparse vector $\mu$ corrupted by
additive Gaussian noise. A central fusion machine is connected to the $M$
machines in a star topology, and its goal is to estimate the vector $\mu$ with
a low communication budget. Previous works have shown that to achieve the
centralized minimax rate for the $\ell_2$ risk, the total communication must be
high - at least linear in the dimension $d$. This phenomenon occurs, however,
at very weak signals. We show that once the signal-to-noise ratio (SNR) is
slightly higher, the support of $\mu$ can be correctly recovered with much less
communication. Specifically, we present two algorithms for the distributed
sparse normal means problem, and prove that above a certain SNR threshold, with
high probability, they recover the correct support with total communication
that is sublinear in the dimension $d$. Furthermore, the communication
decreases exponentially as a function of signal strength. If in addition $KM\ll
d$, then with an additional round of sublinear communication, our algorithms
achieve the centralized rate for the $\ell_2$ risk. Finally, we present
simulations that illustrate the performance of our algorithms in different
parameter regimes.
| stat |
The Statistics of Circular Optimal Transport | Empirical optimal transport (OT) plans and distances provide effective tools
to compare and statistically match probability measures defined on a given
ground space. Fundamental to this are distributional limit laws and we derive a
central limit theorem for the empirical OT distance of circular data. Our limit
results require only mild assumptions in general and include prominent examples
such as the von Mises or wrapped Cauchy family. Most notably, no assumptions
are required when data are sampled from the probability measure to be compared
with, which is in strict contrast to the real line. A bootstrap principle
follows immediately as our proof relies on Hadamard differentiability of the OT
functional. This paves the way for a variety of statistical inference tasks and
is exemplified for asymptotic OT based goodness of fit testing for circular
distributions. We discuss numerical implementation, consistency and investigate
its statistical power. For testing uniformity, it turns out that this approach
performs particularly well for unimodal alternatives and is almost as powerful
as Rayleigh's test, the most powerful invariant test for von Mises
alternatives. For regimes with many modes the circular OT test is less powerful
which is explained by the shape of the corresponding transport plan.
| stat |
Learning sparse linear dynamic networks in a hyper-parameter free
setting | We address the issue of estimating the topology and dynamics of sparse linear
dynamic networks in a hyperparameter-free setting. We propose a method to
estimate the network dynamics in a computationally efficient and parameter
tuning-free iterative framework known as SPICE (Sparse Iterative Covariance
Estimation). The estimated dynamics directly reveal the underlying topology.
Our approach does not assume that the network is undirected and is applicable
even with varying noise levels across the modules of the network. We also do
not assume any explicit prior knowledge on the network dynamics. Numerical
experiments with realistic dynamic networks illustrate the usefulness of our
method.
| stat |
Adapting BH to One- and Two-Way Classified Structures of Hypotheses | Multiple testing literature contains ample research on controlling false
discoveries for hypotheses classified according to one criterion, which we
refer to as one-way classified hypotheses. Although simultaneous classification
of hypotheses according to two different criteria, resulting in two-way
classified hypotheses, do often occur in scientific studies, no such research
has taken place yet, as far as we know, under this structure. This article
produces procedures, both in their oracle and data-adaptive forms, for
controlling the overall false discovery rate (FDR) across all hypotheses
effectively capturing the underlying one- or two-way classification structure.
They have been obtained by using results associated with weighted
Benjamini-Hochberg (BH) procedure in their more general forms providing
guidance on how to adapt the original BH procedure to the underlying one- or
two-way classification structure through an appropriate choice of the weights.
The FDR is maintained non-asymptotically by our proposed procedures in their
oracle forms under positive regression dependence on subset of null $p$-values
(PRDS) and in their data-adaptive forms under independence of the $p$-values.
Possible control of FDR for our data-adaptive procedures in certain scenarios
involving dependent $p$-values have been investigated through simulations. The
fact that our suggested procedures can be superior to contemporary practices
has been demonstrated through their applications in simulated scenarios and to
real-life data sets. While the procedures proposed here for two-way classified
hypotheses are new, the data-adaptive procedure obtained for one-way classified
hypotheses is alternative to and often more powerful than those proposed in Hu
et al. (2010).
| stat |
Estimating Tukey Depth Using Incremental Quantile Estimators | The concept of depth represents methods to measure how deep an arbitrary
point is positioned in a dataset and can be seen as the opposite of
outlyingness. It has proved very useful and a wide range of methods have been
developed based on the concept.
To address the well-known computational challenges associated with the depth
concept, we suggest to estimate Tukey depth contours using recently developed
incremental quantile estimators. The suggested algorithm can estimate depth
contours when the dataset in known in advance, but also recursively update and
even track Tukey depth contours for dynamically varying data stream
distributions. Tracking was demonstrated in a real-life data example where
changes in human activity was detected in real-time from accelerometer
observations.
| stat |
Learning Theory for Estimation of Animal Motion Submanifolds | This paper describes the formulation and experimental testing of a novel
method for the estimation and approximation of submanifold models of animal
motion. It is assumed that the animal motion is supported on a configuration
manifold $Q$ that is a smooth, connected, regularly embedded Riemannian
submanifold of Euclidean space $X\approx \mathbb{R}^d$ for some $d>0$, and that
the manifold $Q$ is homeomorphic to a known smooth, Riemannian manifold $S$.
Estimation of the manifold is achieved by finding an unknown mapping
$\gamma:S\rightarrow Q\subset X$ that maps the manifold $S$ into $Q$. The
overall problem is cast as a distribution-free learning problem over the
manifold of measurements $\mathbb{Z}=S\times X$. That is, it is assumed that
experiments generate a finite sets $\{(s_i,x_i)\}_{i=1}^m\subset \mathbb{Z}^m$
of samples that are generated according to an unknown probability density $\mu$
on $\mathbb{Z}$. This paper derives approximations $\gamma_{n,m}$ of $\gamma$
that are based on the $m$ samples and are contained in an $N(n)$ dimensional
space of approximants. The paper defines sufficient conditions that shows that
the rates of convergence in $L^2_\mu(S)$ correspond to those known for
classical distribution-free learning theory over Euclidean space. Specifically,
the paper derives sufficient conditions that guarantee rates of convergence
that have the form $$\mathbb{E} \left
(\|\gamma_\mu^j-\gamma_{n,m}^j\|_{L^2_\mu(S)}^2\right )\leq C_1 N(n)^{-r} + C_2
\frac{N(n)\log(N(n))}{m}$$for constants $C_1,C_2$ with
$\gamma_\mu:=\{\gamma^1_\mu,\ldots,\gamma^d_\mu\}$ the regressor function
$\gamma_\mu:S\rightarrow Q\subset X$ and
$\gamma_{n,m}:=\{\gamma^1_{n,j},\ldots,\gamma^d_{n,m}\}$.
| stat |
A novel topology design approach using an integrated deep learning
network architecture | Topology design optimization offers tremendous opportunity in design and
manufacturing freedoms by designing and producing a part from the ground-up
without a meaningful initial design as required by conventional shape design
optimization approaches. Ideally, with adequate problem statements, to
formulate and solve the topology design problem using a standard topology
optimization process, such as SIMP (Simplified Isotropic Material with
Penalization) is possible. In reality, an estimated over thousands of design
iterations is often required for just a few design variables, the conventional
optimization approach is in general impractical or computationally unachievable
for real world applications significantly diluting the development of the
topology optimization technology. There is, therefore, a need for a different
approach that will be able to optimize the initial design topology effectively
and rapidly. Therefore, this work presents a new topology design procedure to
generate optimal structures using an integrated Generative Adversarial Networks
(GANs) and convolutional neural network architecture.
| stat |
Semi-supervised learning in unbalanced and heterogeneous networks | Community detection was a hot topic on network analysis, where the main aim
is to perform unsupervised learning or clustering in networks. Recently,
semi-supervised learning has received increasing attention among researchers.
In this paper, we propose a new algorithm, called weighted inverse Laplacian
(WIL), for predicting labels in partially labeled networks. The idea comes from
the first hitting time in random walk, and it also has nice explanations both
in information propagation and the regularization framework. We propose a
partially labeled degree-corrected block model (pDCBM) to describe the
generation of partially labeled networks. We show that WIL ensures the
misclassification rate is of order $O(\frac{1}{d})$ for the pDCBM with average
degree $d=\Omega(\log n),$ and that it can handle situations with greater
unbalanced than traditional Laplacian methods. WIL outperforms other
state-of-the-art methods in most of our simulations and real datasets,
especially in unbalanced networks and heterogeneous networks.
| stat |
On the Stability of Random Matrix Product with Markovian Noise:
Application to Linear Stochastic Approximation and TD Learning | This paper studies the exponential stability of random matrix products driven
by a general (possibly unbounded) state space Markov chain. It is a cornerstone
in the analysis of stochastic algorithms in machine learning (e.g. for
parameter tracking in online learning or reinforcement learning). The existing
results impose strong conditions such as uniform boundedness of the
matrix-valued functions and uniform ergodicity of the Markov chains. Our main
contribution is an exponential stability result for the $p$-th moment of random
matrix product, provided that (i) the underlying Markov chain satisfies a
super-Lyapunov drift condition, (ii) the growth of the matrix-valued functions
is controlled by an appropriately defined function (related to the drift
condition). Using this result, we give finite-time $p$-th moment bounds for
constant and decreasing stepsize linear stochastic approximation schemes with
Markovian noise on general state space. We illustrate these findings for linear
value-function estimation in reinforcement learning. We provide finite-time
$p$-th moment bound for various members of temporal difference (TD) family of
algorithms.
| stat |
How little data do we need for patient-level prediction? | Objective: Provide guidance on sample size considerations for developing
predictive models by empirically establishing the adequate sample size, which
balances the competing objectives of improving model performance and reducing
model complexity as well as computational requirements.
Materials and Methods: We empirically assess the effect of sample size on
prediction performance and model complexity by generating learning curves for
81 prediction problems in three large observational health databases, requiring
training of 17,248 prediction models. The adequate sample size was defined as
the sample size for which the performance of a model equalled the maximum model
performance minus a small threshold value.
Results: The adequate sample size achieves a median reduction of the number
of observations between 9.5% and 78.5% for threshold values between 0.001 and
0.02. The median reduction of the number of predictors in the models at the
adequate sample size varied between 8.6% and 68.3%, respectively.
Discussion: Based on our results a conservative, yet significant, reduction
in sample size and model complexity can be estimated for future prediction
work. Though, if a researcher is willing to generate a learning curve a much
larger reduction of the model complexity may be possible as suggested by a
large outcome-dependent variability.
Conclusion: Our results suggest that in most cases only a fraction of the
available data was sufficient to produce a model close to the performance of
one developed on the full data set, but with a substantially reduced model
complexity.
| stat |
Bayesian Inference with Generative Adversarial Network Priors | Bayesian inference is used extensively to infer and to quantify the
uncertainty in a field of interest from a measurement of a related field when
the two are linked by a physical model. Despite its many applications, Bayesian
inference faces challenges when inferring fields that have discrete
representations of large dimension, and/or have prior distributions that are
difficult to represent mathematically. In this manuscript we consider the use
of Generative Adversarial Networks (GANs) in addressing these challenges. A GAN
is a type of deep neural network equipped with the ability to learn the
distribution implied by multiple samples of a given field. Once trained on
these samples, the generator component of a GAN maps the iid components of a
low-dimensional latent vector to an approximation of the distribution of the
field of interest. In this work we demonstrate how this approximate
distribution may be used as a prior in a Bayesian update, and how it addresses
the challenges associated with characterizing complex prior distributions and
the large dimension of the inferred field. We demonstrate the efficacy of this
approach by applying it to the problem of inferring and quantifying uncertainty
in the initial temperature field in a heat conduction problem from a noisy
measurement of the temperature at later time.
| stat |
Stopping Criterion for Active Learning Based on Error Stability | Active learning is a framework for supervised learning to improve the
predictive performance by adaptively annotating a small number of samples. To
realize efficient active learning, both an acquisition function that determines
the next datum and a stopping criterion that determines when to stop learning
should be considered. In this study, we propose a stopping criterion based on
error stability, which guarantees that the change in generalization error upon
adding a new sample is bounded by the annotation cost and can be applied to any
Bayesian active learning. We demonstrate that the proposed criterion stops
active learning at the appropriate timing for various learning models and real
datasets.
| stat |
Deep Latent-Variable Kernel Learning | Deep kernel learning (DKL) leverages the connection between Gaussian process
(GP) and neural networks (NN) to build an end-to-end, hybrid model. It combines
the capability of NN to learn rich representations under massive data and the
non-parametric property of GP to achieve automatic regularization that
incorporates a trade-off between model fit and model complexity. However, the
deterministic encoder may weaken the model regularization of the following GP
part, especially on small datasets, due to the free latent representation. We
therefore present a complete deep latent-variable kernel learning (DLVKL) model
wherein the latent variables perform stochastic encoding for regularized
representation. We further enhance the DLVKL from two aspects: (i) the
expressive variational posterior through neural stochastic differential
equation (NSDE) to improve the approximation quality, and (ii) the hybrid prior
taking knowledge from both the SDE prior and the posterior to arrive at a
flexible trade-off. Intensive experiments imply that the DLVKL-NSDE performs
similarly to the well calibrated GP on small datasets, and outperforms existing
deep GPs on large datasets.
| stat |
Analysis of a DNA mixture case involving Romani reference populations | Here we present an Italian criminal case that shows how statistical methods
can be used to extract information from a series of mixed DNA profiles. The
case involves several different individuals and a set of different DNA traces.
The case possibly involves persons of interest of a small, inbred population of
Romani origin. First, a brief description of the case is provided. Secondly, we
introduce some heuristic tools that can be used to evaluate the data and
briefly outline the statistical model used for analysing DNA mixtures. Finally,
we illustrate some of the findings on the case and discuss further directions
of research. The results show how the use of different population database
allele frequencies for analysing the DNA mixtures can lead to very different
results, some seemingly inculpatory and some seemingly exculpatory. We also
illustrate the results obtained from combining the evidence from different
samples.
| stat |
Bayesian Mixed Effect Sparse Tensor Response Regression Model with Joint
Estimation of Activation and Connectivity | Brain activation and connectivity analyses in task-based functional magnetic
resonance imaging (fMRI) experiments with multiple subjects are currently at
the forefront of data-driven neuroscience. In such experiments, interest often
lies in understanding activation of brain voxels due to external stimuli and
strong association or connectivity between the measurements on a set of
pre-specified group of brain voxels, also known as regions of interest (ROI).
This article proposes a joint Bayesian additive mixed modeling framework that
simultaneously assesses brain activation and connectivity patterns from
multiple subjects. In particular, fMRI measurements from each individual
obtained in the form of a multi-dimensional array/tensor at each time are
regressed on functions of the stimuli. We impose a low-rank PARAFAC
decomposition on the tensor regression coefficients corresponding to the
stimuli to achieve parsimony. Multiway stick breaking shrinkage priors are
employed to infer activation patterns and associated uncertainties in each
voxel. Further, the model introduces region specific random effects which are
jointly modeled with a Bayesian Gaussian graphical prior to account for the
connectivity among pairs of ROIs. Empirical investigations under various
simulation studies demonstrate the effectiveness of the method as a tool to
simultaneously assess brain activation and connectivity. The method is then
applied to a multi-subject fMRI dataset from a balloon-analog risk-taking
experiment in order to make inference about how the brain processes risk.
| stat |
Clustering microbiome data using mixtures of logistic normal multinomial
models | Discrete data such as counts of microbiome taxa resulting from
next-generation sequencing are routinely encountered in bioinformatics. Taxa
count data in microbiome studies are typically high-dimensional,
over-dispersed, and can only reveal relative abundance therefore being treated
as compositional. Analyzing compositional data presents many challenges because
they are restricted on a simplex. In a logistic normal multinomial model, the
relative abundance is mapped from a simplex to a latent variable that exists on
the real Euclidean space using the additive log-ratio transformation. While a
logistic normal multinomial approach brings in flexibility for modeling the
data, it comes with a heavy computational cost as the parameter estimation
typically relies on Bayesian techniques. In this paper, we develop a novel
mixture of logistic normal multinomial models for clustering microbiome data.
Additionally, we utilize an efficient framework for parameter estimation using
variational Gaussian approximations (VGA). Adopting a variational Gaussian
approximation for the posterior of the latent variable reduces the
computational overhead substantially. The proposed method is illustrated on
simulated and real datasets.
| stat |
Spatial Functional Data Modeling of Plant Reflectances | Plant reflectance spectra - the profile of light reflected by leaves across
different wavelengths - supply the spectral signature for a species at a
spatial location to enable estimation of functional and taxonomic diversity for
plants. We consider leaf spectra as "responses" to be explained spatially.
These spectra/reflectances are functions over a wavelength band that respond to
the environment.
Our motivating data are gathered for several families from the Cape Floristic
Region (CFR) in South Africa and lead us to develop rich novel spatial models
that can explain spectra for genera within families. Wavelength responses for
an individual leaf are viewed as a function of wavelength, leading to
functional data modeling. Local environmental features become covariates. We
introduce wavelength - covariate interaction since the response to
environmental regressors may vary with wavelength, so may variance. Formal
spatial modeling enables prediction of reflectances for genera at unobserved
locations with known environmental features. We incorporate spatial dependence,
wavelength dependence, and space-wavelength interaction (in the spirit of
space-time interaction). We implement out-of-sample validation to select a best
model, discovering that the model features listed above are all informative for
the functional data analysis. We then supply interpretation of the results
under the selected model.
| stat |
Active learning for level set estimation under cost-dependent input
uncertainty | As part of a quality control process in manufacturing it is often necessary
to test whether all parts of a product satisfy a required property, with as few
inspections as possible. When multiple inspection apparatuses with different
costs and precision exist, it is desirable that testing can be carried out
cost-effectively by properly controlling the trade-off between the costs and
the precision. In this paper, we formulate this as a level set estimation (LSE)
problem under cost-dependent input uncertainty - LSE being a type of active
learning for estimating the level set, i.e., the subset of the input space in
which an unknown function value is greater or smaller than a pre-determined
threshold. Then, we propose a new algorithm for LSE under cost-dependent input
uncertainty with theoretical convergence guarantee. We demonstrate the
effectiveness of the proposed algorithm by applying it to synthetic and real
datasets.
| stat |
Optimal Transport Relaxations with Application to Wasserstein GANs | We propose a family of relaxations of the optimal transport problem which
regularize the problem by introducing an additional minimization step over a
small region around one of the underlying transporting measures. The type of
regularization that we obtain is related to smoothing techniques studied in the
optimization literature. When using our approach to estimate optimal transport
costs based on empirical measures, we obtain statistical learning bounds which
are useful to guide the amount of regularization, while maintaining good
generalization properties. To illustrate the computational advantages of our
regularization approach, we apply our method to training Wasserstein GANs. We
obtain running time improvements, relative to current benchmarks, with no
deterioration in testing performance (via FID). The running time improvement
occurs because our new optimality-based threshold criterion reduces the number
of expensive iterates of the generating networks, while increasing the number
of actor-critic iterations.
| stat |
On the Occasional Exactness of the Distributional Transform
Approximation for Direct Gaussian Copula Models with Discrete Margins | The direct Gaussian copula model with discrete marginal distributions is an
appealing data-analytic tool but poses difficult computational challenges due
to its intractable likelihood. A number of approximations/surrogates for the
likelihood have been proposed, including the continuous extension-based
approximation (CE) and the distributional transform-based approximation (DT).
The continuous extension approach is exact up to Monte Carlo error but does not
scale well computationally. The distributional transform approach permits
efficient computation but offers no theoretical guarantee that it is exact. In
practice, though, the distributional transform-based approximate likelihood is
so very nearly exact for some variants of the model as to permit genuine
maximum likelihood or Bayesian inference. We demonstrate the exactness of the
distributional transform-based objective function for two interesting variants
of the model, and propose a quantity that can be used to assess exactness for
experimentally observed datasets. Said diagnostic will permit practitioners to
determine whether genuine Bayesian inference or ordinary maximum likelihood
inference using the DT-based likelihood is possible for a given dataset.
| stat |
Dynamic Batch Learning in High-Dimensional Sparse Linear Contextual
Bandits | We study the problem of dynamic batch learning in high-dimensional sparse
linear contextual bandits, where a decision maker, under a given
maximum-number-of-batch constraint and only able to observe rewards at the end
of each batch, can dynamically decide how many individuals to include in the
next batch (at the end of the current batch) and what personalized
action-selection scheme to adopt within each batch. Such batch constraints are
ubiquitous in a variety of practical contexts, including personalized product
offerings in marketing and medical treatment selection in clinical trials. We
characterize the fundamental learning limit in this problem via a regret lower
bound and provide a matching upper bound (up to log factors), thus prescribing
an optimal scheme for this problem. To the best of our knowledge, our work
provides the first inroad into a theoretical understanding of dynamic batch
learning in high-dimensional sparse linear contextual bandits. Notably, even a
special case of our result (when no batch constraint is present) yields the
first minimax optimal $\tilde{O}(\sqrt{s_0T})$ regret bound for standard online
learning in high-dimensional linear contextual bandits (for the no-margin
case), where $s_0$ is the sparsity parameter (or an upper bound thereof) and
$T$ is the learning horizon. This result (both that $\tilde{O}(\sqrt{s_0 T})$
is achievable and that $\Omega(\sqrt{s_0 T})$ is a lower bound) appears to be
unknown in the emerging literature of high-dimensional contextual bandits.
| stat |
On gradient regularizers for MMD GANs | We propose a principled method for gradient-based regularization of the
critic of GAN-like models trained by adversarially optimizing the kernel of a
Maximum Mean Discrepancy (MMD). We show that controlling the gradient of the
critic is vital to having a sensible loss function, and devise a method to
enforce exact, analytical gradient constraints at no additional cost compared
to existing approximate techniques based on additive regularizers. The new loss
function is provably continuous, and experiments show that it stabilizes and
accelerates training, giving image generation models that outperform
state-of-the art methods on $160 \times 160$ CelebA and $64 \times 64$
unconditional ImageNet.
| stat |
Heteroscedastic Gaussian Process Regression on the Alkenone over Sea
Surface Temperatures | To restore the historical sea surface temperatures (SSTs) better, it is
important to construct a good calibration model for the associated proxies. In
this paper, we introduce a new model for alkenone (${\rm{U}}_{37}^{\rm{K}'}$)
based on the heteroscedastic Gaussian process (GP) regression method. Our
nonparametric approach not only deals with the variable pattern of noises over
SSTs but also contains a Bayesian method of classifying potential outliers.
| stat |
Mendelian Randomization with Incomplete Exposure Data: a Bayesian
Approach | We expand Mendelian Randomization (MR) methodology to deal with randomly
missing data on either the exposure or the outcome variable, and furthermore
with data from nonindependent individuals (eg components of a family). Our
method rests on the Bayesian MR framework proposed by Berzuini et al (2018),
which we apply in a study of multiplex Multiple Sclerosis (MS) Sardinian
families to characterise the role of certain plasma proteins in MS causation.
The method is robust to presence of pleiotropic effects in an unknown number of
instruments, and is able to incorporate inter-individual kinship information.
Introduction of missing data allows us to overcome the bias introduced by the
(reverse) effect of treatment (in MS cases) on level of protein. From a
substantive point of view, our study results confirm recent suspicion that an
increase in circulating IL12A and STAT4 protein levels does not cause an
increase in MS risk, as originally believed, suggesting that these two proteins
may not be suitable drug targets for MS.
| stat |
Revisiting ICH E9 (R1) during the COVID-19 pandemic | The current COVID-19 pandemic poses numerous challenges for ongoing clinical
trials and provides a stress-testing environment for the existing principles
and practice of estimands in clinical trials. The pandemic may increase the
rate of intercurrent events (ICEs) and missing values, spurring a great deal of
discussion on amending protocols and statistical analysis plans to address
these issues. In this article we revisit recent research on estimands and
handling of missing values, especially the ICH E9 (R1) on Estimands and
Sensitivity Analysis in Clinical Trials. Based on an in-depth discussion of the
strategies for handling ICEs using a causal inference framework, we suggest
some improvements in applying the estimand and estimation framework in ICH E9
(R1). Specifically, we discuss a mix of strategies allowing us to handle ICEs
differentially based on reasons for ICEs. We also suggest ICEs should be
handled primarily by hypothetical strategies and provide examples of different
hypothetical strategies for different types of ICEs as well as a road map for
estimation and sensitivity analyses. We conclude that the proposed framework
helps streamline translating clinical objectives into targets of statistical
inference and automatically resolves many issues with defining estimands and
choosing estimation procedures arising from events such as the pandemic.
| stat |
Simulations evaluating resampling methods for causal discovery: ensemble
performance and calibration | Causal discovery can be a powerful tool for investigating causality when a
system can be observed but is inaccessible to experiments in practice. Despite
this, it is rarely used in any scientific or medical fields. One of the major
hurdles preventing the field of causal discovery from having a larger impact is
that it is difficult to determine when the output of a causal discovery method
can be trusted in a real-world setting. Trust is especially critical when human
health is on the line.
In this paper, we report the results of a series of simulation studies
investigating the performance of different resampling methods as indicators of
confidence in discovered graph features. We found that subsampling and sampling
with replacement both performed surprisingly well, suggesting that they can
serve as grounds for confidence in graph features. We also found that the
calibration of subsampling and sampling with replacement had different
convergence properties, suggesting that one's choice of which to use should
depend on the sample size.
| stat |
Semiparametric Inference for Non-monotone Missing-Not-at-Random Data:
the No Self-Censoring Model | We study the identification and estimation of statistical functionals of
multivariate data missing non-monotonically and not-at-random, taking a
semiparametric approach. Specifically, we assume that the missingness mechanism
satisfies what has been previously called "no self-censoring" or "itemwise
conditionally independent nonresponse," which roughly corresponds to the
assumption that no partially-observed variable directly determines its own
missingness status. We show that this assumption, combined with an odds ratio
parameterization of the joint density, enables identification of functionals of
interest, and we establish the semiparametric efficiency bound for the
nonparametric model satisfying this assumption. We propose a practical
augmented inverse probability weighted estimator, and in the setting with a
(possibly high-dimensional) always-observed subset of covariates, our proposed
estimator enjoys a certain double-robustness property. We explore the
performance of our estimator with simulation experiments and on a
previously-studied data set of HIV-positive mothers in Botswana.
| stat |
Methods and Software for the Multilevel Social Relations Model: A
Tutorial | This tutorial demonstrates the estimation and interpretation of the
Multilevel Social Relations Model for dyadic data. The Social Relations Model
is appropriate for data structures in which individuals appear multiple times
as both the source and recipient of dyadic outcomes. Estimated using Stat-JR
statistical software, the models are fitted to multiple outcome types:
continuous, count, and binary outcomes. In addition, models are demonstrated
for dyadic data from a single group and from multiple groups. The modeling
approaches are illustrated via a series of case studies, and the data and
software to replicate these analyses are available as supplemental files.
| stat |
Elements of Sequential Monte Carlo | A core problem in statistics and probabilistic machine learning is to compute
probability distributions and expectations. This is the fundamental problem of
Bayesian statistics and machine learning, which frames all inference as
expectations with respect to the posterior distribution. The key challenge is
to approximate these intractable expectations. In this tutorial, we review
sequential Monte Carlo (SMC), a random-sampling-based class of methods for
approximate inference. First, we explain the basics of SMC, discuss practical
issues, and review theoretical results. We then examine two of the main user
design choices: the proposal distributions and the so called intermediate
target distributions. We review recent results on how variational inference and
amortization can be used to learn efficient proposals and target distributions.
Next, we discuss the SMC estimate of the normalizing constant, how this can be
used for pseudo-marginal inference and inference evaluation. Throughout the
tutorial we illustrate the use of SMC on various models commonly used in
machine learning, such as stochastic recurrent neural networks, probabilistic
graphical models, and probabilistic programs.
| stat |
A functional autoregressive model based on exogenous hydrometeorological
variables for river flow prediction | In this research, a functional time series model was introduced to predict
future realizations of river flow time series. The proposed model was
constructed based on a functional time series's correlated lags and the
essential exogenous climate variables. Rainfall, temperature, and evaporation
variables were hypothesized to have substantial functionality in river flow
simulation. Because an actual time series model is unspecified and the input
variables' significance for the learning process is unknown in practice, it was
employed a variable selection procedure to determine only the significant
variables for the model. A nonparametric bootstrap model was also proposed to
investigate predictions' uncertainty and construct pointwise prediction
intervals for the river flow curve time series. Historical datasets at three
meteorological stations (Mosul, Baghdad, and Kut) located in the semi-arid
region, Iraq, were used for model development. The prediction performance of
the proposed model was validated against existing functional and traditional
time series models. The numerical analyses revealed that the proposed model
provides competitive or even better performance than the benchmark models.
Also, the incorporated exogenous climate variables have substantially improved
the modeling predictability performance. Overall, the proposed model indicated
a reliable methodology for modeling river flow within the semi-arid region.
| stat |
Fiedler Regularization: Learning Neural Networks with Graph Sparsity | We introduce a novel regularization approach for deep learning that
incorporates and respects the underlying graphical structure of the neural
network. Existing regularization methods often focus on dropping/penalizing
weights in a global manner that ignores the connectivity structure of the
neural network. We propose to use the Fiedler value of the neural network's
underlying graph as a tool for regularization. We provide theoretical support
for this approach via spectral graph theory. We list several useful properties
of the Fiedler value that makes it suitable in regularization. We provide an
approximate, variational approach for fast computation in practical training of
neural networks. We provide bounds on such approximations. We provide an
alternative but equivalent formulation of this framework in the form of a
structurally weighted L1 penalty, thus linking our approach to sparsity
induction. We performed experiments on datasets that compare Fiedler
regularization with traditional regularization methods such as dropout and
weight decay. Results demonstrate the efficacy of Fiedler regularization.
| stat |
Training Invertible Linear Layers through Rank-One Perturbations | Many types of neural network layers rely on matrix properties such as
invertibility or orthogonality. Retaining such properties during optimization
with gradient-based stochastic optimizers is a challenging task, which is
usually addressed by either reparameterization of the affected parameters or by
directly optimizing on the manifold. This work presents a novel approach for
training invertible linear layers. In lieu of directly optimizing the network
parameters, we train rank-one perturbations and add them to the actual weight
matrices infrequently. This P$^{4}$Inv update allows keeping track of inverses
and determinants without ever explicitly computing them. We show how such
invertible blocks improve the mixing and thus the mode separation of the
resulting normalizing flows. Furthermore, we outline how the P$^4$ concept can
be utilized to retain properties other than invertibility.
| stat |
The Jensen Effect and Functional Single Index Models: Estimating the
Ecological Implications of Nonlinear Reaction Norms | This paper develops tools to characterize how species are affected by
environmental variability, based on a functional single index model relating a
response such as growth rate or survival to environmental conditions. In
ecology, the curvature of such responses are used, via Jensen's inequality, to
determine whether environmental variability is harmful or beneficial, and
differing nonlinear responses to environmental variability can contribute to
the coexistence of competing species.
Here, we address estimation and inference for these models with observational
data on individual responses to environmental conditions. Because nonparametric
estimation of the curvature (second derivative) in a nonparametric functional
single index model requires unrealistic sample sizes, we instead focus on
directly estimating the effect of the nonlinearity, by comparing the average
response to a variable environment with the response at the expected
environment, which we call the Jensen Effect. We develop a test statistic to
assess whether this effect is significantly different from zero. In doing so we
re-interpret the SiZer method of Chaudhuri and Marron (1995) by maximizing a
test statistic over smoothing parameters. We show that our proposed method
works well both in simulations and on real ecological data from the long-term
data set described in Drake (2005).
| stat |
Semi-supervised empirical Bayes group-regularized factor regression | The features in high dimensional biomedical prediction problems are often
well described with lower dimensional manifolds. An example is genes that are
organised in smaller functional networks. The outcome can then be described
with the factor regression model. A benefit of the factor model is that is
allows for straightforward inclusion of unlabeled observations in the
estimation of the model, i.e., semi-supervised learning. In addition, the high
dimensional features in biomedical prediction problems are often well
characterised. Examples are genes, for which annotation is available, and
metabolites with $p$-values from a previous study available. In this paper, the
extra information on the features is included in the prior model for the
features. The extra information is weighted and included in the estimation
through empirical Bayes, with Variational approximations to speed up the
computation. The method is demonstrated in simulations and two applications.
One application considers influenza vaccine efficacy prediction based on
microarray data. The second application predictions oral cancer metastatsis
from RNAseq data.
| stat |
Tensor denoising and completion based on ordinal observations | Higher-order tensors arise frequently in applications such as neuroimaging,
recommendation system, social network analysis, and psychological studies. We
consider the problem of low-rank tensor estimation from possibly incomplete,
ordinal-valued observations. Two related problems are studied, one on tensor
denoising and the other on tensor completion. We propose a multi-linear
cumulative link model, develop a rank-constrained M-estimator, and obtain
theoretical accuracy guarantees. Our mean squared error bound enjoys a faster
convergence rate than previous results, and we show that the proposed estimator
is minimax optimal under the class of low-rank models. Furthermore, the
procedure developed serves as an efficient completion method which guarantees
consistent recovery of an order-$K$ $(d,\ldots,d)$-dimensional low-rank tensor
using only $\tilde{\mathcal{O}}(Kd)$ noisy, quantized observations. We
demonstrate the outperformance of our approach over previous methods on the
tasks of clustering and collaborative filtering.
| stat |
Estimating linear covariance models with numerical nonlinear algebra | Numerical nonlinear algebra is applied to maximum likelihood estimation for
Gaussian models defined by linear constraints on the covariance matrix. We
examine the generic case as well as special models (e.g. Toeplitz, sparse,
trees) that are of interest in statistics. We study the maximum likelihood
degree and its dual analogue, and we introduce a new software package
LinearCovarianceModels.jl for solving the score equations. All local maxima can
thus be computed reliably. In addition we identify several scenarios for which
the estimator is a rational function.
| stat |
A Function Emulation Approach for Doubly Intractable Distributions | Doubly intractable distributions arise in many settings, for example in
Markov models for point processes and exponential random graph models for
networks. Bayesian inference for these models is challenging because they
involve intractable normalising "constants" that are actually functions of the
parameters of interest. Although several clever computational methods have been
developed for these models, each method suffers from computational issues that
makes it computationally burdensome or even infeasible for many problems. We
propose a novel algorithm that provides computational gains over existing
methods by replacing Monte Carlo approximations to the normalising function
with a Gaussian process-based approximation. We provide theoretical
justification for this method. We also develop a closely related algorithm that
is applicable more broadly to any likelihood function that is expensive to
evaluate. We illustrate the application of our methods to a variety of
challenging simulated and real data examples, including an exponential random
graph model, a Markov point process, and a model for infectious disease
dynamics. The algorithm shows significant gains in computational efficiency
over existing methods, and has the potential for greater gains for more
challenging problems. For a random graph model example, we show how this gain
in efficiency allows us to carry out accurate Bayesian inference when other
algorithms are computationally impractical.
| stat |
Separating common (global and local) and distinct variation in multiple
mixed types data sets | Multiple sets of measurements on the same objects obtained from different
platforms may reflect partially complementary information of the studied
system. The integrative analysis of such data sets not only provides us with
the opportunity of a deeper understanding of the studied system, but also
introduces some new statistical challenges. First, the separation of
information that is common across all or some of the data sets, and the
information that is specific to each data set is problematic. Furthermore,
these data sets are often a mix of quantitative and discrete (binary or
categorical) data types, while commonly used data fusion methods require all
data sets to be quantitative. In this paper, we propose an exponential family
simultaneous component analysis (ESCA) model to tackle the potential mixed data
types problem of multiple data sets. In addition, a structured sparse pattern
of the loading matrix is induced through a nearly unbiased group concave
penalty to disentangle the global, local common and distinct information of the
multiple data sets. A Majorization-Minimization based algorithm is derived to
fit the proposed model. Analytic solutions are derived for updating all the
parameters of the model in each iteration, and the algorithm will decrease the
objective function in each iteration monotonically. For model selection, a
missing value based cross validation procedure is implemented. The advantages
of the proposed method in comparison with other approaches are assessed using
comprehensive simulations as well as the analysis of real data from a chronic
lymphocytic leukaemia (CLL) study.
Availability: the codes to reproduce the results in this article are
available at https://gitlab.com/uvabda.
| stat |
Beyond unidimensional poverty analysis using distributional copula
models for mixed ordered-continuous outcomes | Poverty is a multidimensional concept often comprising a monetary outcome and
other welfare dimensions such as education, subjective well-being or health,
that are measured on an ordinal scale. In applied research, multidimensional
poverty is ubiquitously assessed by studying each poverty dimension
independently in univariate regression models or by combining several poverty
dimensions into a scalar index. This inhibits a thorough analysis of the
potentially varying interdependence between the poverty dimensions. We propose
a multivariate copula generalized additive model for location, scale and shape
(copula GAMLSS or distributional copula model) to tackle this challenge. By
relating the copula parameter to covariates, we specifically examine if certain
factors determine the dependence between poverty dimensions. Furthermore,
specifying the full conditional bivariate distribution, allows us to derive
several features such as poverty risks and dependence measures coherently from
one model for different individuals. We demonstrate the approach by studying
two important poverty dimensions: income and education. Since the level of
education is measured on an ordinal scale while income is continuous, we extend
the bivariate copula GAMLSS to the case of mixed ordered-continuous outcomes.
The new model is integrated into the GJRM package in R and applied to data from
Indonesia. Particular emphasis is given to the spatial variation of the
income-education dependence and groups of individuals at risk of being
simultaneously poor in both education and income dimensions.
| stat |
Clustering multivariate functional data using unsupervised binary trees | We propose a model-based clustering algorithm for a general class of
functional data for which the components could be curves or images. The random
functional data realizations could be measured with error at discrete, and
possibly random, points in the definition domain. The idea is to build a set of
binary trees by recursive splitting of the observations. The number of groups
are determined in a data-driven way. The new algorithm provides easily
interpretable results and fast predictions for online data sets. Results on
simulated datasets reveal good performance in various complex settings. The
methodology is applied to the analysis of vehicle trajectories on a German
roundabout.
| stat |
Provable Convex Co-clustering of Tensors | Cluster analysis is a fundamental tool for pattern discovery of complex
heterogeneous data. Prevalent clustering methods mainly focus on vector or
matrix-variate data and are not applicable to general-order tensors, which
arise frequently in modern scientific and business applications. Moreover,
there is a gap between statistical guarantees and computational efficiency for
existing tensor clustering solutions due to the nature of their non-convex
formulations. In this work, we bridge this gap by developing a provable convex
formulation of tensor co-clustering. Our convex co-clustering (CoCo) estimator
enjoys stability guarantees and its computational and storage costs are
polynomial in the size of the data. We further establish a non-asymptotic error
bound for the CoCo estimator, which reveals a surprising "blessing of
dimensionality" phenomenon that does not exist in vector or matrix-variate
cluster analysis. Our theoretical findings are supported by extensive simulated
studies. Finally, we apply the CoCo estimator to the cluster analysis of
advertisement click tensor data from a major online company. Our clustering
results provide meaningful business insights to improve advertising
effectiveness.
| stat |
Atlantic Causal Inference Conference (ACIC) Data Analysis Challenge 2017 | This brief note documents the data generating processes used in the 2017 Data
Analysis Challenge associated with the Atlantic Causal Inference Conference
(ACIC). The focus of the challenge was estimation and inference for conditional
average treatment effects (CATEs) in the presence of targeted selection, which
leads to strong confounding. The associated data files and further plots can be
found on the first author's web page.
| stat |
BaCOUn: Bayesian Classifers with Out-of-Distribution Uncertainty | Traditional training of deep classifiers yields overconfident models that are
not reliable under dataset shift. We propose a Bayesian framework to obtain
reliable uncertainty estimates for deep classifiers. Our approach consists of a
plug-in "generator" used to augment the data with an additional class of points
that lie on the boundary of the training data, followed by Bayesian inference
on top of features that are trained to distinguish these "out-of-distribution"
points.
| stat |
Efficient multiply robust imputation in the presence of influential
units in surveys | Item nonresponse is a common issue in surveys. Because unadjusted estimators
may be biased in the presence of nonresponse, it is common practice to impute
the missing values with the objective of reducing the nonresponse bias as much
as possible. However, commonly used imputation procedures may lead to unstable
estimators of population totals/means when influential units are present in the
set of respondents. In this article, we consider the class of multiply robust
imputation procedures that provide some protection against the failure of
underlying model assumptions. We develop an efficient version of multiply
robust estimators based on the concept of conditional bias, a measure of
influence. We present the results of a simulation study to show the benefits of
the proposed method in terms of bias and efficiency.
| stat |
Ask-n-Learn: Active Learning via Reliable Gradient Representations for
Image Classification | Deep predictive models rely on human supervision in the form of labeled
training data. Obtaining large amounts of annotated training data can be
expensive and time consuming, and this becomes a critical bottleneck while
building such models in practice. In such scenarios, active learning (AL)
strategies are used to achieve faster convergence in terms of labeling efforts.
Existing active learning employ a variety of heuristics based on uncertainty
and diversity to select query samples. Despite their wide-spread use, in
practice, their performance is limited by a number of factors including
non-calibrated uncertainties, insufficient trade-off between data exploration
and exploitation, presence of confirmation bias etc. In order to address these
challenges, we propose Ask-n-Learn, an active learning approach based on
gradient embeddings obtained using the pesudo-labels estimated in each
iteration of the algorithm. More importantly, we advocate the use of prediction
calibration to obtain reliable gradient embeddings, and propose a data
augmentation strategy to alleviate the effects of confirmation bias during
pseudo-labeling. Through empirical studies on benchmark image classification
tasks (CIFAR-10, SVHN, Fashion-MNIST, MNIST), we demonstrate significant
improvements over state-of-the-art baselines, including the recently proposed
BADGE algorithm.
| stat |
Comparison Between Bayesian and Frequentist Tail Probability Estimates | In this paper, we investigate the reasons that the Bayesian estimator of the
tail probability is always higher than the frequentist estimator. Sufficient
conditions for this phenomenon are established both by using Jensen's
Inequality and by looking at Taylor series approximations, both of which point
to the convexity of the distribution function.
| stat |
A Test for Independence Via Bayesian Nonparametric Estimation of Mutual
Information | Mutual information is a well-known tool to measure the mutual dependence
between variables. In this paper, a Bayesian nonparametric estimation of mutual
information is established by means of the Dirichlet process and the
$k$-nearest neighbor distance. As a direct outcome of the estimation, an
easy-to-implement test of independence is introduced through the relative
belief ratio. Several theoretical properties of the approach are presented. The
procedure is investigated through various examples where the results are
compared to its frequentist counterpart and demonstrate a good performance.
| stat |
Comparing clusterings and numbers of clusters by aggregation of
calibrated clustering validity indexes | A key issue in cluster analysis is the choice of an appropriate clustering
method and the determination of the best number of clusters. Different
clusterings are optimal on the same data set according to different criteria,
and the choice of such criteria depends on the context and aim of clustering.
Therefore, researchers need to consider what data analytic characteristics the
clusters they are aiming at are supposed to have, among others within-cluster
homogeneity, between-clusters separation, and stability. Here, a set of
internal clustering validity indexes measuring different aspects of clustering
quality is proposed, including some indexes from the literature. Users can
choose the indexes that are relevant in the application at hand. In order to
measure the overall quality of a clustering (for comparing clusterings from
different methods and/or different numbers of clusters), the index values are
calibrated for aggregation. Calibration is relative to a set of random
clusterings on the same data. Two specific aggregated indexes are proposed and
compared with existing indexes on simulated and real data.
| stat |
Empirical and Constrained Empirical Bayes Variance Estimation Under A
One Unit Per Stratum Sample Design | A single primary sampling unit (PSU) per stratum design is a popular design
for estimating the parameter of interest. Although, the point estimator of the
design is unbiased and efficient, an unbiased variance estimator does not
exist. A common practice to solve this is to collapse or combine the two
adjacent strata, but the attained estimator of variance is not design-unbiased,
and the bias increases as the population means of the collapsed strata become
more variant. Therefore, the one PSU per stratum design with collapsed stratum
variance estimator might not be a good choice, and some statisticians prefer a
design in which two PSUs per stratum are selected. In this paper, we first
compare a one PSU per stratum design to a two PSUs per stratum design. Then, we
propose an empirical Bayes estimator for the variance of one PSU per stratum
design, where it over-shrinks towards the prior mean. To protect against this,
we investigate the potential of a constrained empirical Bayes estimator.
Through a simulation study, we show that the empirical Bayes and constrained
empirical Bayes estimators outperform the classical collapsed one in terms of
empirical relative mean squared error.
| stat |
DOPE: D-Optimal Pooling Experimental design with application for
SARS-CoV-2 screening | Testing individuals for the presence of severe acute respiratory syndrome
coronavirus 2 (SARS-CoV-2), the pathogen causing the coronavirus disease 2019
(COVID-19), is crucial for curtailing transmission chains. Moreover, rapidly
testing many potentially infected individuals is often a limiting factor in
controlling COVID-19 outbreaks. Hence, pooling strategies, wherein individuals
are grouped and tested simultaneously, are employed. We present a novel pooling
strategy that implements D-Optimal Pooling Experimental design (DOPE). DOPE
defines optimal pooled tests as those maximizing the mutual information between
data and infection states. We estimate said mutual information via Monte-Carlo
sampling and employ a discrete optimization heuristic for maximizing it. DOPE
outperforms common pooling strategies both in terms of lower error rates and
fewer tests utilized. DOPE holds several additional advantages: it provides
posterior distributions of the probability of infection, rather than only
binary classification outcomes; it naturally incorporates prior information of
infection probabilities and test error rates; and finally, it can be easily
extended to include other, newly discovered information regarding COVID-19.
Hence, we believe that implementation of Bayesian D-optimal experimental design
holds a great promise for the efforts of combating COVID-19 and other future
pandemics.
| stat |
Bayesian analysis of ranking data with the constrained Extended
Plackett-Luce model | Multistage ranking models, including the popular Plackett-Luce distribution
(PL), rely on the assumption that the ranking process is performed
sequentially, by assigning the positions from the top to the bottom one
(forward order). A recent contribution to the ranking literature relaxed this
assumption with the addition of the discrete-valued reference order parameter,
yielding the novel Extended Plackett-Luce model (EPL). Inference on the EPL and
its generalization into a finite mixture framework was originally addressed
from the frequentist perspective. In this work, we propose the Bayesian
estimation of the EPL with order constraints on the reference order parameter.
The proposed restrictions reflect a meaningful rank assignment process. By
combining the restrictions with the data augmentation strategy and the
conjugacy of the Gamma prior distribution with the EPL, we facilitate the
construction of a tuned joint Metropolis-Hastings algorithm within Gibbs
sampling to simulate from the posterior distribution. The Bayesian approach
allows to address more efficiently the inference on the additional
discrete-valued parameter and the assessment of its estimation uncertainty. The
usefulness of the proposal is illustrated with applications to simulated and
real datasets.
| stat |
Characterizing the Probability Law on Time Until Core Damage With PRA | Certain modeling assumptions underlying Probabilistic Risk Assessment (PRA)
allow a simple computation of core damage frequency (CDF). These assumptions
also guarantee that the time remaining until a core damage event follows an
exponential distribution having parameter value equal to that computed for the
CDF. While it is commonly understood that these modeling assumptions lead to an
approximate characterization of uncertainty, we offer a simple argument that
explains why the resulting exponential time until core damage distribution
under-estimates risk. Our explanation will first review operational physics
properties of hazard functions, and then offer a non-measure-theoretic argument
to reveal the the consequences of these properties for PRA. The conclusions
offered, here, hold for any possible operating history that respects the
underlying assumptions of PRA. Hence, the measure-theoretic constructs on
filtered probability spaces is unnecessary for our developments. We will then
conclude with a brief discussion that connects intuition with our analytical
development.
| stat |
Nonparametric Estimation of Repeated Densities with Heterogeneous Sample
Sizes | We consider the estimation of densities in multiple subpopulations, where the
available sample size in each subpopulation greatly varies. For example, in
epidemiology, different diseases may share similar pathogenic mechanism but
differ in their prevalence. Without specifying a parametric form, our proposed
approach pools information from the population and estimate the density in each
subpopulation in a data-driven fashion. Low-dimensional approximating density
families in the form of exponential families are constructed from the principal
modes of variation in the log-densities, within which subpopulation densities
are then fitted based on likelihood principles and shrinkage. The approximating
families increase in their flexibility as the number of components increases
and can approximate arbitrary infinite-dimensional densities with discrete
observations, for which we derived convergence results. The proposed methods
are shown to be interpretable and efficient in simulation as well as
applications to electronic medical record and rainfall data.
| stat |
Statistical Inference on the Cure Time | In population-based cancer survival analysis, the net survival is important
for government to assess health care programs. For decades, it is observed that
the net survival reaches a plateau after long-term follow-up, this is so called
``statistical cure''. Several methods were proposed to address the statistical
cure. Besides, the cure time can be used to evaluate the time period of a
health care program for a specific patient population, and it also can be
helpful for a clinician to explain the prognosis for patients, therefore the
cure time is an important health care index. However, those proposed methods
assume the cure time to be infinity, thus it is inconvenient to make inference
on the cure time. In this dissertation, we define a more general concept of
statistical cure via conditional survival. Based on the newly defined
statistical cure, the cure time is well defined. We develop cure time model
methodologies and show a variety of properties through simulation. In data
analysis, cure times are estimated for 22 major cancers in Taiwan, we further
use colorectal cancer data as an example to conduct statistical inference via
cure time model with covariate sex, age group, and stage. This dissertation
provides a methodology to obtain cure time estimate, which can contribute to
public health policy making.
| stat |
Volatility Analysis with Realized GARCH-Ito Models | This paper introduces a unified approach for modeling high-frequency
financial data that can accommodate both the continuous-time jump-diffusion and
discrete-time realized GARCH model by embedding the discrete realized GARCH
structure in the continuous instantaneous volatility process. The key feature
of the proposed model is that the corresponding conditional daily integrated
volatility adopts an autoregressive structure where both integrated volatility
and jump variation serve as innovations. We name it as the realized GARCH-Ito
model. Given the autoregressive structure in the conditional daily integrated
volatility, we propose a quasi-likelihood function for parameter estimation and
establish its asymptotic properties. To improve the parameter estimation, we
propose a joint quasi-likelihood function that is built on the marriage of
daily integrated volatility estimated by high-frequency data and nonparametric
volatility estimator obtained from option data. We conduct a simulation study
to check the finite sample performance of the proposed methodologies and an
empirical study with the S&P500 stock index and option data.
| stat |
Experimental Evaluation of Individualized Treatment Rules | The increasing availability of individual-level data has led to numerous
applications of individualized (or personalized) treatment rules (ITRs). Policy
makers often wish to empirically evaluate ITRs and compare their relative
performance before implementing them in a target population. We propose a new
evaluation metric, the population average prescriptive effect (PAPE). The PAPE
compares the performance of ITR with that of non-individualized treatment rule,
which randomly treats the same proportion of units. Averaging the PAPE over a
range of budget constraints yields our second evaluation metric, the area under
the prescriptive effect curve (AUPEC). The AUPEC represents an overall
performance measure for evaluation, like the area under the receiver and
operating characteristic curve (AUROC) does for classification, and is a
generalization of the QINI coefficient utilized in uplift modeling. We use
Neyman's repeated sampling framework to estimate the PAPE and AUPEC and derive
their exact finite-sample variances based on random sampling of units and
random assignment of treatment. We extend our methodology to a common setting,
in which the same experimental data is used to both estimate and evaluate ITRs.
In this case, our variance calculation incorporates the additional uncertainty
due to random splits of data used for cross-validation. The proposed evaluation
metrics can be estimated without requiring modeling assumptions, asymptotic
approximation, or resampling methods. As a result, it is applicable to any ITR
including those based on complex machine learning algorithms. The open-source
software package is available for implementing the proposed methodology.
| stat |
SGA: A Robust Algorithm for Partial Recovery of Tree-Structured
Graphical Models with Noisy Samples | We consider learning Ising tree models when the observations from the nodes
are corrupted by independent but non-identically distributed noise with unknown
statistics. Katiyar et al. (2020) showed that although the exact tree structure
cannot be recovered, one can recover a partial tree structure; that is, a
structure belonging to the equivalence class containing the true tree. This
paper presents a systematic improvement of Katiyar et al. (2020). First, we
present a novel impossibility result by deriving a bound on the necessary
number of samples for partial recovery. Second, we derive a significantly
improved sample complexity result in which the dependence on the minimum
correlation $\rho_{\min}$ is $\rho_{\min}^{-8}$ instead of $\rho_{\min}^{-24}$.
Finally, we propose Symmetrized Geometric Averaging (SGA), a more statistically
robust algorithm for partial tree recovery. We provide error exponent analyses
and extensive numerical results on a variety of trees to show that the sample
complexity of SGA is significantly better than the algorithm of Katiyar et al.
(2020). SGA can be readily extended to Gaussian models and is shown via
numerical experiments to be similarly superior.
| stat |
Categorical Exploratory Data Analysis: From Multiclass Classification
and Response Manifold Analytics perspectives of baseball pitching dynamics | From two coupled Multiclass Classification (MCC) and Response Manifold
Analytics (RMA) perspectives, we develop Categorical Exploratory Data Analysis
(CEDA) on PITCHf/x database for the information content of Major League
Baseball's (MLB) pitching dynamics. MCC and RMA information contents are
represented by one collection of multi-scales pattern categories from mixing
geometries and one collection of global-to-local geometric localities from
response-covariate manifolds, respectively. These collectives shed light on the
pitching dynamics and maps out uncertainty of popular machine learning
approaches. On MCC setting, an indirect-distance-measure based label embedding
tree leads to discover asymmetry of mixing geometries among labels'
point-clouds. A selected chain of complementary covariate feature groups
collectively brings out multi-order mixing geometric pattern categories. Such
categories then reveal the true nature of MCC predictive inferences. On RMA
setting, multiple response features couple with multiple major covariate
features to demonstrate physical principles bearing manifolds with a lattice of
natural localities. With minor features' heterogeneous effects being locally
identified, such localities jointly weave their focal characteristics into
system understanding and provide a platform for RMA predictive inferences. Our
CEDA works for universal data types, adopts non-linear associations and
facilitates efficient feature-selections and inferences.
| stat |
Perfect reconstruction of sparse signals with piecewise continuous
nonconvex penalties and nonconvexity control | We consider compressed sensing formulated as a minimization problem of
nonconvex sparse penalties, Smoothly Clipped Absolute deviation (SCAD) and
Minimax Concave Penalty (MCP). The nonconvexity of these penalties is
controlled by nonconvexity parameters, and L1 penalty is contained as a limit
with respect to these parameters. The analytically derived reconstruction limit
overcomes that of L1 and the algorithmic limit in the Bayes-optimal setting,
when the nonconvexity parameters have suitable values. However, for small
nonconvexity parameters, where the reconstruction of the relatively dense
signals is theoretically guaranteed, the corresponding approximate message
passing (AMP) cannot achieve perfect reconstruction. We identify that the
shrinks in the basin of attraction to the perfect reconstruction causes the
discrepancy between the AMP and corresponding theory using state evolution. A
part of the discrepancy is resolved by introducing the control of the
nonconvexity parameters to guide the AMP trajectory to the basin of the
attraction.
| stat |
Generalized Multiple Importance Sampling | Importance Sampling methods are broadly used to approximate posterior
distributions or some of their moments. In its standard approach, samples are
drawn from a single proposal distribution and weighted properly. However, since
the performance depends on the mismatch between the targeted and the proposal
distributions, several proposal densities are often employed for the generation
of samples. Under this Multiple Importance Sampling (MIS) scenario, many works
have addressed the selection or adaptation of the proposal distributions,
interpreting the sampling and the weighting steps in different ways. In this
paper, we establish a general framework for sampling and weighing procedures
when more than one proposal are available. The most relevant MIS schemes in the
literature are encompassed within the new framework, and, moreover novel valid
schemes appear naturally. All the MIS schemes are compared and ranked in terms
of the variance of the associated estimators. Finally, we provide illustrative
examples which reveal that, even with a good choice of the proposal densities,
a careful interpretation of the sampling and weighting procedures can make a
significant difference in the performance of the method.
| stat |
Interpretable brain age prediction using linear latent variable models
of functional connectivity | Neuroimaging-driven prediction of brain age, defined as the predicted
biological age of a subject using only brain imaging data, is an exciting
avenue of research. In this work we seek to build models of brain age based on
functional connectivity while prioritizing model interpretability and
understanding. This way, the models serve to both provide accurate estimates of
brain age as well as allow us to investigate changes in functional connectivity
which occur during the ageing process. The methods proposed in this work
consist of a two-step procedure: first, linear latent variable models, such as
PCA and its extensions, are employed to learn reproducible functional
connectivity networks present across a cohort of subjects. The activity within
each network is subsequently employed as a feature in a linear regression model
to predict brain age. The proposed framework is employed on the data from the
CamCAN repository and the inferred brain age models are further demonstrated to
generalize using data from two open-access repositories: the Human Connectome
Project and the ATR Wide-Age-Range.
| stat |
Building high accuracy emulators for scientific simulations with deep
neural architecture search | Computer simulations are invaluable tools for scientific discovery. However,
accurate simulations are often slow to execute, which limits their
applicability to extensive parameter exploration, large-scale data analysis,
and uncertainty quantification. A promising route to accelerate simulations by
building fast emulators with machine learning requires large training datasets,
which can be prohibitively expensive to obtain with slow simulations. Here we
present a method based on neural architecture search to build accurate
emulators even with a limited number of training data. The method successfully
accelerates simulations by up to 2 billion times in 10 scientific cases
including astrophysics, climate science, biogeochemistry, high energy density
physics, fusion energy, and seismology, using the same super-architecture,
algorithm, and hyperparameters. Our approach also inherently provides emulator
uncertainty estimation, adding further confidence in their use. We anticipate
this work will accelerate research involving expensive simulations, allow more
extensive parameters exploration, and enable new, previously unfeasible
computational discovery.
| stat |
Fisher Auto-Encoders | It has been conjectured that the Fisher divergence is more robust to model
uncertainty than the conventional Kullback-Leibler (KL) divergence. This
motivates the design of a new class of robust generative auto-encoders (AE)
referred to as Fisher auto-encoders. Our approach is to design Fisher AEs by
minimizing the Fisher divergence between the intractable joint distribution of
observed data and latent variables, with that of the postulated/modeled joint
distribution. In contrast to KL-based variational AEs (VAEs), the Fisher AE can
exactly quantify the distance between the true and the model-based posterior
distributions. Qualitative and quantitative results are provided on both MNIST
and celebA datasets demonstrating the competitive performance of Fisher AEs in
terms of robustness compared to other AEs such as VAEs and Wasserstein AEs.
| stat |
A statistical test to reject the structural interpretation of a latent
factor model | Factor analysis is often used to assess whether a single univariate latent
variable is sufficient to explain most of the covariance among a set of
indicators for some underlying construct. When evidence suggests that a single
factor is adequate, research often proceeds by using a univariate summary of
the indicators in subsequent research. Implicit in such practices is the
assumption that it is the underlying latent, rather than the indicators, that
is causally efficacious. The assumption that the indicators do not have effects
on anything subsequent, and that they are themselves only affected by
antecedents through the underlying latent is a strong assumption, effectively
imposing a structural interpretation on the latent factor model. In this paper,
we show that this structural assumption has empirically testable implications,
even though the latent is unobserved. We develop a statistical test to
potentially reject the structural interpretation of a latent factor model. We
apply this test to data concerning associations between the
Satisfaction-with-Life-Scale and subsequent all-cause mortality, which provides
strong evidence against a structural interpretation for a univariate latent
underlying the scale. Discussion is given to the implications of this result
for the development, evaluation, and use of measures related to latent factor
models.
| stat |