title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Geographically Weighted Cox Regression for Prostate Cancer Survival Data
in Louisiana | The Cox proportional hazard model is one of the most popular tools in
analyzing time-to-event data in public health studies. When outcomes observed
in clinical data from different regions yield a varying pattern correlated with
location, it is often of great interest to investigate spatially varying
effects of covariates. In this paper, we propose a geographically weighted Cox
regression model for sparse spatial survival data. In addition, a stochastic
neighborhood weighting scheme is introduced at the county level. Theoretical
properties of the proposed geographically weighted estimators are examined in
detail. A model selection scheme based on the Takeuchi's model robust
information criteria (TIC) is discussed. Extensive simulation studies are
carried out to examine the empirical performance of the proposed methods. We
further apply the proposed methodology to analyze real data on prostate cancer
from the Surveillance, Epidemiology, and End Results cancer registry for the
state of Louisiana.
| stat |
Optimal Training of Fair Predictive Models | Recently there has been sustained interest in modifying prediction algorithms
to satisfy fairness constraints. These constraints are typically complex
nonlinear functionals of the observed data distribution. Focusing on the causal
constraints proposed by Nabi and Shpitser (2018), we introduce new theoretical
results and optimization techniques to make model training easier and more
accurate. Specifically, we show how to reparameterize the observed data
likelihood such that fairness constraints correspond directly to parameters
that appear in the likelihood, transforming a complex constrained optimization
objective into a simple optimization problem with box constraints. We also
exploit methods from empirical likelihood theory in statistics to improve
predictive performance, without requiring parametric models for
high-dimensional feature vectors.
| stat |
Generalized Poisson Difference Autoregressive Processes | This paper introduces a new stochastic process with values in the set Z of
integers with sign. The increments of process are Poisson differences and the
dynamics has an autoregressive structure. We study the properties of the
process and exploit the thinning representation to derive stationarity
conditions and the stationary distribution of the process. We provide a
Bayesian inference method and an efficient posterior approximation procedure
based on Monte Carlo. Numerical illustrations on both simulated and real data
show the effectiveness of the proposed inference.
| stat |
A Probabilistic Approach to Identifying Run Scoring Advantage in the
Order of Playing Cricket | In the game of cricket, the result of coin toss is assumed to be one of the
determinants of match outcome. The decision to bat first after winning the toss
is often taken to make the best use of superior pitch conditions and set a big
target for the opponent. However, the opponent may fail to show their natural
batting performance in the second innings due to a number of factors, including
deteriorated pitch conditions and excessive pressure of chasing a high target
score. The advantage of batting first has been highlighted in the literature
and expert opinions, however, the effect of batting and bowling order on match
outcome has not been investigated well enough to recommend a solution to any
potential bias. This study proposes a probability theory-based model to study
venue-specific scoring and chasing characteristics of teams under different
match outcomes. A total of 1117 one-day international matches held in ten
popular venues are analyzed to show substantially high scoring advantage and
likelihood when the winning team bat in the first innings. Results suggest that
the same 'bat-first' winning team is very unlikely to score or chase such a
high score if they were to bat in the second innings. Therefore, the coin toss
decision may favor one team over the other. A Bayesian model is proposed to
revise the target score for each venue such that the winning and scoring
likelihood is equal regardless of the toss decision. The data and source codes
have been shared publicly for future research in creating competitive match
outcomes by eliminating the advantage of batting order in run scoring.
| stat |
Regularization Matters: A Nonparametric Perspective on Overparametrized
Neural Network | Overparametrized neural networks trained by gradient descent (GD) can
provably overfit any training data. However, the generalization guarantee may
not hold for noisy data. From a nonparametric perspective, this paper studies
how well overparametrized neural networks can recover the true target function
in the presence of random noises. We establish a lower bound on the $L_2$
estimation error with respect to the GD iteration, which is away from zero
without a delicate choice of early stopping. In turn, through a comprehensive
analysis of $\ell_2$-regularized GD trajectories, we prove that for
overparametrized one-hidden-layer ReLU neural network with the $\ell_2$
regularization: (1) the output is close to that of the kernel ridge regression
with the corresponding neural tangent kernel; (2) minimax {optimal} rate of
$L_2$ estimation error is achieved. Numerical experiments confirm our theory
and further demonstrate that the $\ell_2$ regularization approach improves the
training robustness and works for a wider range of neural networks.
| stat |
Bayesian Neural Network Priors Revisited | Isotropic Gaussian priors are the de facto standard for modern Bayesian
neural network inference. However, such simplistic priors are unlikely to
either accurately reflect our true beliefs about the weight distributions, or
to give optimal performance. We study summary statistics of neural network
weights in different networks trained using SGD. We find that fully connected
networks (FCNNs) display heavy-tailed weight distributions, while convolutional
neural network (CNN) weights display strong spatial correlations. Building
these observations into the respective priors leads to improved performance on
a variety of image classification datasets. Moreover, we find that these priors
also mitigate the cold posterior effect in FCNNs, while in CNNs we see strong
improvements at all temperatures, and hence no reduction in the cold posterior
effect.
| stat |
Clustering of Big Data with Mixed Features | Clustering large, mixed data is a central problem in data mining. Many
approaches adopt the idea of k-means, and hence are sensitive to
initialisation, detect only spherical clusters, and require a priori the
unknown number of clusters. We here develop a new clustering algorithm for
large data of mixed type, aiming at improving the applicability and efficiency
of the peak-finding technique. The improvements are threefold: (1) the new
algorithm is applicable to mixed data; (2) the algorithm is capable of
detecting outliers and clusters of relatively lower density values; (3) the
algorithm is competent at deciding the correct number of clusters. The
computational complexity of the algorithm is greatly reduced by applying a fast
k-nearest neighbors method and by scaling down to component sets. We present
experimental results to verify that our algorithm works well in practice.
Keywords: Clustering; Big Data; Mixed Attribute; Density Peaks;
Nearest-Neighbor Graph; Conductance.
| stat |
Instance-dependent $\ell_\infty$-bounds for policy evaluation in tabular
reinforcement learning | Markov reward processes (MRPs) are used to model stochastic phenomena arising
in operations research, control engineering, robotics, and artificial
intelligence, as well as communication and transportation networks. In many of
these cases, such as in the policy evaluation problem encountered in
reinforcement learning, the goal is to estimate the long-term value function of
such a process without access to the underlying population transition and
reward functions. Working with samples generated under the synchronous model,
we study the problem of estimating the value function of an infinite-horizon,
discounted MRP on finitely many states in the $\ell_\infty$-norm. We analyze
both the standard plug-in approach to this problem and a more robust variant,
and establish non-asymptotic bounds that depend on the (unknown) problem
instance, as well as data-dependent bounds that can be evaluated based on the
observations of state-transitions and rewards. We show that these approaches
are minimax-optimal up to constant factors over natural sub-classes of MRPs.
Our analysis makes use of a leave-one-out decoupling argument tailored to the
policy evaluation problem, one which may be of independent interest.
| stat |
Centralizing-Unitizing Standardized High-Dimensional Directional
Statistics and Its Applications in Finance | Cross-sectional "Information Coefficient" (IC) is a widely and deeply
accepted measure in portfolio management. The paper gives an insight into IC in
view of high-dimensional directional statistics: IC is a linear operator on the
components of a centralizing-unitizing standardized random vector of
next-period cross-sectional returns. Our primary research first clearly defines
IC with the high-dimensional directional statistics, discussing its first two
moments. We derive the closed-form expressions of the directional statistics'
covariance matrix and IC's variance in a homoscedastic condition. Also, we
solve the optimization of IC's maximum expectation and minimum variance.
Simulation intuitively characterizes the standardized directional statistics
and IC's p.d.f.. The empirical analysis of the Chinese stock market uncovers
interesting facts about the standardized vectors of cross-sectional returns and
helps obtain the time series of the measure in the real market. The paper
discovers a potential application of directional statistics in finance, proves
explicit results of the projected normal distribution, and reveals IC's nature.
| stat |
A fresh look at introductory data science | The proliferation of vast quantities of available datasets that are large and
complex in nature has challenged universities to keep up with the demand for
graduates trained in both the statistical and the computational set of skills
required to effectively plan, acquire, manage, analyze, and communicate the
findings of such data. To keep up with this demand, attracting students early
on to data science as well as providing them a solid foray into the field
becomes increasingly important. We present a case study of an introductory
undergraduate course in data science that is designed to address these needs.
Offered at Duke University, this course has no pre-requisites and serves a wide
audience of aspiring statistics and data science majors as well as humanities,
social sciences, and natural sciences students. We discuss the unique set of
challenges posed by offering such a course and in light of these challenges, we
present a detailed discussion into the pedagogical design elements, content,
structure, computational infrastructure, and the assessment methodology of the
course. We also offer a repository containing all teaching materials that are
open-source, along with supplemental materials and the R code for reproducing
the figures found in the paper.
| stat |
Assumption-Lean Analysis of Cluster Randomized Trials in Infectious
Diseases for Intent-to-Treat Effects and Spillover Effects Among A Vulnerable
Subpopulation | Cluster randomized trials (CRTs) are a popular design to study the effect of
interventions in infectious disease settings. However, standard analysis of
CRTs primarily relies on strong parametric methods, usually a Normal mixed
effect models to account for the clustering structure, and focus on the overall
intent-to-treat (ITT) effect to evaluate effectiveness. The paper presents two
methods to analyze two types of effects in CRTs, the overall and heterogeneous
ITT effects and the spillover effect among never-takers who cannot or refuse to
take the intervention. For the ITT effects, we make a modest extension of an
existing method where we do not impose parametric models or asymptotic
restrictions on cluster size. For the spillover effect among never-takers, we
propose a new bound-based method that uses pre-treatment covariates,
classification algorithms, and a linear program to obtain sharp bounds. A key
feature of our method is that the bounds can become dramatically narrower as
the classification algorithm improves and the method may also be useful for
studies of partial identification with instrumental variables. We conclude by
reanalyzing a CRT studying the effect of face masks and hand sanitizers on
transmission of 2008 interpandemic influenza in Hong Kong.
| stat |
Where to find needles in a haystack? | In many existing methods in multiple comparison, one starts with either
Fisher's p-values or the local fdr scores. The former one, with a usual
definition as the tail probability exceeding the observed test statistic under
the null distribution, fails to use the information from the alternative
hypothesis and the targeted region of signals could be completely wrong
especially when the likelihood ratio function is not monotone. The local fdr
based approaches, usually relying on the density functions, are optimal
oracally. However, the targeted region of the signals of the data-driven
version is problematic because of the slow convergence of the non-parametric
density estimation especially on the boundaries. In this paper, we propose a
new method: Cdf and Local fdr Assisted multiple Testing method (CLAT), which is
optimal for cases when the p-values based method are not. Additionally, the
data-driven version only relies on the estimation of the cumulative
distribution function and converges to the oracle version quickly. Both
simulations and real data analysis demonstrate the superior performance of the
proposed method than the existing ones. Furthermore, the computation is
instantaneous based on a novel algorithm and is scalable to the large data set.
| stat |
Parameter elimination in particle Gibbs sampling | Bayesian inference in state-space models is challenging due to
high-dimensional state trajectories. A viable approach is particle Markov chain
Monte Carlo, combining MCMC and sequential Monte Carlo to form "exact
approximations" to otherwise intractable MCMC methods. The performance of the
approximation is limited to that of the exact method. We focus on particle
Gibbs and particle Gibbs with ancestor sampling, improving their performance
beyond that of the underlying Gibbs sampler (which they approximate) by
marginalizing out one or more parameters. This is possible when the parameter
prior is conjugate to the complete data likelihood. Marginalization yields a
non-Markovian model for inference, but we show that, in contrast to the general
case, this method still scales linearly in time. While marginalization can be
cumbersome to implement, recent advances in probabilistic programming have
enabled its automation. We demonstrate how the marginalized methods are viable
as efficient inference backends in probabilistic programming, and demonstrate
with examples in ecology and epidemiology.
| stat |
A robust nonlinear mixed-effects model for COVID-19 deaths data | The analysis of complex longitudinal data such as COVID-19 deaths is
challenging due to several inherent features: (i) Similarly-shaped profiles
with different decay patterns; (ii) Unexplained variation among repeated
measurements within each country, these repeated measurements may be viewed as
clustered data since they are taken on the same country at roughly the same
time; (iii) Skewness, outliers or skew-heavy-tailed noises are possibly
embodied within response variables. This article formulates a robust nonlinear
mixed-effects model based in the class of scale mixtures of skew-normal
distributions for modeling COVID-19 deaths, which allows the analysts to model
such data in the presence of the above described features simultaneously. An
efficient EM-type algorithm is proposed to carry out maximum likelihood
estimation of model parameters. The bootstrap method is used to determine
inherent characteristics of the nonlinear individual profiles such as
confidence interval of the predicted deaths and fitted curves. The target is to
model COVID-19 deaths curves from some Latin American countries since this
region is the new epicenter of the disease. Moreover, since a mixed-effect
framework borrows information from the population-average effects, in our
analysis we include some countries from Europe and North America that are in a
more advanced stage of their COVID-19 deaths curve.
| stat |
Transformed Central Quantile Subspace | Quantile regression (QR) is becoming increasingly popular due to its
relevance in many scientific investigations. However, application of QR can
become very challenging when dealing with high-dimensional data, making it
necessary to use dimension reduction techniques. Existing dimension reduction
techniques focus on the entire conditional distribution. We turn our attention
to dimension reduction techniques for conditional quantiles and introduce a
method that serves as an intermediate step between linear and nonlinear
dimension reduction. The idea is to apply existing linear dimension reduction
techniques on the transformed predictors. The proposed estimator, which is
shown to be root-n consistent, is demonstrated through simulation examples and
real data applications. Our results suggest that this method outperforms linear
dimension reduction for conditional quantiles.
| stat |
Estimation of Spatial Deformation for Nonstationary Processes via
Variogram Alignment | In modeling spatial processes, a second-order stationarity assumption is
often made. However, for spatial data observed on a vast domain, the covariance
function often varies over space, leading to a heterogeneous spatial dependence
structure, therefore requiring nonstationary modeling. Spatial deformation is
one of the main methods for modeling nonstationary processes, assuming the
nonstationary process has a stationary counterpart in the deformed space. The
estimation of the deformation function poses severe challenges. Here, we
introduce a novel approach for nonstationary geostatistical modeling, using
space deformation, when a single realization of the spatial process is
observed. Our method is based, at a fundamental level, on aligning regional
variograms, where warping variability of the distance from each subregion
explains the spatial nonstationarity. We propose to use multi-dimensional
scaling to map the warped distances to spatial locations. We asses the
performance of our new method using multiple simulation studies. Additionally,
we illustrate our methodology on precipitation data to estimate the
heterogeneous spatial dependence and to perform spatial predictions.
| stat |
A review and evaluation of secondary school accountability in England:
Statistical strengths, weaknesses, and challenges for 'Progress 8' | School performance measures are published annually in England to hold schools
to account and to support parental school choice. This article reviews and
evaluates the Progress 8 secondary school accountability system for
state-funded schools. We assess the statistical strengths and weaknesses of
Progress 8 relating to: choice of pupil outcome attainment measures; potential
adjustments for pupil input attainment and background characteristics;
decisions around which schools and pupils are excluded from the measure;
presentation of Progress 8 to users, choice of statistical model, and
calculation of statistical uncertainty; and issues related to the volatility of
school performance over time, including scope for reporting multi-year
averages. We then discuss challenges for Progress 8 raised by the COVID-19
pandemic. Six simple recommendations follow to improve Progress 8 and school
accountability in England.
| stat |
Granger Causality: A Review and Recent Advances | Introduced more than a half century ago, Granger causality has become a
popular tool for analyzing time series data in many application domains, from
economics and finance to genomics and neuroscience. Despite this popularity,
the validity of this notion for inferring causal relationships among time
series has remained the topic of continuous debate. Moreover, while the
original definition was general, limitations in computational tools have
primarily limited the applications of Granger causality to simple bivariate
vector auto-regressive processes or pairwise relationships among a set of
variables. Starting with a review of early developments and debates, this paper
discusses recent advances that address various shortcomings of the earlier
approaches, from models for high-dimensional time series to more recent
developments that account for nonlinear and non-Gaussian observations and allow
for sub-sampled and mixed frequency time series.
| stat |
Regularized ERM on random subspaces | We study a natural extension of classical empirical risk minimization, where
the hypothesis space is a random subspace of a given space. In particular, we
consider possibly data dependent subspaces spanned by a random subset of the
data, recovering as a special case Nystr\"om approaches for kernel methods.
Considering random subspaces naturally leads to computational savings, but the
question is whether the corresponding learning accuracy is degraded. These
statistical-computational tradeoffs have been recently explored for the least
squares loss and self-concordant loss functions, such as the logistic loss.
Here, we work to extend these results to convex Lipschitz loss functions, that
might not be smooth, such as the hinge loss used in support vector machines.
This extension requires developing new proofs, that use different technical
tools. Our main results show the existence of different settings, depending on
how hard the learning problem is, for which computational efficiency can be
improved with no loss in performance. Theoretical results are illustrated with
simple numerical experiments.
| stat |
Identify treatment effect patterns for personalised decisions | In personalised decision making, evidence is required to determine suitable
actions for individuals. Such evidence can be obtained by identifying treatment
effect heterogeneity in different subgroups of the population. In this paper,
we design a new type of pattern, treatment effect pattern to represent and
discover treatment effect heterogeneity from data for determining whether a
treatment will work for an individual or not. Our purpose is to use the
computational power to find the most specific and relevant conditions for
individuals with respect to a treatment or an action to assist with
personalised decision making. Most existing work on identifying treatment
effect heterogeneity takes a top-down or partitioning based approach to search
for subgroups with heterogeneous treatment effects. We propose a bottom-up
generalisation algorithm to obtain the most specific patterns that fit
individual circumstances the best for personalised decision making. For the
generalisation, we follow a consistency driven strategy to maintain inner-group
homogeneity and inter-group heterogeneity of treatment effects. We also employ
graphical causal modelling technique to identify adjustment variables for
reliable treatment effect pattern discovery. Our method can find the treatment
effect patterns reliably as validated by the experiments. The method is faster
than the two existing machine learning methods for heterogeneous treatment
effect identification and it produces subgroups with higher inner-group
treatment effect homogeneity.
| stat |
Deep Neural Networks as Point Estimates for Deep Gaussian Processes | Deep Gaussian processes (DGPs) have struggled for relevance in applications
due to the challenges and cost associated with Bayesian inference. In this
paper we propose a sparse variational approximation for DGPs for which the
approximate posterior mean has the same mathematical structure as a Deep Neural
Network (DNN). We make the forward pass through a DGP equivalent to a ReLU DNN
by finding an interdomain transformation that represents the GP posterior mean
as a sum of ReLU basis functions. This unification enables the initialisation
and training of the DGP as a neural network, leveraging the well established
practice in the deep learning community, and so greatly aiding the inference
task. The experiments demonstrate improved accuracy and faster training
compared to current DGP methods, while retaining favourable predictive
uncertainties.
| stat |
Geodesically parameterized covariance estimation | Statistical modeling of spatiotemporal phenomena often requires selecting a
covariance matrix from a covariance class. Yet standard parametric covariance
families can be insufficiently flexible for practical applications, while
non-parametric approaches may not easily allow certain kinds of prior knowledge
to be incorporated. We propose instead to build covariance families out of
geodesic curves. These covariances offer more flexibility for problem-specific
tailoring than classical parametric families, and are preferable to simple
convex combinations. Once the covariance family has been chosen, one typically
needs to select a representative member by solving an optimization problem,
e.g., by maximizing the likelihood of a data set. We consider instead a
differential geometric interpretation of this problem: minimizing the geodesic
distance to a sample covariance matrix ("natural projection"). Our approach is
consistent with the notion of distance employed to build the covariance family
and does not require assuming a particular probability distribution for the
data. We show that natural projection and maximum likelihood estimation within
the covariance family are locally equivalent up to second order. We also
demonstrate that natural projection may yield more accurate estimates with
noise-corrupted data.
| stat |
Variable screening based on Gaussian Centered L-moments | An important challenge in big data is identification of important variables.
In this paper, we propose methods of discovering variables with non-standard
univariate marginal distributions. The conventional moments-based summary
statistics can be well-adopted for that purpose, but their sensitivity to
outliers can lead to selection based on a few outliers rather than
distributional shape such as bimodality. To address this type of
non-robustness, we consider the L-moments. Using these in practice, however,
has a limitation because they do not take zero values at the Gaussian
distributions to which the shape of a marginal distribution is most naturally
compared. As a remedy, we propose Gaussian Centered L-moments which share
advantages of the L-moments but have zeros at the Gaussian distributions. The
strength of Gaussian Centered L-moments over other conventional moments is
shown in theoretical and practical aspects such as their performances in
screening important genes in cancer genetics data.
| stat |
Bayesian Context Trees: Modelling and exact inference for discrete time
series | We develop a new Bayesian modelling framework for the class of higher-order,
variable-memory Markov chains, and introduce an associated collection of
methodological tools for exact inference with discrete time series. We show
that a version of the context tree weighting algorithm can compute the prior
predictive likelihood exactly (averaged over both models and parameters), and
two related algorithms are introduced, which identify the a posteriori most
likely models and compute their exact posterior probabilities. All three
algorithms are deterministic and have linear-time complexity. A family of
variable-dimension Markov chain Monte Carlo samplers is also provided,
facilitating further exploration of the posterior. The performance of the
proposed methods in model selection, Markov order estimation and prediction is
illustrated through simulation experiments and real-world applications with
data from finance, genetics, neuroscience, and animal communication.
| stat |
Permutation test for the multivariate coefficient of variation in
factorial designs | New inference methods for the multivariate coefficient of variation and its
reciprocal, the standardized mean, are presented. While there are various
testing procedures for both parameters in the univariate case, it is less known
how to do inference in the multivariate setting appropriately. There are some
existing procedures but they rely on restrictive assumptions on the underlying
distributions. We tackle this problem by applying Wald-type statistics in the
context of general, potentially heteroscedastic factorial designs. In addition
to the $k$-sample case, higher-way layouts can be incorporated into this
framework allowing the discussion of main and interaction effects. The
resulting procedures are shown to be asymptotically valid under the null
hypothesis and consistent under general alternatives. To improve the finite
sample performance, we suggest permutation versions of the tests and shown that
the tests' asymptotic properties can be transferred to them. An exhaustive
simulation study compares the new tests, their permutation counterparts and
existing methods. To further analyse the differences between the tests, we
conduct two illustrative real data examples.
| stat |
Linking Bank Clients using Graph Neural Networks Powered by Rich
Transactional Data | Financial institutions obtain enormous amounts of data about user
transactions and money transfers, which can be considered as a large graph
dynamically changing in time. In this work, we focus on the task of predicting
new interactions in the network of bank clients and treat it as a link
prediction problem. We propose a new graph neural network model, which uses not
only the topological structure of the network but rich time-series data
available for the graph nodes and edges. We evaluate the developed method using
the data provided by a large European bank for several years. The proposed
model outperforms the existing approaches, including other neural network
models, with a significant gap in ROC AUC score on link prediction problem and
also allows to improve the quality of credit scoring.
| stat |
Metropolized Knockoff Sampling | Model-X knockoffs is a wrapper that transforms essentially any feature
importance measure into a variable selection algorithm, which discovers true
effects while rigorously controlling the expected fraction of false positives.
A frequently discussed challenge to apply this method is to construct knockoff
variables, which are synthetic variables obeying a crucial exchangeability
property with the explanatory variables under study. This paper introduces
techniques for knockoff generation in great generality: we provide a sequential
characterization of all possible knockoff distributions, which leads to a
Metropolis-Hastingsformulation of an exact knockoff sampler. We further show
how to use conditional independence structure to speed up computations.
Combining these two threads, we introduce an explicit set of sequential
algorithms and empirically demonstrate their effectiveness. Our theoretical
analysis proves that our algorithms achieve near-optimal computational
complexity in certain cases. The techniques we develop are sufficiently rich to
enable knockoff sampling in challenging models including cases where the
covariates are continuous and heavy-tailed, and follow a graphical model such
as the Ising model.
| stat |
Sparse Structures for Multivariate Extremes | Extreme value statistics provides accurate estimates for the small occurrence
probabilities of rare events. While theory and statistical tools for univariate
extremes are well-developed, methods for high-dimensional and complex data sets
are still scarce. Appropriate notions of sparsity and connections to other
fields such as machine learning, graphical models and high-dimensional
statistics have only recently been established. This article reviews the new
domain of research concerned with the detection and modeling of sparse patterns
in rare events. We first describe the different forms of extremal dependence
that can arise between the largest observations of a multivariate random
vector. We then discuss the current research topics including clustering,
principal component analysis and graphical modeling for extremes.
Identification of groups of variables which can be concomitantly extreme is
also addressed. The methods are illustrated with an application to flood risk
assessment.
| stat |
Testing Mediation Effects Using Logic of Boolean Matrices | Mediation analysis is becoming an increasingly important tool in scientific
studies. A central question in high-dimensional mediation analysis is to infer
the significance of individual mediators. The main challenge is the sheer
number of possible paths that go through all combinations of mediators. Most
existing mediation inference solutions either explicitly impose that the
mediators are conditionally independent given the exposure, or ignore any
potential directed paths among the mediators. In this article, we propose a
novel hypothesis testing procedure to evaluate individual mediation effects,
while taking into account potential interactions among the mediators. Our
proposal thus fills a crucial gap, and greatly extends the scope of existing
mediation tests. Our key idea is to construct the test statistic using the
logic of Boolean matrices, which enables us to establish the proper limiting
distribution under the null hypothesis. We further employ screening, data
splitting, and decorrelated estimation to reduce the bias and increase the
power of the test. We show our test can control both the size and false
discovery rate asymptotically, and the power of the test approaches one,
meanwhile allowing the number of mediators to diverge to infinity with the
sample size. We demonstrate the efficacy of our method through both simulations
and a neuroimaging study of Alzheimer's disease.
| stat |
Combining Non-probability and Probability Survey Samples Through Mass
Imputation | This paper presents theoretical results on combining non-probability and
probability survey samples through mass imputation, an approach originally
proposed by Rivers (2007) as sample matching without rigorous theoretical
justification. Under suitable regularity conditions, we establish the
consistency of the mass imputation estimator and derive its asymptotic variance
formula. Variance estimators are developed using either linearization or
bootstrap. Finite sample performances of the mass imputation estimator are
investigated through simulation studies and an application to analyzing a
non-probability sample collected by the Pew Research Centre.
| stat |
Bayesian Regression and Classification Using Gaussian Process Priors
Indexed by Probability Density Functions | In this paper, we introduce the notion of Gaussian processes indexed by
probability density functions for extending the Mat\'ern family of covariance
functions. We use some tools from information geometry to improve the
efficiency and the computational aspects of the Bayesian learning model. We
particularly show how a Bayesian inference with a Gaussian process prior
(covariance parameters estimation and prediction) can be put into action on the
space of probability density functions. Our framework has the capacity of
classifiying and infering on data observations that lie on nonlinear subspaces.
Extensive experiments on multiple synthetic, semi-synthetic and real data
demonstrate the effectiveness and the efficiency of the proposed methods in
comparison with current state-of-the-art methods.
| stat |
Rapid Numerical Approximation Method for Integrated Covariance Functions
Over Irregular Data Regions | In many practical applications, spatial data are often collected at areal
levels (i.e., block data) and the inferences and predictions about the variable
at points or blocks different from those at which it has been observed
typically depend on integrals of the underlying continuous spatial process. In
this paper we describe a method based on Fourier transform by which multiple
integrals of covariance functions over irregular data regions may be
numerically approximated with the same level of accuracy to traditional
methods, but at a greatly reduced computational expense.
| stat |
Incertitudes et mesures | Educational guide focused on the statistical treatment of measurement
uncertainties. The conditions of application of current practices are detailed
and precised: mean values, central limit theorem, linear regression. The last
two chapters are devoted to an introduction to the Bayesian inference and a
series of application cases: machine failure date, elimination of a background
noise, linear adjustment with elimination of outliers.
| stat |
Kernel Dependence Regularizers and Gaussian Processes with Applications
to Algorithmic Fairness | Current adoption of machine learning in industrial, societal and economical
activities has raised concerns about the fairness, equity and ethics of
automated decisions. Predictive models are often developed using biased
datasets and thus retain or even exacerbate biases in their decisions and
recommendations. Removing the sensitive covariates, such as gender or race, is
insufficient to remedy this issue since the biases may be retained due to other
related covariates. We present a regularization approach to this problem that
trades off predictive accuracy of the learned models (with respect to biased
labels) for the fairness in terms of statistical parity, i.e. independence of
the decisions from the sensitive covariates. In particular, we consider a
general framework of regularized empirical risk minimization over reproducing
kernel Hilbert spaces and impose an additional regularizer of dependence
between predictors and sensitive covariates using kernel-based measures of
dependence, namely the Hilbert-Schmidt Independence Criterion (HSIC) and its
normalized version. This approach leads to a closed-form solution in the case
of squared loss, i.e. ridge regression. Moreover, we show that the dependence
regularizer has an interpretation as modifying the corresponding Gaussian
process (GP) prior. As a consequence, a GP model with a prior that encourages
fairness to sensitive variables can be derived, allowing principled
hyperparameter selection and studying of the relative relevance of covariates
under fairness constraints. Experimental results in synthetic examples and in
real problems of income and crime prediction illustrate the potential of the
approach to improve fairness of automated decisions.
| stat |
Transfer Learning for High-dimensional Linear Regression: Prediction,
Estimation, and Minimax Optimality | This paper considers the estimation and prediction of a high-dimensional
linear regression in the setting of transfer learning, using samples from the
target model as well as auxiliary samples from different but possibly related
regression models. When the set of "informative" auxiliary samples is known, an
estimator and a predictor are proposed and their optimality is established. The
optimal rates of convergence for prediction and estimation are faster than the
corresponding rates without using the auxiliary samples. This implies that
knowledge from the informative auxiliary samples can be transferred to improve
the learning performance of the target problem. In the case that the set of
informative auxiliary samples is unknown, we propose a data-driven procedure
for transfer learning, called Trans-Lasso, and reveal its robustness to
non-informative auxiliary samples and its efficiency in knowledge transfer. The
proposed procedures are demonstrated in numerical studies and are applied to a
dataset concerning the associations among gene expressions. It is shown that
Trans-Lasso leads to improved performance in gene expression prediction in a
target tissue by incorporating the data from multiple different tissues as
auxiliary samples.
| stat |
The Nu Class of Low-Degree-Truncated, Rational, Generalized Functions.
I. IMSPE in Design of Computer Experiments: Integrals and Very-Low-N,
Single-Factor, Free-Ranging Designs | We provide detailed algebra for determining the integrated mean-squared
prediction error (IMSPE) of designs of computer experiments, with one factor
and one or two points, under the exponential, Gaussian, or either of two Matern
correlation functions. This algebra shall provide the basis for the
identification of the IMSPE as a member of a special class of
low-degree-truncated rational functions, which we name, here, the Nu class. We
shall detail this function class in a series of papers, of which this is the
first.
| stat |
Transformed Fay-Herriot Model with Measurement Error in Covariates | Statistical agencies are often asked to produce small area estimates (SAEs)
for positively skewed variables. When domain sample sizes are too small to
support direct estimators, effects of skewness of the response variable can be
large. As such, it is important to appropriately account for the distribution
of the response variable given available auxiliary information. Motivated by
this issue and in order to stabilize the skewness and achieve normality in the
response variable, we propose an area-level log-measurement error model on the
response variable. Then, under our proposed modeling framework, we derive an
empirical Bayes (EB) predictor of positive small area quantities subject to the
covariates containing measurement error. We propose a corresponding mean
squared prediction error (MSPE) of EB predictor using both a jackknife and a
bootstrap method. We show that the order of the bias is $O(m^{-1})$, where $m$
is the number of small areas. Finally, we investigate the performance of our
methodology using both design-based and model-based simulation studies.
| stat |
Particle filter with rejection control and unbiased estimator of the
marginal likelihood | We consider the combined use of resampling and partial rejection control in
sequential Monte Carlo methods, also known as particle filters. While the
variance reducing properties of rejection control are known, there has not been
(to the best of our knowledge) any work on unbiased estimation of the marginal
likelihood (also known as the model evidence or the normalizing constant) in
this type of particle filter. Being able to estimate the marginal likelihood
without bias is highly relevant for model comparison, computation of
interpretable and reliable confidence intervals, and in exact approximation
methods, such as particle Markov chain Monte Carlo. In the paper we present a
particle filter with rejection control that enables unbiased estimation of the
marginal likelihood.
| stat |
Actions Speak Louder Than Goals: Valuing Player Actions in Soccer | Assessing the impact of the individual actions performed by soccer players
during games is a crucial aspect of the player recruitment process.
Unfortunately, most traditional metrics fall short in addressing this task as
they either focus on rare actions like shots and goals alone or fail to account
for the context in which the actions occurred. This paper introduces (1) a new
language for describing individual player actions on the pitch and (2) a
framework for valuing any type of player action based on its impact on the game
outcome while accounting for the context in which the action happened. By
aggregating soccer players' action values, their total offensive and defensive
contributions to their team can be quantified. We show how our approach
considers relevant contextual information that traditional player evaluation
metrics ignore and present a number of use cases related to scouting and
playing style characterization in the 2016/2017 and 2017/2018 seasons in
Europe's top competitions.
| stat |
Any Target Function Exists in a Neighborhood of Any Sufficiently Wide
Random Network: A Geometrical Perspective | It is known that any target function is realized in a sufficiently small
neighborhood of any randomly connected deep network, provided the width (the
number of neurons in a layer) is sufficiently large. There are sophisticated
theories and discussions concerning this striking fact, but rigorous theories
are very complicated. We give an elementary geometrical proof by using a simple
model for the purpose of elucidating its structure. We show that
high-dimensional geometry plays a magical role: When we project a
high-dimensional sphere of radius 1 to a low-dimensional subspace, the uniform
distribution over the sphere reduces to a Gaussian distribution of negligibly
small covariances.
| stat |
Copula-based functional Bayes classification with principal components
and partial least squares | We present a new functional Bayes classifier that uses principal component
(PC) or partial least squares (PLS) scores from the common covariance function,
that is, the covariance function marginalized over groups. When the groups have
different covariance functions, the PC or PLS scores need not be independent or
even uncorrelated. We use copulas to model the dependence. Our method is
semiparametric; the marginal densities are estimated nonparametrically by
kernel smoothing and the copula is modeled parametrically. We focus on Gaussian
and t-copulas, but other copulas could be used. The strong performance of our
methodology is demonstrated through simulation, real data examples, and
asymptotic properties.
| stat |
Fast Adaptation with Linearized Neural Networks | The inductive biases of trained neural networks are difficult to understand
and, consequently, to adapt to new settings. We study the inductive biases of
linearizations of neural networks, which we show to be surprisingly good
summaries of the full network functions. Inspired by this finding, we propose a
technique for embedding these inductive biases into Gaussian processes through
a kernel designed from the Jacobian of the network. In this setting, domain
adaptation takes the form of interpretable posterior inference, with
accompanying uncertainty estimation. This inference is analytic and free of
local optima issues found in standard techniques such as fine-tuning neural
network weights to a new task. We develop significant computational speed-ups
based on matrix multiplies, including a novel implementation for scalable
Fisher vector products. Our experiments on both image classification and
regression demonstrate the promise and convenience of this framework for
transfer learning, compared to neural network fine-tuning. Code is available at
https://github.com/amzn/xfer/tree/master/finite_ntk.
| stat |
Graphical model inference: Sequential Monte Carlo meets deterministic
approximations | Approximate inference in probabilistic graphical models (PGMs) can be grouped
into deterministic methods and Monte-Carlo-based methods. The former can often
provide accurate and rapid inferences, but are typically associated with biases
that are hard to quantify. The latter enjoy asymptotic consistency, but can
suffer from high computational costs. In this paper we present a way of
bridging the gap between deterministic and stochastic inference. Specifically,
we suggest an efficient sequential Monte Carlo (SMC) algorithm for PGMs which
can leverage the output from deterministic inference methods. While generally
applicable, we show explicitly how this can be done with loopy belief
propagation, expectation propagation, and Laplace approximations. The resulting
algorithm can be viewed as a post-correction of the biases associated with
these methods and, indeed, numerical results show clear improvements over the
baseline deterministic methods as well as over "plain" SMC.
| stat |
Qini-based Uplift Regression | Uplift models provide a solution to the problem of isolating the marketing
effect of a campaign. For customer churn reduction, uplift models are used to
identify the customers who are likely to respond positively to a retention
activity only if targeted, and to avoid wasting resources on customers that are
very likely to switch to another company. We introduce a Qini-based uplift
regression model to analyze a large insurance company's retention marketing
campaign. Our approach is based on logistic regression models. We show that a
Qini-optimized uplift model acts as a regularizing factor for uplift, much as a
penalized likelihood model does for regression. This results in interpretable
parsimonious models with few relevant xplanatory variables. Our results show
that performing Qini-based parameters estimation significantly improves the
uplift models performance.
| stat |
Trading off Accuracy for Speedup: Multiplier Bootstraps for Subgraph
Counts | We propose a new class of multiplier bootstraps for count functionals. We
consider bootstrap procedures with linear and quadratic weights. These
correspond to the first and second-order terms of the Hoeffding decomposition
of the bootstrapped statistic arising from the multiplier bootstrap,
respectively. We show that the quadratic bootstrap procedure achieves
higher-order correctness for appropriately sparse graphs. The linear bootstrap
procedure requires fewer estimated network statistics, leading to improved
accuracy over its higher-order correct counterpart in sparser regimes. To
improve the computational properties of the linear bootstrap further, we
consider fast sketching methods to conduct approximate subgraph counting and
establish consistency of the resulting bootstrap procedure. We complement our
theoretical results with a simulation study and real data analysis and verify
that our procedure offers state-of-the-art performance for several functionals.
| stat |
Semi-Structured Deep Distributional Regression: Combining Structured
Additive Models and Deep Learning | Combining additive models and neural networks allows to broaden the scope of
statistical regression and extends deep learning-based approaches by
interpretable structured additive predictors at the same time. Existing
approaches uniting the two modeling approaches are, however, limited to very
specific combinations and, more importantly, involve an identifiability issue.
As a consequence, interpretability and stable estimation is typically lost. We
propose a general framework to combine structured regression models and deep
neural networks into a unifying network architecture. To overcome the inherent
identifiability issues between different model parts, we construct an
orthogonalization cell that projects the deep neural network into the
orthogonal complement of the statistical model predictor. This enables proper
estimation of structured model parts and thereby interpretability. We
demonstrate the framework's efficacy in numerical experiments and illustrate
its special merits in benchmarks and real-world applications.
| stat |
Automatic Relevance Determination Bayesian Neural Networks for Credit
Card Default Modelling | Credit risk modelling is an integral part of the global financial system.
While there has been great attention paid to neural network models for credit
default prediction, such models often lack the required interpretation
mechanisms and measures of the uncertainty around their predictions. This work
develops and compares Bayesian Neural Networks(BNNs) for credit card default
modelling. This includes a BNNs trained by Gaussian approximation and the first
implementation of BNNs trained by Hybrid Monte Carlo(HMC) in credit risk
modelling. The results on the Taiwan Credit Dataset show that BNNs with
Automatic Relevance Determination(ARD) outperform normal BNNs without ARD. The
results also show that BNNs trained by Gaussian approximation display similar
predictive performance to those trained by the HMC. The results further show
that BNN with ARD can be used to draw inferences about the relative importance
of different features thus critically aiding decision makers in explaining
model output to consumers. The robustness of this result is reinforced by high
levels of congruence between the features identified as important using the two
different approaches for training BNNs.
| stat |
Sensor-based localization of epidemic sources on human mobility networks | We investigate the source detection problem in epidemiology, which is one of
the most important issues for control of epidemics. Mathematically, we
reformulate the problem as one of identifying the relevant component in a
multivariate Gaussian mixture model. Focusing on the study of cholera and
diseases with similar modes of transmission, we calibrate the parameters of our
mixture model using human mobility networks within a stochastic, spatially
explicit epidemiological model for waterborne disease. Furthermore, we adopt a
Bayesian perspective, so that prior information on source location can be
incorporated (e.g., reflecting the impact of local conditions). Posterior-based
inference is performed, which permits estimates in the form of either
individual locations or regions. Importantly, our estimator only requires
first-arrival times of the epidemic by putative observers, typically located
only at a small proportion of nodes. The proposed method is demonstrated within
the context of the 2000-2002 cholera outbreak in the KwaZulu-Natal province of
South Africa.
| stat |
An Automatic Relevance Determination Prior Bayesian Neural Network for
Controlled Variable Selection | We present an Automatic Relevance Determination prior Bayesian Neural
Network(BNN-ARD) weight l2-norm measure as a feature importance statistic for
the model-x knockoff filter. We show on both simulated data and the Norwegian
wind farm dataset that the proposed feature importance statistic yields
statistically significant improvements relative to similar feature importance
measures in both variable selection power and predictive performance on a real
world dataset.
| stat |
Comparative Analysis of Machine Learning Approaches to Analyze and
Predict the Covid-19 Outbreak | Background. Forecasting the time of forthcoming pandemic reduces the impact
of diseases by taking precautionary steps such as public health messaging and
raising the consciousness of doctors. With the continuous and rapid increase in
the cumulative incidence of COVID-19, statistical and outbreak prediction
models including various machine learning (ML) models are being used by the
research community to track and predict the trend of the epidemic, and also in
developing appropriate strategies to combat and manage its spread. Methods. In
this paper, we present a comparative analysis of various ML approaches
including Support Vector Machine, Random Forest, K-Nearest Neighbor and
Artificial Neural Network in predicting the COVID-19 outbreak in the
epidemiological domain. We first apply the autoregressive distributed lag
(ARDL) method to identify and model the short and long-run relationships of the
time-series COVID-19 datasets. That is, we determine the lags between a
response variable and its respective explanatory time series variables as
independent variables. Then, the resulting significant variables concerning
their lags are used in the regression model selected by the ARDL for predicting
and forecasting the trend of the epidemic. Results. Statistical measures i.e.,
Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Mean Absolute
Percentage Error (MAPE) are used for model accuracy. The values of MAPE for the
best selected models for confirmed, recovered and deaths cases are 0.407, 0.094
and 0.124 respectively, which falls under the category of highly accurate
forecasts. In addition, we computed fifteen days ahead forecast for the daily
deaths, recover, and confirm patients and the cases fluctuated across time in
all aspects. Besides, the results reveal the advantages of ML algorithms for
supporting decision making of evolving short term policies.
| stat |
Group Testing for COVID-19: How to Stop Worrying and Test More | The corona virus disease 2019 (COVID-19) caused by the novel corona virus has
an exponential rate of infection. COVID-19 is particularly notorious as the
onset of symptoms in infected patients are usually delayed and there exists a
large number of asymptomatic carriers. In order to prevent overwhelming of
medical facilities and large fatality rate, early stage testing and diagnosis
are key requirements. In this article, we discuss the methodologies from the
group testing literature and its relevance to COVID-19 diagnosis. Specifically,
we investigate the efficiency of group testing using polymerase chain reaction
(PCR) for COVID-19. Group testing is a method in which multiple samples are
pooled together in groups and fewer tests are performed on these groups to
discern all the infected samples. We study the effect of dilution due to
pooling in group testing and show that group tests can perform well even in the
presence of dilution effects. We present multiple group testing algorithms that
could reduce the number of tests performed for COVID-19 diagnosis. We analyze
the efficiency of these tests and provide insights on their practical
relevance. With the use of algorithms described here, test plans can be
developed that can enable testing centers to increase the number of diagnosis
performed without increasing the number of PCR tests. The codes for generating
test plans are available online at [1].
| stat |
Clinical trials impacted by the COVID-19 pandemic: Adaptive designs to
the rescue? | Very recently the new pathogen severe acute respiratory syndrome coronavirus
2 (SARS-CoV-2) was identified and the coronavirus disease 2019 (COVID-19)
declared a pandemic by the World Health Organization. The pandemic has a number
of consequences for the ongoing clinical trials in non-COVID-19 conditions.
Motivated by four currently ongoing clinical trials in a variety of disease
areas we illustrate the challenges faced by the pandemic and sketch out
possible solutions including adaptive designs. Guidance is provided on (i)
where blinded adaptations can help; (ii) how to achieve type I error rate
control, if required; (iii) how to deal with potential treatment effect
heterogeneity; (iv) how to utilize early readouts; and (v) how to utilize
Bayesian techniques. In more detail approaches to resizing a trial affected by
the pandemic are developed including considerations to stop a trial early, the
use of group-sequential designs or sample size adjustment. All methods
considered are implemented in a freely available R shiny app. Furthermore,
regulatory and operational issues including the role of data monitoring
committees are discussed.
| stat |
Independence Testing for Multivariate Time Series | Complex data structures such as time series are increasingly present in
modern data science problems. A fundamental question is whether two such
time-series are statistically dependent. Many current approaches make
parametric assumptions on the random processes, only detect linear association,
require multiple tests, or forfeit power in high-dimensional, nonlinear
settings. Estimating the distribution of any test statistic under the null is
non-trivial, as the permutation test is invalid. This work juxtaposes distance
correlation (Dcorr) and multiscale graph correlation (MGC) from independence
testing literature and block permutation from time series analysis to address
these challenges. The proposed nonparametric procedure is valid and consistent,
building upon prior work by characterizing the geometry of the relationship,
estimating the time lag at which dependence is maximized, avoiding the need for
multiple testing, and exhibiting superior power in high-dimensional, low sample
size, nonlinear settings. Neural connectivity is analyzed via fMRI data,
revealing linear dependence of signals within the visual network and default
mode network, and nonlinear relationships in other networks. This work uncovers
a first-resort data analysis tool with open-source code available, directly
impacting a wide range of scientific disciplines.
| stat |
Forecasting functional time series using weighted likelihood methodology | Functional time series whose sample elements are recorded sequentially over
time are frequently encountered with increasing technology. Recent studies have
shown that analyzing and forecasting of functional time series can be performed
easily using functional principal component analysis and existing
univariate/multivariate time series models. However, the forecasting
performance of such functional time series models may be affected by the
presence of outlying observations which are very common in many scientific
fields. Outliers may distort the functional time series model structure, and
thus, the underlying model may produce high forecast errors. We introduce a
robust forecasting technique based on weighted likelihood methodology to obtain
point and interval forecasts in functional time series in the presence of
outliers. The finite sample performance of the proposed method is illustrated
by Monte Carlo simulations and four real-data examples. Numerical results
reveal that the proposed method exhibits superior performance compared with the
existing method(s).
| stat |
Extraction Urban Clusters from Geospatial Data: A Case Study from
Switzerland | Different techniques were developed to extract urban agglomerations from a
big dataset. The urban agglomerations are used to understand the structure and
growth of cities. However, the major challenge is to extract urban
agglomerations from big data, which can reflect human activities. Community
urban cluster refers to spatially clustered geographic events, such as human
settlements or activities. It provides a powerful and innovative insight to
analyze the structure and growth of the real city. In order to understand the
shape and growth of urban agglomerations in Switzerland from spatial and
temporal aspects, this work identifies urban clusters from nighttime light data
and street network data. Nighttime light data record lights emitted from human
settlements at night on the earth's surface. This work uses DMSP-OLS Nighttime
light data to extract urban clusters from 1992 to 2013. The street is one of
the most important factors to reflect human activities. Hence, urban clusters
are also extracted from street network data to understand the structure of
cities. Both of these data have a heavy-tailed distribution, which includes
power laws as well as lognormal and exponential distributions. The head/tail
breaks is a classification method to find the hierarchy of data with a
heavy-tailed distribution. This work uses head/tail breaks classification to
extract urban clusters of Switzerland. At last, the power law distribution of
all the urban clusters was detected at the country level.
| stat |
NLS: an accurate and yet easy-to-interpret regression method | An important feature of successful supervised machine learning applications
is to be able to explain the predictions given by the regression or
classification model being used. However, most state-of-the-art models that
have good predictive power lead to predictions that are hard to interpret.
Thus, several model-agnostic interpreters have been developed recently as a way
of explaining black-box classifiers. In practice, using these methods is a slow
process because a novel fitting is required for each new testing instance, and
several non-trivial choices must be made. We develop NLS (neural local
smoother), a method that is complex enough to give good predictions, and yet
gives solutions that are easy to be interpreted without the need of using a
separate interpreter. The key idea is to use a neural network that imposes a
local linear shape to the output layer. We show that NLS leads to predictive
power that is comparable to state-of-the-art machine learning models, and yet
is easier to interpret.
| stat |
Multi-Level Micro-Randomized Trial: Detecting the Proximal Effect of
Messages on Physical Activity | Technological advancements in mobile devices have made it possible to deliver
mobile health interventions to individuals. A novel intervention framework that
emerges from such advancements is the just-in-time adaptive intervention
(JITAI), where it aims to suggest the right support to the individual "just in
time", when their needs arise, thus having proximal, near future effects. The
micro-randomized trial (MRT) design was proposed recently to test the proximal
effects of these JITAIs. In an MRT, participants are repeatedly randomized to
one of the intervention options of various in the intervention components, at a
scale of hundreds or thousands of decision time points over the course of the
study. However, the extant MRT framework only tests the proximal effect of
two-level intervention components (e.g. control vs intervention). In this
paper, we propose a novel version of MRT design with multiple levels per
intervention component, which we call "multi-level micro-randomized trial"
(MLMRT) design. The MLMRT extends the existing MRT design by allowing
multi-level intervention components, and the addition of more levels to the
components during the study period. We apply generalized estimating equation
type methodology on the longitudinal data arising from an MLMRT to develop the
novel test statistics for assessing the proximal effects and deriving the
associated sample size calculators. We conduct simulation studies to evaluate
the sample size calculators based on both power and precision. We have
developed an R shiny application of the sample size calculators. This proposed
design is motivated by our involvement in the Diabetes and Mental Health
Adaptive Notification Tracking and Evaluation (DIAMANTE) study. This study uses
a novel mobile application, also called "DIAMANTE", which delivers adaptive
text messages to encourage physical activity.
| stat |
Feature quantization for parsimonious and interpretable predictive
models | For regulatory and interpretability reasons, logistic regression is still
widely used. To improve prediction accuracy and interpretability, a
preprocessing step quantizing both continuous and categorical data is usually
performed: continuous features are discretized and, if numerous, levels of
categorical features are grouped. An even better predictive accuracy can be
reached by embedding this quantization estimation step directly into the
predictive estimation step itself. But doing so, the predictive loss has to be
optimized on a huge set. To overcome this difficulty, we introduce a specific
two-step optimization strategy: first, the optimization problem is relaxed by
approximating discontinuous quantization functions by smooth functions; second,
the resulting relaxed optimization problem is solved via a particular neural
network. The good performances of this approach, which we call glmdisc, are
illustrated on simulated and real data from the UCI library and Cr\'edit
Agricole Consumer Finance (a major European historic player in the consumer
credit market).
| stat |
An Analysis of LIME for Text Data | Text data are increasingly handled in an automated fashion by machine
learning algorithms. But the models handling these data are not always
well-understood due to their complexity and are more and more often referred to
as "black-boxes." Interpretability methods aim to explain how these models
operate. Among them, LIME has become one of the most popular in recent years.
However, it comes without theoretical guarantees: even for simple models, we
are not sure that LIME behaves accurately. In this paper, we provide a first
theoretical analysis of LIME for text data. As a consequence of our theoretical
findings, we show that LIME indeed provides meaningful explanations for simple
models, namely decision trees and linear models.
| stat |
Adjusting for Spatial Effects in Genomic Prediction | This paper investigates the problem of adjusting for spatial effects in
genomic prediction. Despite being seldomly considered in genomic prediction,
spatial effects often affect phenotypic measurements of plants. We consider a
Gaussian random field model with an additive covariance structure that
incorporates genotype effects, spatial effects and subpopulation effects. An
empirical study shows the existence of spatial effects and heterogeneity across
different subpopulation families, while simulations illustrate the improvement
in selecting genotypically superior plants by adjusting for spatial effects in
genomic prediction.
| stat |
Model-based Computed Tomography Image Estimation: Partitioning Approach | There is a growing interest to get a fully MR based radiotherapy. The most
important development needed is to obtain improved bone tissue estimation. The
existing model-based methods perform poorly on bone tissues. This paper was
aimed at obtaining improved bone tissue estimation. Skew Gaussian mixture model
and Gaussian mixture model were proposed to investigate CT image estimation
from MR images by partitioning the data into two major tissue types. The
performance of the proposed models was evaluated using leave-one-out
cross-validation method on real data. In comparison with the existing
model-based approaches, the model-based partitioning approach outperformed in
bone tissue estimation, especially in dense bone tissue estimation.
| stat |
Modeling long-term capacity degradation of lithium-ion batteries | Capacity degradation of lithium-ion batteries under long-term cyclic aging is
modelled via a flexible sigmoidal-type regression set-up, where the regression
parameters can be interpreted. Different approaches known from the literature
are discussed and compared with the new proposal. Statistical procedures, such
as parameter estimation, confidence and prediction intervals are presented and
applied to real data. The long-term capacity degradation model may be applied
in second-life scenarios of batteries. Using some prior information or training
data on the complete degradation path, the model can be fitted satisfactorily
even if only short-term degradation data is available. The training data may
arise from a single battery.
| stat |
High-Dimensional Regression with Binary Coefficients. Estimating Squared
Error and a Phase Transition | We consider a sparse linear regression model Y=X\beta^{*}+W where X has a
Gaussian entries, W is the noise vector with mean zero Gaussian entries, and
\beta^{*} is a binary vector with support size (sparsity) k. Using a novel
conditional second moment method we obtain a tight up to a multiplicative
constant approximation of the optimal squared error
\min_{\beta}\|Y-X\beta\|_{2}, where the minimization is over all k-sparse
binary vectors \beta. The approximation reveals interesting structural
properties of the underlying regression problem. In particular, a) We establish
that n^*=2k\log p/\log (2k/\sigma^{2}+1) is a phase transition point with the
following "all-or-nothing" property. When n exceeds n^{*},
(2k)^{-1}\|\beta_{2}-\beta^*\|_0\approx 0, and when n is below n^{*},
(2k)^{-1}\|\beta_{2}-\beta^*\|_0\approx 1, where \beta_2 is the optimal
solution achieving the smallest squared error. With this we prove that n^{*} is
the asymptotic threshold for recovering \beta^* information theoretically. b)
We compute the squared error for an intermediate problem
\min_{\beta}\|Y-X\beta\|_{2} where minimization is restricted to vectors \beta
with \|\beta-\beta^{*}\|_0=2k \zeta, for \zeta\in [0,1]. We show that a lower
bound part \Gamma(\zeta) of the estimate, which corresponds to the estimate
based on the first moment method, undergoes a phase transition at three
different thresholds, namely n_{\text{inf,1}}=\sigma^2\log p, which is
information theoretic bound for recovering \beta^* when k=1 and \sigma is
large, then at n^{*} and finally at n_{\text{LASSO/CS}}. c) We establish a
certain Overlap Gap Property (OGP) on the space of all binary vectors \beta
when n\le ck\log p for sufficiently small constant c. We conjecture that OGP is
the source of algorithmic hardness of solving the minimization problem
\min_{\beta}\|Y-X\beta\|_{2} in the regime n<n_{\text{LASSO/CS}}.
| stat |
Transfer Learning Robustness in Multi-Class Categorization by
Fine-Tuning Pre-Trained Contextualized Language Models | This study compares the effectiveness and robustness of multi-class
categorization of Amazon product data using transfer learning on pre-trained
contextualized language models. Specifically, we fine-tuned BERT and XLNet, two
bidirectional models that have achieved state-of-the-art performance on many
natural language tasks and benchmarks, including text classification. While
existing classification studies and benchmarks focus on binary targets, with
the exception of ordinal ranking tasks, here we examine the robustness of such
models as the number of classes grows from 1 to 20. Our experiments demonstrate
an approximately linear decrease in performance metrics (i.e., precision,
recall, $F_1$ score, and accuracy) with the number of class labels. BERT
consistently outperforms XLNet using identical hyperparameters on the entire
range of class label quantities for categorizing products based on their
textual descriptions. BERT is also more affordable than XLNet in terms of the
computational cost (i.e., time and memory) required for training. In all cases
studied, the performance degradation rates were estimated to be 1% per
additional class label.
| stat |
Understanding and adjusting the selection bias from a proof-of-concept
study to a more confirmatory study | It has long been noticed that the efficacy observed in small early phase
studies is generally better than that observed in later larger studies.
Historically, the inflation of the efficacy results from early proof-of-concept
studies is either ignored, or adjusted empirically using a frequentist or
Bayesian approach. In this article, we systematically explained the underlying
reason for the inflation of efficacy results in small early phase studies from
the perspectives of measurement error models and selection bias. A systematic
method was built to adjust the early phase study results from both frequentist
and Bayesian perspectives. A hierarchical model was proposed to estimate the
distribution of the efficacy for a portfolio of compounds, which can serve as
the prior distribution for the Bayesian approach. We showed through theory that
the systematic adjustment provides an unbiased estimator for the true mean
efficacy for a portfolio of compounds. The adjustment was applied to paired
data for the efficacy in early small and later larger studies for a set of
compounds in diabetes and immunology. After the adjustment, the bias in the
early phase small studies seems to be diminished.
| stat |
A coherent likelihood parametrization for doubly robust estimation of a
causal effect with missing confounders | Missing data and confounding are two problems researchers face in
observational studies for comparative effectiveness. Williamson et al. (2012)
recently proposed a unified approach to handle both issues concurrently using a
multiply-robust (MR) methodology under the assumption that confounders are
missing at random. Their approach considers a union of models in which any
submodel has a parametric component while the remaining models are
unrestricted. We show that while their estimating function is MR in theory, the
possibility for multiply robust inference is complicated by the fact that
parametric models for different components of the union model are not variation
independent and therefore the MR property is unlikely to hold in practice. To
address this, we propose an alternative transparent parametrization of the
likelihood function, which makes explicit the model dependencies between
various nuisance functions needed to evaluate the MR efficient score. The
proposed method is genuinely doubly-robust (DR) in that it is consistent and
asymptotic normal if one of two sets of modeling assumptions holds. We evaluate
the performance and doubly robust property of the DR method via a simulation
study.
| stat |
Estimating Atmospheric Motion Winds from Satellite Image Data using
Space-time Drift Models | Geostationary satellites collect high-resolution weather data comprising a
series of images which can be used to estimate wind speed and direction at
different altitudes. The Derived Motion Winds (DMW) Algorithm is commonly used
to process these data and estimate atmospheric winds by tracking features in
images taken by the GOES-R series of the NOAA geostationary meteorological
satellites. However, the wind estimates from the DMW Algorithm are sparse and
do not come with uncertainty measures. This motivates us to statistically model
wind motions as a spatial process drifting in time. We propose a covariance
function that depends on spatial and temporal lags and a drift parameter to
capture the wind speed and wind direction. We estimate the parameters by local
maximum likelihood. Our method allows us to compute standard errors of the
estimates, enabling spatial smoothing of the estimates using a Gaussian kernel
weighted by the inverses of the estimated variances. We conduct extensive
simulation studies to determine the situations where our method performs well.
The proposed method is applied to the GOES-15 brightness temperature data over
Colorado and reduces prediction error of brightness temperature compared to the
DMW Algorithm.
| stat |
Nudging the particle filter | We investigate a new sampling scheme aimed at improving the performance of
particle filters whenever (a) there is a significant mismatch between the
assumed model dynamics and the actual system, or (b) the posterior probability
tends to concentrate in relatively small regions of the state space. The
proposed scheme pushes some particles towards specific regions where the
likelihood is expected to be high, an operation known as nudging in the
geophysics literature. We re-interpret nudging in a form applicable to any
particle filtering scheme, as it does not involve any changes in the rest of
the algorithm. Since the particles are modified, but the importance weights do
not account for this modification, the use of nudging leads to additional bias
in the resulting estimators. However, we prove analytically that nudged
particle filters can still attain asymptotic convergence with the same error
rates as conventional particle methods. Simple analysis also yields an
alternative interpretation of the nudging operation that explains its
robustness to model errors. Finally, we show numerical results that illustrate
the improvements that can be attained using the proposed scheme. In particular,
we present nonlinear tracking examples with synthetic data and a model
inference example using real-world financial data.
| stat |
Performance Metric Elicitation from Pairwise Classifier Comparisons | Given a binary prediction problem, which performance metric should the
classifier optimize? We address this question by formalizing the problem of
Metric Elicitation. The goal of metric elicitation is to discover the
performance metric of a practitioner, which reflects her innate rewards (costs)
for correct (incorrect) classification. In particular, we focus on eliciting
binary classification performance metrics from pairwise feedback, where a
practitioner is queried to provide relative preference between two classifiers.
By exploiting key geometric properties of the space of confusion matrices, we
obtain provably query efficient algorithms for eliciting linear and
linear-fractional performance metrics. We further show that our method is
robust to feedback and finite sample noise.
| stat |
Multinomial Logit Contextual Bandits: Provable Optimality and
Practicality | We consider a sequential assortment selection problem where the user choice
is given by a multinomial logit (MNL) choice model whose parameters are
unknown. In each period, the learning agent observes a $d$-dimensional
contextual information about the user and the $N$ available items, and offers
an assortment of size $K$ to the user, and observes the bandit feedback of the
item chosen from the assortment. We propose upper confidence bound based
algorithms for this MNL contextual bandit. The first algorithm is a simple and
practical method which achieves an $\tilde{\mathcal{O}}(d\sqrt{T})$ regret over
$T$ rounds. Next, we propose a second algorithm which achieves a
$\tilde{\mathcal{O}}(\sqrt{dT})$ regret. This matches the lower bound for the
MNL bandit problem, up to logarithmic terms, and improves on the best known
result by a $\sqrt{d}$ factor. To establish this sharper regret bound, we
present a non-asymptotic confidence bound for the maximum likelihood estimator
of the MNL model that may be of independent interest as its own theoretical
contribution. We then revisit the simpler, significantly more practical, first
algorithm and show that a simple variant of the algorithm achieves the optimal
regret for a broad class of important applications.
| stat |
Online Action Learning in High Dimensions: A New Exploration Rule for
Contextual $\epsilon_t$-Greedy Heuristics | Bandit problems are pervasive in various fields of research and are also
present in several practical applications. Examples, including dynamic pricing
and assortment and the design of auctions and incentives, permeate a large
number of sequential treatment experiments. Different applications impose
distinct levels of restrictions on viable actions. Some favor diversity of
outcomes, while others require harmful actions to be closely monitored or
mainly avoided. In this paper, we extend one of the most popular bandit
solutions, the original $\epsilon_t$-greedy heuristics, to high-dimensional
contexts. Moreover, we introduce a competing exploration mechanism that counts
with searching sets based on order statistics. We view our proposals as
alternatives for cases where pluralism is valued or, in the opposite direction,
cases where the end-user should carefully tune the range of exploration of new
actions. We find reasonable bounds for the cumulative regret of a decaying
$\epsilon_t$-greedy heuristic in both cases and we provide an upper bound for
the initialization phase that implies the regret bounds when order statistics
are considered to be at most equal but mostly better than the case when random
searching is the sole exploration mechanism. Additionally, we show that
end-users have sufficient flexibility to avoid harmful actions since any
cardinality for the higher-order statistics can be used to achieve an stricter
upper bound. We illustrate the algorithms proposed in this paper both with
simulated and real data.
| stat |
Inference for high-dimensional linear mixed-effects models: A
quasi-likelihood approach | Linear mixed-effects models are widely used in analyzing clustered or
repeated measures data. We propose a quasi-likelihood approach for estimation
and inference of the unknown parameters in linear mixed-effects models with
high-dimensional fixed effects. The proposed method is applicable to general
settings where the dimension of the random effects and the cluster sizes are
possibly large. Regarding the fixed effects, we provide rate optimal estimators
and valid inference procedures that do not rely on the structural information
of the variance components. We also study the estimation of variance components
with high-dimensional fixed effects in general settings. The algorithms are
easy to implement and computationally fast. The proposed methods are assessed
in various simulation settings and are applied to a real study regarding the
associations between body mass index and genetic polymorphic markers in a
heterogeneous stock mice population.
| stat |
Convolutional Phase Retrieval via Gradient Descent | We study the convolutional phase retrieval problem, of recovering an unknown
signal $\mathbf x \in \mathbb C^n $ from $m$ measurements consisting of the
magnitude of its cyclic convolution with a given kernel $\mathbf a \in \mathbb
C^m $. This model is motivated by applications such as channel estimation,
optics, and underwater acoustic communication, where the signal of interest is
acted on by a given channel/filter, and phase information is difficult or
impossible to acquire. We show that when $\mathbf a$ is random and the number
of observations $m$ is sufficiently large, with high probability $\mathbf x$
can be efficiently recovered up to a global phase shift using a combination of
spectral initialization and generalized gradient descent. The main challenge is
coping with dependencies in the measurement operator. We overcome this
challenge by using ideas from decoupling theory, suprema of chaos processes and
the restricted isometry property of random circulant matrices, and recent
analysis of alternating minimization methods.
| stat |
A Locally Adaptive Interpretable Regression | Machine learning models with both good predictability and high
interpretability are crucial for decision support systems. Linear regression is
one of the most interpretable prediction models. However, the linearity in a
simple linear regression worsens its predictability. In this work, we introduce
a locally adaptive interpretable regression (LoAIR). In LoAIR, a metamodel
parameterized by neural networks predicts percentile of a Gaussian distribution
for the regression coefficients for a rapid adaptation. Our experimental
results on public benchmark datasets show that our model not only achieves
comparable or better predictive performance than the other state-of-the-art
baselines but also discovers some interesting relationships between input and
target variables such as a parabolic relationship between CO2 emissions and
Gross National Product (GNP). Therefore, LoAIR is a step towards bridging the
gap between econometrics, statistics, and machine learning by improving the
predictive ability of linear regression without depreciating its
interpretability.
| stat |
Unnormalized Variational Bayes | We unify empirical Bayes and variational Bayes for approximating unnormalized
densities. This framework, named unnormalized variational Bayes (UVB), is based
on formulating a latent variable model for the random variable
$Y=X+N(0,\sigma^2 I_d)$ and using the evidence lower bound (ELBO), computed by
a variational autoencoder, as a parametrization of the energy function of $Y$
which is then used to estimate $X$ with the empirical Bayes least-squares
estimator. In this intriguing setup, the $\textit{gradient}$ of the ELBO with
respect to noisy inputs plays the central role in learning the energy function.
Empirically, we demonstrate that UVB has a higher capacity to approximate
energy functions than the parametrization with MLPs as done in neural empirical
Bayes (DEEN). We especially showcase $\sigma=1$, where the differences between
UVB and DEEN become visible and qualitative in the denoising experiments. For
this high level of noise, the distribution of $Y$ is very smoothed and we
demonstrate that one can traverse in a single run $-$ without a restart $-$ all
MNIST classes in a variety of styles via walk-jump sampling with a fast-mixing
Langevin MCMC sampler. We finish by probing the encoder/decoder of the trained
models and confirm UVB $\neq$ VAE.
| stat |
Targeted Estimation of Heterogeneous Treatment Effect in Observational
Survival Analysis | The aim of clinical effectiveness research using repositories of electronic
health records is to identify what health interventions 'work best' in
real-world settings. Since there are several reasons why the net benefit of
intervention may differ across patients, current comparative effectiveness
literature focuses on investigating heterogeneous treatment effect and
predicting whether an individual might benefit from an intervention. The
majority of this literature has concentrated on the estimation of the effect of
treatment on binary outcomes. However, many medical interventions are evaluated
in terms of their effect on future events, which are subject to loss to
follow-up. In this study, we describe a framework for the estimation of
heterogeneous treatment effect in terms of differences in time-to-event
(survival) probabilities. We divide the problem into three phases: (1)
estimation of treatment effect conditioned on unique sets of the covariate
vector; (2) identification of features important for heterogeneity using an
ensemble of non-parametric variable importance methods; and (3) estimation of
treatment effect on the reference classes defined by the previously selected
features, using one-step Targeted Maximum Likelihood Estimation. We conducted a
series of simulation studies and found that this method performs well when
either sample size or event rate is high enough and the number of covariates
contributing to the effect heterogeneity is moderate. An application of this
method to a clinical case study was conducted by estimating the effect of oral
anticoagulants on newly diagnosed non-valvular atrial fibrillation patients
using data from the UK Clinical Practice Research Datalink.
| stat |
Asymptotic Errors for Teacher-Student Convex Generalized Linear Models
(or : How to Prove Kabashima's Replica Formula) | There has been a recent surge of interest in the study of asymptotic
reconstruction performance in various cases of generalized linear estimation
problems in the teacher-student setting, especially for the case of i.i.d
standard normal matrices. Here, we go beyond these matrices, and prove an
analytical formula for the reconstruction performance of convex generalized
linear models with rotationally-invariant data matrices with arbitrary bounded
spectrum, rigorously confirming a conjecture originally derived using the
replica method from statistical physics. The formula includes many problems
such as compressed sensing or sparse logistic classification. The proof is
achieved by leveraging on message passing algorithms and the statistical
properties of their iterates, allowing to characterize the asymptotic empirical
distribution of the estimator. Our proof is crucially based on the construction
of converging sequences of an oracle multi-layer vector approximate message
passing algorithm, where the convergence analysis is done by checking the
stability of an equivalent dynamical system. We illustrate our claim with
numerical examples on mainstream learning methods such as sparse logistic
regression and linear support vector classifiers, showing excellent agreement
between moderate size simulation and the asymptotic prediction.
| stat |
Reduced-order modeling using Dynamic Mode Decomposition and Least Angle
Regression | Dynamic Mode Decomposition (DMD) yields a linear, approximate model of a
system's dynamics that is built from data. We seek to reduce the order of this
model by identifying a reduced set of modes that best fit the output. We adopt
a model selection algorithm from statistics and machine learning known as Least
Angle Regression (LARS). We modify LARS to be complex-valued and utilize LARS
to select DMD modes. We refer to the resulting algorithm as Least Angle
Regression for Dynamic Mode Decomposition (LARS4DMD). Sparsity-Promoting
Dynamic Mode Decomposition (DMDSP), a popular mode-selection algorithm, serves
as a benchmark for comparison. Numerical results from a Poiseuille flow test
problem show that LARS4DMD yields reduced-order models that have comparable
performance to DMDSP. LARS4DMD has the added benefit that the regularization
weighting parameter required for DMDSP is not needed.
| stat |
Hierarchical causal variance decomposition for institution and provider
comparisons in healthcare | Disease-specific quality indicators (QIs) are used to compare institutions
and health care providers in terms processes or outcomes relevant to treatment
of a particular condition. In the context of surgical cancer treatments, the
performance variations can be due to hospital and/or surgeon level differences,
creating a hierarchical clustering. We consider how the observed variation in
care received at patient level can be decomposed into that causally explained
by the hospital performance, surgeon performance within hospital, patient
case-mix, and unexplained (residual) variation. For this purpose, we derive a
four-way variance decomposition, with particular attention to the causal
interpretation of the components. For estimation, we use inputs from a
mixed-effect model with nested random hospital/surgeon-specific effects, and a
multinomial logistic model for the hospital/surgeon-specific patient
populations. We investigate the performance of our methods in a simulation
study.
| stat |
Scalable penalized spatiotemporal land-use regression for ground-level
nitrogen dioxide | Nitrogen dioxide (NO$_2$) is a primary constituent of traffic-related air
pollution and has well established harmful environmental and human-health
impacts. Knowledge of the spatiotemporal distribution of NO$_2$ is critical for
exposure and risk assessment. A common approach for assessing air pollution
exposure is linear regression involving spatially referenced covariates, known
as land-use regression (LUR). We develop a scalable approach for simultaneous
variable selection and estimation of LUR models with spatiotemporally
correlated errors, by combining a general-Vecchia Gaussian process
approximation with a penalty on the LUR coefficients. In comparisons to
existing methods using simulated data, our approach resulted in higher
model-selection specificity and sensitivity and in better prediction in terms
of calibration and sharpness, for a wide range of relevant settings. In our
spatiotemporal analysis of daily, US-wide, ground-level NO$_2$ data, our
approach was more accurate, and produced a sparser and more interpretable
model. Our daily predictions elucidate spatiotemporal patterns of NO$_2$
concentrations across the United States, including significant variations
between cities and intra-urban variation. Thus, our predictions will be useful
for epidemiological and risk-assessment studies seeking daily, national-scale
predictions, and they can be used in acute-outcome health-risk assessments.
| stat |
Testing Conditional Independence in Supervised Learning Algorithms | We propose the conditional predictive impact (CPI), a consistent and unbiased
estimator of the association between one or several features and a given
outcome, conditional on a reduced feature set. Building on the knockoff
framework of Cand\`es et al. (2018), we develop a novel testing procedure that
works in conjunction with any valid knockoff sampler, supervised learning
algorithm, and loss function. The CPI can be efficiently computed for
high-dimensional data without any sparsity constraints. We demonstrate
convergence criteria for the CPI and develop statistical inference procedures
for evaluating its magnitude, significance, and precision. These tests aid in
feature and model selection, extending traditional frequentist and Bayesian
techniques to general supervised learning tasks. The CPI may also be applied in
causal discovery to identify underlying multivariate graph structures. We test
our method using various algorithms, including linear regression, neural
networks, random forests, and support vector machines. Empirical results show
that the CPI compares favorably to alternative variable importance measures and
other nonparametric tests of conditional independence on a diverse array of
real and simulated datasets. Simulations confirm that our inference procedures
successfully control Type I error and achieve nominal coverage probability. Our
method has been implemented in an R package, cpi, which can be downloaded from
https://github.com/dswatson/cpi.
| stat |
Approximate Bayesian Model Inversion for PDEs with Heterogeneous and
State-Dependent Coefficients | We present two approximate Bayesian inference methods for parameter
estimation in partial differential equation (PDE) models with space-dependent
and state-dependent parameters. We demonstrate that these methods provide
accurate and cost-effective alternatives to Markov Chain Monte Carlo
simulation. We assume a parameterized Gaussian prior on the unknown functions,
and approximate the posterior density by a parameterized multivariate Gaussian
density. The parameters of the prior and posterior are estimated from sparse
observations of the PDE model's states and the unknown functions themselves by
maximizing the evidence lower bound (ELBO), a lower bound on the log marginal
likelihood of the observations. The first method, Laplace-EM, employs the
expectation maximization algorithm to maximize the ELBO, with a Laplace
approximation of the posterior on the E-step, and minimization of a
Kullback-Leibler divergence on the M-step. The second method, DSVI-EB, employs
the doubly stochastic variational inference (DSVI) algorithm, in which the ELBO
is maximized via gradient-based stochastic optimization, with nosiy gradients
computed via simple Monte Carlo sampling and Gaussian backpropagation. We apply
these methods to identifying diffusion coefficients in linear and nonlinear
diffusion equations, and we find that both methods provide accurate estimates
of posterior densities and the hyperparameters of Gaussian priors. While the
Laplace-EM method is more accurate, it requires computing Hessians of the
physics model. The DSVI-EB method is found to be less accurate but only
requires gradients of the physics model.
| stat |
DTR Bandit: Learning to Make Response-Adaptive Decisions With Low Regret | Dynamic treatment regimes (DTRs) are personalized, adaptive, multi-stage
treatment plans that adapt treatment decisions both to an individual's initial
features and to intermediate outcomes and features at each subsequent stage,
which are affected by decisions in prior stages. Examples include personalized
first- and second-line treatments of chronic conditions like diabetes, cancer,
and depression, which adapt to patient response to first-line treatment,
disease progression, and individual characteristics. While existing literature
mostly focuses on estimating the optimal DTR from offline data such as from
sequentially randomized trials, we study the problem of developing the optimal
DTR in an online manner, where the interaction with each individual affect both
our cumulative reward and our data collection for future learning. We term this
the DTR bandit problem. We propose a novel algorithm that, by carefully
balancing exploration and exploitation, is guaranteed to achieve rate-optimal
regret when the transition and reward models are linear. We demonstrate our
algorithm and its benefits both in synthetic experiments and in a case study of
adaptive treatment of major depressive disorder using real-world data.
| stat |
Low-Rank Autoregressive Tensor Completion for Multivariate Time Series
Forecasting | Time series prediction has been a long-standing research topic and an
essential application in many domains. Modern time series collected from sensor
networks (e.g., energy consumption and traffic flow) are often large-scale and
incomplete with considerable corruption and missing values, making it difficult
to perform accurate predictions. In this paper, we propose a low-rank
autoregressive tensor completion (LATC) framework to model multivariate time
series data. The key of LATC is to transform the original multivariate time
series matrix (e.g., sensor$\times$time point) to a third-order tensor
structure (e.g., sensor$\times$time of day$\times$day) by introducing an
additional temporal dimension, which allows us to model the inherent rhythms
and seasonality of time series as global patterns. With the tensor structure,
we can transform the time series prediction and missing data imputation
problems into a universal low-rank tensor completion problem. Besides
minimizing tensor rank, we also integrate a novel autoregressive norm on the
original matrix representation into the objective function. The two components
serve different roles. The low-rank structure allows us to effectively capture
the global consistency and trends across all the three dimensions (i.e.,
similarity among sensors, similarity of different days, and current time v.s.
the same time of historical days). The autoregressive norm can better model the
local temporal trends. Our numerical experiments on three real-world data sets
demonstrate the superiority of the integration of global and local trends in
LATC in both missing data imputation and rolling prediction tasks.
| stat |
Unfolded Algorithms for Deep Phase Retrieval | Exploring the idea of phase retrieval has been intriguing researchers for
decades, due to its appearance in a wide range of applications. The task of a
phase retrieval algorithm is typically to recover a signal from linear
phaseless measurements. In this paper, we approach the problem by proposing a
hybrid model-based data-driven deep architecture, referred to as Unfolded Phase
Retrieval (UPR), that exhibits significant potential in improving the
performance of state-of-the art data-driven and model-based phase retrieval
algorithms. The proposed method benefits from versatility and interpretability
of well-established model-based algorithms, while simultaneously benefiting
from the expressive power of deep neural networks. In particular, our proposed
model-based deep architecture is applied to the conventional phase retrieval
problem (via the incremental reshaped Wirtinger flow algorithm) and the sparse
phase retrieval problem (via the sparse truncated amplitude flow algorithm),
showing immense promise in both cases. Furthermore, we consider a joint design
of the sensing matrix and the signal processing algorithm and utilize the deep
unfolding technique in the process. Our numerical results illustrate the
effectiveness of such hybrid model-based and data-driven frameworks and
showcase the untapped potential of data-aided methodologies to enhance the
existing phase retrieval algorithms.
| stat |
Which Minimizer Does My Neural Network Converge To? | The loss surface of an overparameterized neural network (NN) possesses many
global minima of zero training error. We explain how common variants of the
standard NN training procedure change the minimizer obtained. First, we make
explicit how the size of the initialization of a strongly overparameterized NN
affects the minimizer and can deteriorate its final test performance. We
propose a strategy to limit this effect. Then, we demonstrate that for adaptive
optimization such as AdaGrad, the obtained minimizer generally differs from the
gradient descent (GD) minimizer. This adaptive minimizer is changed further by
stochastic mini-batch training, even though in the non-adaptive case GD and
stochastic GD result in essentially the same minimizer. Lastly, we explain that
these effects remain relevant for less overparameterized NNs. While
overparameterization has its benefits, our work highlights that it induces
sources of error absent from underparameterized models, some of which can be
challenging to control.
| stat |
Optimizing Graphical Procedures for Multiplicity Control in a
Confirmatory Clinical Trial via Deep Learning | In confirmatory clinical trials, it has been proposed [Bretz et al., 2009] to
use a simple iterative graphical approach to constructing and performing
intersection hypotheses tests using weighted Bonferroni-type procedures to
control type I errors in the strong sense. Given Phase II study results or
prior knowledge, it is usually of main interest to find the optimal graph that
maximizes a certain objective function in a future Phase III study. However,
lack of a closed form expression in the objective function makes the
optimization challenging. In this manuscript, we propose a general optimization
framework to obtain the global maximum via feedforward neural networks (FNNs)
in deep learning. Simulation studies show that our FNN-based approach has a
better balance between robustness and time efficiency than some existing
derivative-free constrained optimization algorithms. Compared to the
traditional window searching approach, our optimizer has moderate multiplicity
adjusted power gain when the number of hypotheses is relatively large or when
their correlations are high. We further apply it to a case study to illustrate
how to optimize a multiple test procedure with respect to a specific study
objective.
| stat |
What can the millions of random treatments in nonexperimental data
reveal about causes? | We propose a new method to estimate causal effects from nonexperimental data.
Each pair of sample units is first associated with a stochastic 'treatment' -
differences in factors between units - and an effect - a resultant outcome
difference. It is then proposed that all such pairs can be combined to provide
more accurate estimates of causal effects in observational data, provided a
statistical model connecting combinatorial properties of treatments to the
accuracy and unbiasedness of their effects. The article introduces one such
model and a Bayesian approach to combine the $O(n^2)$ pairwise observations
typically available in nonexperimnetal data. This also leads to an
interpretation of nonexperimental datasets as incomplete, or noisy, versions of
ideal factorial experimental designs.
This approach to causal effect estimation has several advantages: (1) it
expands the number of observations, converting thousands of individuals into
millions of observational treatments; (2) starting with treatments closest to
the experimental ideal, it identifies noncausal variables that can be ignored
in the future, making estimation easier in each subsequent iteration while
departing minimally from experiment-like conditions; (3) it recovers individual
causal effects in heterogeneous populations. We evaluate the method in
simulations and the National Supported Work (NSW) program, an intensively
studied program whose effects are known from randomized field experiments. We
demonstrate that the proposed approach recovers causal effects in common NSW
samples, as well as in arbitrary subpopulations and an order-of-magnitude
larger supersample with the entire national program data, outperforming
Statistical, Econometrics and Machine Learning estimators in all cases...
| stat |
Evaluating Catchment Models as Multiple Working Hypotheses: on the Role
of Error Metrics, Parameter Sampling, Model Structure, and Data Information
Content | To evaluate models as hypotheses, we developed the method of Flux Mapping to
construct a hypothesis space based on dominant runoff generating mechanisms.
Acceptable model runs, defined as total simulated flow with similar (and
minimal) model error, are mapped to the hypothesis space given their simulated
runoff components. In each modeling case, the hypothesis space is the result of
an interplay of factors: model structure and parameterization, chosen error
metric, and data information content. The aim of this study is to disentangle
the role of each factor in model evaluation. We used two model structures
(SACRAMENTO and SIMHYD), two parameter sampling approaches (Latin Hypercube
Sampling of the parameter space and guided-search of the solution space), three
widely used error metrics (Nash-Sutcliffe Efficiency - NSE, Kling-Gupta
Efficiency skill score - KGEss, and Willmott refined Index of Agreement - WIA),
and hydrological data from a large sample of Australian catchments. First, we
characterized how the three error metrics behave under different error types
and magnitudes independent of any modeling. We then conducted a series of
controlled experiments to unpack the role of each factor in runoff generation
hypotheses. We show that KGEss is a more reliable metric compared to NSE and
WIA for model evaluation. We further demonstrate that only changing the error
metric -- while other factors remain constant -- can change the model solution
space and hence vary model performance, parameter sampling sufficiency, and or
the flux map. We show how unreliable error metrics and insufficient parameter
sampling impair model-based inferences, particularly runoff generation
hypotheses.
| stat |
Estimating Stochastic Poisson Intensities Using Deep Latent Models | We present methodology for estimating the stochastic intensity of a doubly
stochastic Poisson process. Statistical and theoretical analyses of traffic
traces show that these processes are appropriate models of high intensity
traffic arriving at an array of service systems. The statistical estimation of
the underlying latent stochastic intensity process driving the traffic model
involves a rather complicated nonlinear filtering problem. We develop a novel
simulation methodology, using deep neural networks to approximate the path
measures induced by the stochastic intensity process, for solving this
nonlinear filtering problem. Our simulation studies demonstrate that the method
is quite accurate on both in-sample estimation and on an out-of-sample
performance prediction task for an infinite server queue.
| stat |
Repeated measurements with unintended feedback: The Dutch new herring
scandals | An econometric analysis of consumer research data which hit newspaper
headlines in the Netherlands illustrates almost everything that can go wrong
when statistical models are fit to the superficial characteristics of a
data-set with no attention paid to the data generation mechanism. This paper is
dedicated to Ornulf Borgan on the occasion of his virtual 65th birthday
celebrations.
| stat |
A presmoothing approach for estimation in mixture cure models | A challenge when dealing with survival analysis data is accounting for a cure
fraction, meaning that some subjects will never experience the event of
interest. Mixture cure models have been frequently used to estimate both the
probability of being cured and the time to event for the susceptible subjects,
by usually assuming a parametric (logistic) form of the incidence. We propose a
new estimation procedure for a parametric cure rate that relies on a
preliminary smooth estimator and is independent of the model assumed for the
latency. We investigate the theoretical properties of the estimators and show
through simulations that, in the logistic/Cox model, presmoothing leads to more
accurate results compared to the maximum likelihood estimator. To illustrate
the practical use, we apply the new estimation procedure to two studies of
melanoma survival data.
| stat |
Causal inference methods for small non-randomized studies: Methods and
an application in COVID-19 | The usual development cycles are too slow for the development of vaccines,
diagnostics and treatments in pandemics such as the ongoing SARS-CoV-2
pandemic. Given the pressure in such a situation, there is a risk that findings
of early clinical trials are overinterpreted despite their limitations in terms
of size and design. Motivated by a non-randomized open-label study
investigating the efficacy of hydroxychloroquine in patients with COVID-19, we
describe in a unified fashion various alternative approaches to the analysis of
non-randomized studies. A widely used tool to reduce the impact of
treatment-selection bias are so-called propensity score (PS) methods.
Conditioning on the propensity score allows one to replicate the design of a
randomized controlled trial, conditional on observed covariates. Extensions
include the g-computation approach, which is less frequently applied, in
particular in clinical studies. Moreover, doubly robust estimators provide
additional advantages. Here, we investigate the properties of propensity score
based methods including three variations of doubly robust estimators in small
sample settings, typical for early trials, in a simulation study. R code for
the simulations is provided.
| stat |
HCmodelSets: An R package for specifying sets of well-fitting models in
regression with a large number of potential explanatory variables | In the context of regression with a large number of explanatory variables,
Cox and Battey (2017) emphasize that if there are alternative reasonable
explanations of the data that are statistically indistinguishable, one should
aim to specify as many of these explanations as is feasible. The standard
practice, by contrast, is to report a single model effective for prediction.
The present paper illustrates the R implementation of the new ideas in the
package `HCmodelSets', using simple reproducible examples and real data.
Results of some simulation experiments are also reported.
| stat |
On Shapley Credit Allocation for Interpretability | We emphasize the importance of asking the right question when interpreting
the decisions of a learning model. We discuss a natural extension of the
theoretical machinery from Janzing et. al. 2020, which answers the question
"Why did my model predict a person has cancer?" for answering a more involved
question, "What caused my model to predict a person has cancer?" While the
former quantifies the direct effects of variables on the model, the latter also
accounts for indirect effects, thereby providing meaningful insights wherever
human beings can reason in terms of cause and effect. We propose three broad
categories for interpretations: observational, model-specific and causal each
of which are significant in their own right. Furthermore, this paper quantifies
feature relevance by weaving different natures of interpretations together with
different measures as characteristic functions for Shapley symmetrization.
Besides the widely used expected value of the model, we also discuss measures
of statistical uncertainty and dispersion as informative candidates, and their
merits in generating explanations for each data point, some of which are used
in this context for the first time. These measures are not only useful for
studying the influence of variables on the model output, but also on the
predictive performance of the model, and for that we propose relevant
characteristic functions that are also used for the first time.
| stat |
Mining gold from implicit models to improve likelihood-free inference | Simulators often provide the best description of real-world phenomena.
However, they also lead to challenging inverse problems because the density
they implicitly define is often intractable. We present a new suite of
simulation-based inference techniques that go beyond the traditional
Approximate Bayesian Computation approach, which struggles in a
high-dimensional setting, and extend methods that use surrogate models based on
neural networks. We show that additional information, such as the joint
likelihood ratio and the joint score, can often be extracted from simulators
and used to augment the training data for these surrogate models. Finally, we
demonstrate that these new techniques are more sample efficient and provide
higher-fidelity inference than traditional methods.
| stat |
Solving high-dimensional parabolic PDEs using the tensor train format | High-dimensional partial differential equations (PDEs) are ubiquitous in
economics, science and engineering. However, their numerical treatment poses
formidable challenges since traditional grid-based methods tend to be
frustrated by the curse of dimensionality. In this paper, we argue that tensor
trains provide an appealing approximation framework for parabolic PDEs: the
combination of reformulations in terms of backward stochastic differential
equations and regression-type methods in the tensor format holds the promise of
leveraging latent low-rank structures enabling both compression and efficient
computation. Following this paradigm, we develop novel iterative schemes,
involving either explicit and fast or implicit and accurate updates. We
demonstrate in a number of examples that our methods achieve a favorable
trade-off between accuracy and computational efficiency in comparison with
state-of-the-art neural network based approaches.
| stat |
Statistically-Robust Clustering Techniques for Mapping Spatial Hotspots:
A Survey | Mapping of spatial hotspots, i.e., regions with significantly higher rates or
probability density of generating certain events (e.g., disease or crime
cases), is a important task in diverse societal domains, including public
health, public safety, transportation, agriculture, environmental science, etc.
Clustering techniques required by these domains differ from traditional
clustering methods due to the high economic and social costs of spurious
results (e.g., false alarms of crime clusters). As a result, statistical rigor
is needed explicitly to control the rate of spurious detections. To address
this challenge, techniques for statistically-robust clustering have been
extensively studied by the data mining and statistics communities. In this
survey we present an up-to-date and detailed review of the models and
algorithms developed by this field. We first present a general taxonomy of the
clustering process with statistical rigor, covering key steps of data and
statistical modeling, region enumeration and maximization, significance
testing, and data update. We further discuss different paradigms and methods
within each of key steps. Finally, we highlight research gaps and potential
future directions, which may serve as a stepping stone in generating new ideas
and thoughts in this growing field and beyond.
| stat |
Optimal curtailed designs for single arm phase II clinical trials | In single-arm phase II oncology trials, the most popular choice of design is
Simon's two-stage design, which allows early stopping at one interim analysis.
However, the expected trial sample size can be reduced further by allowing
curtailment. Curtailment is stopping when the final go or no-go decision is
certain, so-called non-stochastic curtailment, or very likely, known as
stochastic curtailment.
In the context of single-arm phase II oncology trials, stochastic curtailment
has previously been restricted to stopping in the second stage and/or stopping
for a no-go decision only. We introduce two designs that incorporate stochastic
curtailment and allow stopping after every observation, for either a go or
no-go decision. We obtain optimal stopping boundaries by searching over a range
of potential conditional powers, beyond which the trial will stop for a go or
no-go decision. This search is novel: firstly, the search is undertaken over a
range of values unique to each possible design realisation. Secondly, these
values are evaluated taking into account the possibility of early stopping.
Finally, each design realisation's operating characteristics are obtained
exactly.
The proposed designs are compared to existing designs in a real data example.
They are also compared under three scenarios, both with respect to four single
optimality criteria and using a loss function.
The proposed designs are superior in almost all cases. Optimising for the
expected sample size under either the null or alternative hypothesis, the
saving compared to the popular Simon's design ranges from 22% to 55%.
| stat |
Generalization in Deep Learning | This paper provides theoretical insights into why and how deep learning can
generalize well, despite its large capacity, complexity, possible algorithmic
instability, nonrobustness, and sharp minima, responding to an open question in
the literature. We also discuss approaches to provide non-vacuous
generalization guarantees for deep learning. Based on theoretical observations,
we propose new open problems and discuss the limitations of our results.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.