title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Noise-robust classification with hypergraph neural network | This paper presents a novel version of the hypergraph neural network method.
This method is utilized to solve the noisy label learning problem. First, we
apply the PCA dimensional reduction technique to the feature matrices of the
image datasets in order to reduce the "noise" and the redundant features in the
feature matrices of the image datasets and to reduce the runtime constructing
the hypergraph of the hypergraph neural network method. Then, the classic
graph-based semi-supervised learning method, the classic hypergraph based
semi-supervised learning method, the graph neural network, the hypergraph
neural network, and our proposed hypergraph neural network are employed to
solve the noisy label learning problem. The accuracies of these five methods
are evaluated and compared. Experimental results show that the hypergraph
neural network methods achieve the best performance when the noise level
increases. Moreover, the hypergraph neural network methods are at least as good
as the graph neural network.
| stat |
Simulation study of estimating between-study variance and overall effect
in meta-analysis of standardized mean difference | Methods for random-effects meta-analysis require an estimate of the
between-study variance, $\tau^2$. The performance of estimators of $\tau^2$
(measured by bias and coverage) affects their usefulness in assessing
heterogeneity of study-level effects, and also the performance of related
estimators of the overall effect. For the effect measure standardized mean
difference (SMD), we provide the results from extensive simulations on five
point estimators of $\tau^2$ (the popular methods of DerSimonian-Laird,
restricted maximum likelihood, and Mandel and Paule (MP); the less-familiar
method of Jackson; the new method (KDB) based on the improved approximation to
the distribution of the Q statistic by Kulinskaya, Dollinger and
Bj{\o}rkest{\o}l (2011) ), five interval estimators for $\tau^2$ (profile
likelihood, Q-profile, Biggerstaff and Jackson, Jackson, and the new KDB
method), six point estimators of the overall effect (the five related to the
point estimators of $\tau^2$ and an estimator whose weights use only
study-level sample sizes), and eight interval estimators for the overall effect
(five based on the point estimators for $\tau^2$; the
Hartung-Knapp-Sidik-Jonkman (HKSJ) interval; a modification of HKSJ; and an
interval based on the sample-size-weighted estimator).
| stat |
A fast algorithm for maximum likelihood estimation of mixture
proportions using sequential quadratic programming | Maximum likelihood estimation of mixture proportions has a long history, and
continues to play an important role in modern statistics, including in
development of nonparametric empirical Bayes methods. Maximum likelihood of
mixture proportions has traditionally been solved using the expectation
maximization (EM) algorithm, but recent work by Koenker & Mizera shows that
modern convex optimization techniques -- in particular, interior point methods
-- are substantially faster and more accurate than EM. Here, we develop a new
solution based on sequential quadratic programming (SQP). It is substantially
faster than the interior point method, and just as accurate. Our approach
combines several ideas: first, it solves a reformulation of the original
problem; second, it uses an SQP approach to make the best use of the expensive
gradient and Hessian computations; third, the SQP iterations are implemented
using an active set method to exploit the sparse nature of the quadratic
subproblems; fourth, it uses accurate low-rank approximations for more
efficient gradient and Hessian computations. We illustrate the benefits of our
approach in experiments on synthetic data sets as well as a large genetic
association data set. In large data sets (n = 1,000,000 observations, m = 1,000
mixture components), our implementation achieves at least 100-fold reduction in
runtime compared with a state-of-the-art interior point solver. Our methods are
implemented in Julia, and in an R package available on CRAN (see
https://CRAN.R-project.org/package=mixsqp).
| stat |
Bayesian Semi-supervised learning under nonparanormality | Semi-supervised learning is a classification method which makes use of both
labeled data and unlabeled data for training. In this paper, we propose a
semi-supervised learning algorithm using a Bayesian semi-supervised model. We
make a general assumption that the observations will follow two multivariate
normal distributions depending on their true labels after the same unknown
transformation. We use B-splines to put a prior on the transformation function
for each component. To use unlabeled data in a semi-supervised setting, we
assume the labels are missing at random. The posterior distributions can then
be described using our assumptions, which we compute by the Gibbs sampling
technique. The proposed method is then compared with several other available
methods through an extensive simulation study. Finally we apply the proposed
method in real data contexts for diagnosing breast cancer and classify radar
returns. We conclude that the proposed method has better prediction accuracy in
a wide variety of cases.
| stat |
The uncertainty estimation of feature-based forecast combinations | Forecasting is an indispensable element of operational research (OR) and an
important aid to planning. The accurate estimation of the forecast uncertainty
facilitates several operations management activities, predominantly in
supporting decisions in inventory and supply chain management and effectively
setting safety stocks. In this paper, we introduce a feature-based framework,
which links the relationship between time series features and the interval
forecasting performance into providing reliable interval forecasts. We propose
an optimal threshold ratio searching algorithm and a new weight determination
mechanism for selecting an appropriate subset of models and assigning
combination weights for each time series tailored to the observed features. We
evaluate our approach using a large set of time series from the M4 competition.
Our experiments show that our approach significantly outperforms a wide range
of benchmark models, both in terms of point forecasts as well as prediction
intervals.
| stat |
A scale-dependent notion of effective dimension | We introduce a notion of "effective dimension" of a statistical model based
on the number of cubes of size $1/\sqrt{n}$ needed to cover the model space
when endowed with the Fisher Information Matrix as metric, $n$ being the number
of observations. The number of observations fixes a natural scale or
resolution. The effective dimension is then measured via the spectrum of the
Fisher Information Matrix regularized using this natural scale.
| stat |
How Personal is Machine Learning Personalization? | Though used extensively, the concept and process of machine learning (ML)
personalization have generally received little attention from academics,
practitioners, and the general public. We describe the ML approach as relying
on the metaphor of the person as a feature vector and contrast this with
humanistic views of the person. In light of the recent calls by the IEEE to
consider the effects of ML on human well-being, we ask whether ML
personalization can be reconciled with these humanistic views of the person,
which highlight the importance of moral and social identity. As human behavior
increasingly becomes digitized, analyzed, and predicted, to what extent do our
subsequent decisions about what to choose, buy, or do, made both by us and
others, reflect who we are as persons? This paper first explicates the term
personalization by considering ML personalization and highlights its relation
to humanistic conceptions of the person, then proposes several dimensions for
evaluating the degree of personalization of ML personalized scores. By doing
so, we hope to contribute to current debate on the issues of algorithmic bias,
transparency, and fairness in machine learning.
| stat |
Spatial Factor Modeling: A Bayesian Matrix-Normal Approach for
Misaligned Data | Multivariate spatially-oriented data sets are prevalent in the environmental
and physical sciences. Scientists seek to jointly model multiple variables,
each indexed by a spatial location, to capture any underlying spatial
association for each variable and associations among the different dependent
variables. Multivariate latent spatial process models have proved effective in
driving statistical inference and rendering better predictive inference at
arbitrary locations for the spatial process. High-dimensional multivariate
spatial data, which is the theme of this article, refers to data sets where the
number of spatial locations and the number of spatially dependent variables is
very large. The field has witnessed substantial developments in scalable models
for univariate spatial processes, but such methods for multivariate spatial
processes, especially when the number of outcomes are moderately large, are
limited in comparison. Here, we extend scalable modeling strategies for a
single process to multivariate processes. We pursue Bayesian inference which is
attractive for full uncertainty quantification of the latent spatial process.
Our approach exploits distribution theory for the Matrix-Normal distribution,
which we use to construct scalable versions of a hierarchical linear model of
coregionalization (LMC) and spatial factor models that deliver inference over a
high-dimensional parameter space including the latent spatial process. We
illustrate the computational and inferential benefits of our algorithms over
competing methods using simulation studies and an analysis of a massive
vegetation index data set.
| stat |
Targeted Maximum Likelihood Estimation of Community-based Causal Effect
of Community-Level Stochastic Interventions | Unlike the commonly used parametric regression models such as mixed models,
that can easily violate the required statistical assumptions and result in
invalid statistical inference, target maximum likelihood estimation allows more
realistic data-generative models and provides double-robust, semi-parametric
and efficient estimators. Target maximum likelihood estimators (TMLEs) for the
causal effect of a community-level static exposure were previously proposed by
Balzer et al. In this manuscript, we build on this work and present
identifiability results and develop two semi-parametric efficient TMLEs for the
estimation of the causal effect of the single time-point community-level
stochastic intervention whose assignment mechanism can depend on measured and
unmeasured environmental factors and its individual-level covariates. The first
community-level TMLE is developed under a general hierarchical non-parametric
structural equation model, which can incorporate pooled individual-level
regressions for estimating the outcome mechanism. The second individual-level
TMLE is developed under a restricted hierarchical model in which the additional
assumption of no covariate interference within communities holds. The proposed
TMLEs have several crucial advantages. First, both TMLEs can make use of
individual level data in the hierarchical setting, and potentially reduce
finite sample bias and improve estimator efficiency. Second, the stochastic
intervention framework provides a natural way for defining and estimating
casual effects where the exposure variables are continuous or discrete with
multiple levels, or even cannot be directly intervened on. Also, the positivity
assumption needed for our proposed causal parameters can be weaker than the
version of positivity required for other casual parameters.
| stat |
Spectra of the Conjugate Kernel and Neural Tangent Kernel for
linear-width neural networks | We study the eigenvalue distributions of the Conjugate Kernel and Neural
Tangent Kernel associated to multi-layer feedforward neural networks. In an
asymptotic regime where network width is increasing linearly in sample size,
under random initialization of the weights, and for input samples satisfying a
notion of approximate pairwise orthogonality, we show that the eigenvalue
distributions of the CK and NTK converge to deterministic limits. The limit for
the CK is described by iterating the Marcenko-Pastur map across the hidden
layers. The limit for the NTK is equivalent to that of a linear combination of
the CK matrices across layers, and may be described by recursive fixed-point
equations that extend this Marcenko-Pastur map. We demonstrate the agreement of
these asymptotic predictions with the observed spectra for both synthetic and
CIFAR-10 training data, and we perform a small simulation to investigate the
evolutions of these spectra over training.
| stat |
CLARITY -- Comparing heterogeneous data using dissimiLARITY | Integrating datasets from different disciplines is hard because the data are
often qualitatively different in meaning, scale, and reliability. When two
datasets describe the same entities, many scientific questions can be phrased
around whether the similarities between entities are conserved. Our method,
CLARITY, quantifies consistency across datasets, identifies where
inconsistencies arise, and aids in their interpretation. We explore three
diverse comparisons: Gene Methylation vs Gene Expression, evolution of language
sounds vs word use, and country-level economic metrics vs cultural beliefs. The
non-parametric approach is robust to noise and differences in scaling, and
makes only weak assumptions about how the data were generated. It operates by
decomposing similarities into two components: the `structural' component
analogous to a clustering, and an underlying `relationship' between those
structures. This allows a `structural comparison' between two similarity
matrices using their predictability from `structure'. The software, CLARITY, is
available as an R package from https://github.com/danjlawson/CLARITY.
| stat |
Nonparametric tests of the causal null with non-discrete exposures | In many scientific studies, it is of interest to determine whether an
exposure has a causal effect on an outcome. In observational studies, this is a
challenging task due to the presence of confounding variables that affect both
the exposure and the outcome. Many methods have been developed to test for the
presence of a causal effect when all such confounding variables are observed
and when the exposure of interest is discrete. In this article, we propose a
class of nonparametric tests of the null hypothesis that there is no average
causal effect of an arbitrary univariate exposure on an outcome in the presence
of observed confounding. Our tests apply to discrete, continuous, and mixed
discrete-continuous exposures. We demonstrate that our proposed tests are
doubly-robust consistent, that they have correct asymptotic type I error if
both nuisance parameters involved in the problem are estimated at fast enough
rates, and that they have power to detect local alternatives approaching the
null at the rate $n^{-1/2}$. We study the performance of our tests in numerical
studies, and use them to test for the presence of a causal effect of smoking on
birthweight among smoking mothers.
| stat |
Meta-Active Learning for Node Response Prediction in Graphs | Meta-learning is an important approach to improve machine learning
performance with a limited number of observations for target tasks. However,
when observations are unbalancedly obtained, it is difficult to improve the
performance even with meta-learning methods. In this paper, we propose an
active learning method for meta-learning on node response prediction tasks in
attributed graphs, where nodes to observe are selected to improve performance
with as few observed nodes as possible. With the proposed method, we use models
based on graph convolutional neural networks for both predicting node responses
and selecting nodes, by which we can predict responses and select nodes even
for graphs with unseen response variables. The response prediction model is
trained by minimizing the expected test error. The node selection model is
trained by maximizing the expected error reduction with reinforcement learning.
We demonstrate the effectiveness of the proposed method with 11 types of road
congestion prediction tasks.
| stat |
To tune or not to tune, a case study of ridge logistic regression in
small or sparse datasets | For finite samples with binary outcomes penalized logistic regression such as
ridge logistic regression (RR) has the potential of achieving smaller mean
squared errors (MSE) of coefficients and predictions than maximum likelihood
estimation. There is evidence, however, that RR is sensitive to small or sparse
data situations, yielding poor performance in individual datasets. In this
paper, we elaborate this issue further by performing a comprehensive simulation
study, investigating the performance of RR in comparison to Firth's correction
that has been shown to perform well in low-dimensional settings. Performance of
RR strongly depends on the choice of complexity parameter that is usually tuned
by minimizing some measure of the out-of-sample prediction error or information
criterion. Alternatively, it may be determined according to prior assumptions
about true effects. As shown in our simulation and illustrated by a data
example, values optimized in small or sparse datasets are negatively correlated
with optimal values and suffer from substantial variability which translates
into large MSE of coefficients and large variability of calibration slopes. In
contrast, if the degree of shrinkage is pre-specified, accurate coefficients
and predictions can be obtained even in non-ideal settings such as encountered
in the context of rare outcomes or sparse predictors.
| stat |
The Challenge of Small Data: Dynamic Mode Decomposition, Redux | We revisit the setting and the assumptions that underlie the methodology of
Dynamic Mode Decomposition (DMD) in order to highlight caveats as well as
potential measures of when the applicability is warranted.
| stat |
Fast Statistical Leverage Score Approximation in Kernel Ridge Regression | Nystr\"om approximation is a fast randomized method that rapidly solves
kernel ridge regression (KRR) problems through sub-sampling the n-by-n
empirical kernel matrix appearing in the objective function. However, the
performance of such a sub-sampling method heavily relies on correctly
estimating the statistical leverage scores for forming the sampling
distribution, which can be as costly as solving the original KRR. In this work,
we propose a linear time (modulo poly-log terms) algorithm to accurately
approximate the statistical leverage scores in the stationary-kernel-based KRR
with theoretical guarantees. Particularly, by analyzing the first-order
condition of the KRR objective, we derive an analytic formula, which depends on
both the input distribution and the spectral density of stationary kernels, for
capturing the non-uniformity of the statistical leverage scores. Numerical
experiments demonstrate that with the same prediction accuracy our method is
orders of magnitude more efficient than existing methods in selecting the
representative sub-samples in the Nystr\"om approximation.
| stat |
Un-normalized hypergraph p-Laplacian based semi-supervised learning
methods | Most network-based machine learning methods assume that the labels of two
adjacent samples in the network are likely to be the same. However, assuming
the pairwise relationship between samples is not complete. The information a
group of samples that shows very similar pattern and tends to have similar
labels is missed. The natural way overcoming the information loss of the above
assumption is to represent the feature dataset of samples as the hypergraph.
Thus, in this paper, we will present the un-normalized hypergraph p-Laplacian
semi-supervised learning methods. These methods will be applied to the zoo
dataset and the tiny version of 20 newsgroups dataset. Experiment results show
that the accuracy performance measures of these un-normalized hypergraph
p-Laplacian based semi-supervised learning methods are significantly greater
than the accuracy performance measure of the un-normalized hypergraph Laplacian
based semi-supervised learning method (the current state of the art method
hypergraph Laplacian based semi-supervised learning method for classification
problem with p=2).
| stat |
Spillover Effects in Experimental Data | We present current methods for estimating treatment effects and spillover
effects under "interference", a term which covers a broad class of situations
in which a unit's outcome depends not only on treatments received by that unit,
but also on treatments received by other units. To the extent that units react
to each other, interact, or otherwise transmit effects of treatments, valid
inference requires that we account for such interference, which is a departure
from the traditional assumption that units' outcomes are affected only by their
own treatment assignment. Interference and associated spillovers may be a
nuisance or they may be of substantive interest to the researcher. In this
chapter, we focus on interference in the context of randomized experiments. We
review methods for when interference happens in a general network setting. We
then consider the special case where interference is contained within a
hierarchical structure. Finally, we discuss the relationship between
interference and contagion. We use the interference R package and simulated
data to illustrate key points. We consider efficient designs that allow for
estimation of the treatment and spillover effects and discuss recent empirical
studies that try to capture such effects.
| stat |
Unbiased estimators for the variance of MMD estimators | The maximum mean discrepancy (MMD) is a kernel-based distance between
probability distributions useful in many applications (Gretton et al. 2012),
bearing a simple estimator with pleasing computational and statistical
properties. Being able to efficiently estimate the variance of this estimator
is very helpful to various problems in two-sample testing. Towards this end,
Bounliphone et al. (2016) used the theory of U-statistics to derive estimators
for the variance of an MMD estimator, and differences between two such
estimators. Their estimator, however, drops lower-order terms, and is
unnecessarily biased. We show in this note - extending and correcting work of
Sutherland et al. (2017) - that we can find a truly unbiased estimator for the
actual variance of both the squared MMD estimator and the difference of two
correlated squared MMD estimators, at essentially no additional computational
cost.
| stat |
Convergence and Stability of Graph Convolutional Networks on Large
Random Graphs | We study properties of Graph Convolutional Networks (GCNs) by analyzing their
behavior on standard models of random graphs, where nodes are represented by
random latent variables and edges are drawn according to a similarity kernel.
This allows us to overcome the difficulties of dealing with discrete notions
such as isomorphisms on very large graphs, by considering instead more natural
geometric aspects. We first study the convergence of GCNs to their continuous
counterpart as the number of nodes grows. Our results are fully non-asymptotic
and are valid for relatively sparse graphs with an average degree that grows
logarithmically with the number of nodes. We then analyze the stability of GCNs
to small deformations of the random graph model. In contrast to previous
studies of stability in discrete settings, our continuous setup allows us to
provide more intuitive deformation-based metrics for understanding stability,
which have proven useful for explaining the success of convolutional
representations on Euclidean domains.
| stat |
Joint modeling with time-dependent treatment and heteroskedasticity:
Bayesian analysis with application to the Framingham Heart Study | Medical studies for chronic disease are often interested in the relation
between longitudinal risk factor profiles and individuals' later life disease
outcomes. These profiles may typically be subject to intermediate structural
changes due to treatment or environmental influences. Analysis of such studies
may be handled by the joint model framework. However, current joint modeling
does not consider structural changes in the residual variability of the risk
profile nor consider the influence of subject-specific residual variability on
the time-to-event outcome. In the present paper, we extend the joint model
framework to address these two heterogeneous intra-individual variabilities. A
Bayesian approach is used to estimate the unknown parameters and simulation
studies are conducted to investigate the performance of the method. The
proposed joint model is applied to the Framingham Heart Study to investigate
the influence of anti-hypertensive medication on the systolic blood pressure
variability together with its effect on the risk of developing cardiovascular
disease. We show that anti-hypertensive medication is associated with elevated
systolic blood pressure variability and increased variability elevates risk of
developing cardiovascular disease.
| stat |
Regret in Online Recommendation Systems | This paper proposes a theoretical analysis of recommendation systems in an
online setting, where items are sequentially recommended to users over time. In
each round, a user, randomly picked from a population of $m$ users, requests a
recommendation. The decision-maker observes the user and selects an item from a
catalogue of $n$ items. Importantly, an item cannot be recommended twice to the
same user. The probabilities that a user likes each item are unknown. The
performance of the recommendation algorithm is captured through its regret,
considering as a reference an Oracle algorithm aware of these probabilities. We
investigate various structural assumptions on these probabilities: we derive
for each structure regret lower bounds, and devise algorithms achieving these
limits. Interestingly, our analysis reveals the relative weights of the
different components of regret: the component due to the constraint of not
presenting the same item twice to the same user, that due to learning the
chances users like items, and finally that arising when learning the underlying
structure.
| stat |
Bayesian model averaging for analysis of lattice field theory results | Statistical modeling is a key component in the extraction of physical results
from lattice field theory calculations. Although the general models used are
often strongly motivated by physics, their precise form is typically
ill-determined, and many model variations can be plausibly considered for the
same lattice data. Model averaging, which amounts to a probability-weighted
average over all model variations, can incorporate systematic errors associated
with model choice without being overly conservative. We discuss the framework
of model averaging from the perspective of Bayesian statistics, and give useful
formulas and approximations for the particular case of least-squares fitting,
commonly used in modeling lattice results. In addition, we frame the common
problem of data subset selection (e.g. choice of minimum and maximum time
separation for fitting a two-point correlation function) as a model selection
problem, and study model averaging as a straightforward alternative to manual
selection of fit ranges. Numerical examples involving both mock and real
lattice data are given.
| stat |
Approximate Bayesian Inference via Sparse grid Quadrature Evaluation for
Hierarchical Models | We combine conditioning techniques with sparse grid quadrature rules to
develop a computationally efficient method to approximate marginal, but not
necessarily univariate, posterior quantities, yielding approximate Bayesian
inference via Sparse grid Quadrature Evaluation (BISQuE) for hierarchical
models. BISQuE reformulates posterior quantities as weighted integrals of
conditional quantities, such as densities and expectations. Sparse grid
quadrature rules allow computationally efficient approximation of high
dimensional integrals, which appear in hierarchical models with many
hyperparameters. BISQuE reduces computational effort relative to standard,
Markov chain Monte Carlo methods by at least two orders of magnitude on several
applied and illustrative models. We also briefly discuss using BISQuE to apply
Integrated Nested Laplace Approximations (INLA) to models with more
hyperparameters than is currently practical.
| stat |
A Bandit Approach to Multiple Testing with False Discovery Control | We propose an adaptive sampling approach for multiple testing which aims to
maximize statistical power while ensuring anytime false discovery control. We
consider $n$ distributions whose means are partitioned by whether they are
below or equal to a baseline (nulls), versus above the baseline (actual
positives). In addition, each distribution can be sequentially and repeatedly
sampled. Inspired by the multi-armed bandit literature, we provide an algorithm
that takes as few samples as possible to exceed a target true positive
proportion (i.e. proportion of actual positives discovered) while giving
anytime control of the false discovery proportion (nulls predicted as actual
positives). Our sample complexity results match known information theoretic
lower bounds and through simulations we show a substantial performance
improvement over uniform sampling and an adaptive elimination style algorithm.
Given the simplicity of the approach, and its sample efficiency, the method has
promise for wide adoption in the biological sciences, clinical testing for drug
discovery, and online A/B/n testing problems.
| stat |
Varying-coefficient stochastic differential equations with applications
in ecology | Stochastic differential equations (SDEs) are popular tools to analyse time
series data in many areas, such as mathematical finance, physics, and biology.
They provide a mechanistic description of the phenomeon of interest, and their
parameters often have a clear interpretation. These advantages come at the cost
of requiring a relatively simple model specification. We propose a flexible
model for SDEs with time-varying dynamics where the parameters of the process
are non-parametric functions of covariates, similar to generalized additive
models. Combining the SDEs and non-parametric approaches allows for the SDE to
capture more detailed, non-stationary, features of the data-generating process.
We present a computationally efficient method of approximate inference, where
the SDE parameters can vary according to fixed covariate effects, random
effects, or basis-penalty smoothing splines. We demonstrate the versatility and
utility of this approach with three applications in ecology, where there is
often a modelling trade-off between interpretability and flexibility.
| stat |
Poisson PCA: Poisson Measurement Error corrected PCA, with Application
to Microbiome Data | In this paper, we study the problem of computing a Principal Component
Analysis of data affected by Poisson noise. We assume samples are drawn from
independent Poisson distributions. We want to estimate principle components of
a fixed transformation of the latent Poisson means. Our motivating example is
microbiome data, though the methods apply to many other situations. We develop
a semiparametric approach to correct the bias of variance estimators, both for
untransformed and transformed (with particular attention to log-transformation)
Poisson means. Furthermore, we incorporate methods for correcting different
exposure or sequencing depth in the data. In addition to identifying the
principal components, we also address the non-trivial problem of computing the
principal scores in this semiparametric framework. Most previous approaches
tend to take a more parametric line. For example the Poisson-log-normal (PLN)
model, approach. We compare our method with the PLN approach and find that our
method is better at identifying the main principal components of the latent
log-transformed Poisson means, and as a further major advantage, takes far less
time to compute. Comparing methods on real data, we see that our method also
appears to be more robust to outliers than the parametric method.
| stat |
The Tilted Beta Binomial Linear Regression Model: a Bayesian Approach | This paper proposes new linear regression models to deal with overdispersed
binomial datasets. These new models, called tilted beta binomial regression
models, are defined from the tilted beta binomial distribution, proposed
assuming that the parameter of the binomial distribution follows a tilted beta
distribution. As a particular case of this regression models, we propose the
beta rectangular binomial regression models, defined from the binomial
distribution assuming that their parameters follow a beta rectangular
distribution. These new linear regression models, defined assuming that the
parameters of these new distributions follow regression structures, are fitted
applying Bayesian methods and using the OpenBUGS software. The proposed
regression models are fitted to an overdispersed binomial dataset of the number
of seeds that germinate depending on the type of chosen seed androot.
| stat |
Characteristic and Necessary Minutiae in Fingerprints | Fingerprints feature a ridge pattern with moderately varying ridge frequency
(RF), following an orientation field (OF), which usually features some
singularities. Additionally at some points, called minutiae, ridge lines end or
fork and this point pattern is usually used for fingerprint identification and
authentication. Whenever the OF features divergent ridge lines (e.g.\ near
singularities), a nearly constant RF necessitates the generation of more ridge
lines, originating at minutiae. We call these the necessary minutiae. It turns
out that fingerprints feature additional minutiae which occur at rather
arbitrary locations. We call these the random minutiae or, since they may
convey fingerprint individuality beyond the OF, the characteristic minutiae. In
consequence, the minutiae point pattern is assumed to be a realization of the
superposition of two stochastic point processes: a Strauss point process (whose
activity function is given by the divergence field) with an additional hard
core, and a homogeneous Poisson point process, modelling the necessary and the
characteristic minutiae, respectively. We perform Bayesian inference using an
MCMC-based minutiae separating algorithm (MiSeal). In simulations, it provides
good mixing and good estimation of underlying parameters. In application to
fingerprints, we can separate the two minutiae patterns and verify by example
of two different prints with similar OF that characteristic minutiae convey
fingerprint individuality.
| stat |
Improved estimators in beta prime regression models | In this paper, we consider the beta prime regression model recently proposed
by \cite{bour18}, which is tailored to situations where the response is
continuous and restricted to the positive real line with skewed and long tails
and the regression structure involves regressors and unknown parameters. We
consider two different strategies of bias correction of the maximum-likelihood
estimators for the parameters that index the model. In particular, we discuss
bias-corrected estimators for the mean and the dispersion parameters of the
model. Furthermore, as an alternative to the two analytically bias-corrected
estimators discussed, we consider a bias correction mechanism based on the
parametric bootstrap. The numerical results show that the bias correction
scheme yields nearly unbiased estimates. An example with real data is presented
and discussed.
| stat |
Optimal designs for third-order interactions in paired comparison
experiments | It is shown how by not losing information on higher order interactions,
optimal paired comparison designs involving alternatives of either full or
partial profiles to reduce information overload as frequently encountered in
applications can be constructed which enable identification of main effects up
to third-order interactions when all attributes have general common number of
levels.
| stat |
A Max-Correlation White Noise Test for Weakly Dependent Time Series | This paper presents a bootstrapped p-value white noise test based on the
maximum correlation, for a time series that may be weakly dependent under the
null hypothesis. The time series may be prefiltered residuals. The test
statistic is a normalized weighted maximum sample correlation, where the
maximum lag increases at a rate slower than the sample size. We only require
uncorrelatedness under the null hypothesis, along with a moment contraction
dependence property that includes mixing and non-mixing sequences. We show
Shao's (2011) dependent wild bootstrap is valid for a much larger class of
processes than originally considered. It is also valid for residuals from a
general class of parametric models as long as the bootstrap is applied to a
first order expansion of the sample correlation. We prove the bootstrap
validity without exploiting extreme value theory (standard in the literature)
or recent Gaussian approximation theory. Finally, we extend Escanciano and
Lobato's (2009) automatic maximum lag selection to our setting with an
unbounded choice set, and find it works strikingly well in controlled
experiments. Our proposed test achieves accurate size under various white noise
null hypotheses and high power under various alternative hypotheses including
distant serial dependence.
| stat |
Some New Results for Poisson Binomial Models | We consider a problem of ecological inference, in which individual-level
covariates are known, but labeled data is available only at the aggregate
level. The intended application is modeling voter preferences in elections. In
Rosenman and Viswanathan (2018), we proposed modeling individual voter
probabilities via a logistic regression, and posing the problem as a maximum
likelihood estimation for the parameter vector beta. The likelihood is a
Poisson binomial, the distribution of the sum of independent but not
identically distributed Bernoulli variables, though we approximate it with a
heteroscedastic Gaussian for computational efficiency. Here, we extend the
prior work by proving results about the existence of the MLE and the curvature
of this likelihood, which is not log-concave in general. We further demonstrate
the utility of our method on a real data example. Using data on voters in
Morris County, NJ, we demonstrate that our approach outperforms other
ecological inference methods in predicting a related, but known outcome:
whether an individual votes.
| stat |
Learning Clique Forests | We propose a topological learning algorithm for the estimation of the
conditional dependency structure of large sets of random variables from sparse
and noisy data. The algorithm, named Maximally Filtered Clique Forest (MFCF),
produces a clique forest and an associated Markov Random Field (MRF) by
generalising Prim's minimum spanning tree algorithm. To the best of our
knowledge, the MFCF presents three elements of novelty with respect to existing
structure learning approaches. The first is the repeated application of a local
topological move, the clique expansion, that preserves the decomposability of
the underlying graph. Through this move the decomposability and calculation of
scores is performed incrementally at the variable (rather than edge) level, and
this provides better computational performance and an intuitive application of
multivariate statistical tests. The second is the capability to accommodate a
variety of score functions and, while this paper is focused on multivariate
normal distributions, it can be directly generalised to different types of
statistics. Finally, the third is the variable range of allowed clique sizes
which is an adjustable topological constraint that acts as a topological
penalizer providing a way to tackle sparsity at $l_0$ semi-norm level; this
allows a clean decoupling of structure learning and parameter estimation. The
MFCF produces a representation of the clique forest, together with a perfect
ordering of the cliques and a perfect elimination ordering for the vertices. As
an example we propose an application to covariance selection models and we show
that the MCFC outperforms the Graphical Lasso for a number of classes of
matrices.
| stat |
Estimation of parameters of the logistic exponential distribution under
progressive type-I hybrid censored sample | The paper addresses the problem of estimation of the model parameters of the
logistic exponential distribution based on progressive type-I hybrid censored
sample. The maximum likelihood estimates are obtained and computed numerically
using Newton-Raphson method. Further, the Bayes estimates are derived under
squared error, LINEX and generalized entropy loss functions. Two types
(independent and bivariate) of prior distributions are considered for the
purpose of Bayesian estimation. It is seen that the Bayes estimates are not of
explicit forms.Thus, Lindley's approximation technique is employed to get
approximate Bayes estimates. Interval estimates of the parameters based on
normal approximate of the maximum likelihood estimates and normal approximation
of the log-transformed maximum likelihood estimates are constructed. The
highest posterior density credible intervals are obtained by using the
importance sampling method. Furthermore, numerical computations are reported to
review some of the results obtained in the paper. A real life dataset is
considered for the purpose of illustrations.
| stat |
A (tight) upper bound for the length of confidence intervals with
conditional coverage | We show that two popular selective inference procedures, namely data carving
(Fithian et al., 2017) and selection with a randomized response (Tian et al.,
2018b), when combined with the polyhedral method (Lee et al., 2016), result in
confidence intervals whose length is bounded. This contrasts results for
confidence intervals based on the polyhedral method alone, whose expected
length is typically infinite (Kivaranovic and Leeb, 2020). Moreover, we show
that these two procedures always dominate corresponding sample-splitting
methods in terms of interval length.
| stat |
Consistent Robust Adversarial Prediction for General Multiclass
Classification | We propose a robust adversarial prediction framework for general multiclass
classification. Our method seeks predictive distributions that robustly
optimize non-convex and non-continuous multiclass loss metrics against the
worst-case conditional label distributions (the adversarial distributions) that
(approximately) match the statistics of the training data. Although the
optimized loss metrics are non-convex and non-continuous, the dual formulation
of the framework is a convex optimization problem that can be recast as a risk
minimization model with a prescribed convex surrogate loss we call the
adversarial surrogate loss. We show that the adversarial surrogate losses fill
an existing gap in surrogate loss construction for general multiclass
classification problems, by simultaneously aligning better with the original
multiclass loss, guaranteeing Fisher consistency, enabling a way to incorporate
rich feature spaces via the kernel trick, and providing competitive performance
in practice.
| stat |
Colombian Women's Life Patterns: A Multivariate Density Regression
Approach | Women in Colombia face difficulties related to the patriarchal traits of
their societies and well-known conflict afflicting the country since 1948. In
this critical context, our aim is to study the relationship between baseline
socio-demographic factors and variables associated to fertility, partnership
patterns, and work activity. To best exploit the explanatory structure, we
propose a Bayesian multivariate density regression model, which can accommodate
mixed responses with censored, constrained, and binary traits. The flexible
nature of the models allows for nonlinear regression functions and non-standard
features in the errors, such as asymmetry or multi-modality. The model has
interpretable covariate-dependent weights constructed through normalization,
allowing for combinations of categorical and continuous covariates.
Computational difficulties for inference are overcome through an adaptive
truncation algorithm combining adaptive Metropolis-Hastings and sequential
Monte Carlo to create a sequence of automatically truncated posterior mixtures.
For our study on Colombian women's life patterns, a variety of quantities are
visualised and described, and in particular, our findings highlight the
detrimental impact of family violence on women's choices and behaviors.
| stat |
Framework for Inferring Following Strategies from Time Series of
Movement Data | How do groups of individuals achieve consensus in movement decisions? Do
individuals follow their friends, the one predetermined leader, or whomever
just happens to be nearby? To address these questions computationally, we
formalize "Coordination Strategy Inference Problem". In this setting, a group
of multiple individuals moves in a coordinated manner towards a target path.
Each individual uses a specific strategy to follow others (e.g. nearest
neighbors, pre-defined leaders, preferred friends). Given a set of time series
that includes coordinated movement and a set of candidate strategies as inputs,
we provide the first methodology (to the best of our knowledge) to infer
whether each individual uses local-agreement-system or dictatorship-like
strategy to achieve movement coordination at the group level. We evaluate and
demonstrate the performance of the proposed framework by predicting the
direction of movement of an individual in a group in both simulated datasets as
well as two real-world datasets: a school of fish and a troop of baboons.
Moreover, since there is no prior methodology for inferring individual-level
strategies, we compare our framework with the state-of-the-art approach for the
task of classification of group-level-coordination models. The results show
that our approach is highly accurate in inferring the correct strategy in
simulated datasets even in complicated mixed strategy settings, which no
existing method can infer. In the task of classification of
group-level-coordination models, our framework performs better than the
state-of-the-art approach in all datasets. Animal data experiments show that
fish, as expected, follow their neighbors, while baboons have a preference to
follow specific individuals. Our methodology generalizes to arbitrary time
series data of real numbers, beyond movement data.
| stat |
Improved Precision and Recall Metric for Assessing Generative Models | The ability to automatically estimate the quality and coverage of the samples
produced by a generative model is a vital requirement for driving algorithm
research. We present an evaluation metric that can separately and reliably
measure both of these aspects in image generation tasks by forming explicit,
non-parametric representations of the manifolds of real and generated data. We
demonstrate the effectiveness of our metric in StyleGAN and BigGAN by providing
several illustrative examples where existing metrics yield uninformative or
contradictory results. Furthermore, we analyze multiple design variants of
StyleGAN to better understand the relationships between the model architecture,
training methods, and the properties of the resulting sample distribution. In
the process, we identify new variants that improve the state-of-the-art. We
also perform the first principled analysis of truncation methods and identify
an improved method. Finally, we extend our metric to estimate the perceptual
quality of individual samples, and use this to study latent space
interpolations.
| stat |
lspartition: Partitioning-Based Least Squares Regression | Nonparametric partitioning-based least squares regression is an important
tool in empirical work. Common examples include regressions based on splines,
wavelets, and piecewise polynomials. This article discusses the main
methodological and numerical features of the R software package lspartition,
which implements modern estimation and inference results for partitioning-based
least squares (series) regression estimation. This article discusses the main
methodological and numerical features of the R software package lspartition,
which implements results for partitioning-based least squares (series)
regression estimation and inference from Cattaneo and Farrell (2013) and
Cattaneo, Farrell, and Feng (2019). These results cover the multivariate
regression function as well as its derivatives. First, the package provides
data-driven methods to choose the number of partition knots optimally,
according to integrated mean squared error, yielding optimal point estimation.
Second, robust bias correction is implemented to combine this point estimator
with valid inference. Third, the package provides estimates and inference for
the unknown function both pointwise and uniformly in the conditioning
variables. In particular, valid confidence bands are provided. Finally, an
extension to two-sample analysis is developed, which can be used in
treatment-control comparisons and related problems
| stat |
Deep dynamic modeling with just two time points: Can we still allow for
individual trajectories? | Longitudinal biomedical data are often characterized by a sparse time grid
and individual-specific development patterns. Specifically, in epidemiological
cohort studies and clinical registries we are facing the question of what can
be learned from the data in an early phase of the study, when only a baseline
characterization and one follow-up measurement are available. Inspired by
recent advances that allow to combine deep learning with dynamic modeling, we
investigate whether such approaches can be useful for uncovering complex
structure, in particular for an extreme small data setting with only two
observations time points for each individual. Irregular spacing in time could
then be used to gain more information on individual dynamics by leveraging
similarity of individuals. We provide a brief overview of how variational
autoencoders (VAEs), as a deep learning approach, can be linked to ordinary
differential equations (ODEs) for dynamic modeling, and then specifically
investigate the feasibility of such an approach that infers individual-specific
latent trajectories by including regularity assumptions and individuals'
similarity. We also provide a description of this deep learning approach as a
filtering task to give a statistical perspective. Using simulated data, we show
to what extent the approach can recover individual trajectories from ODE
systems with two and four unknown parameters and infer groups of individuals
with similar trajectories, and where it breaks down. The results show that such
dynamic deep learning approaches can be useful even in extreme small data
settings, but need to be carefully adapted.
| stat |
Generating Digital Twins with Multiple Sclerosis Using Probabilistic
Neural Networks | Multiple Sclerosis (MS) is a neurodegenerative disorder characterized by a
complex set of clinical assessments. We use an unsupervised machine learning
model called a Conditional Restricted Boltzmann Machine (CRBM) to learn the
relationships between covariates commonly used to characterize subjects and
their disease progression in MS clinical trials. A CRBM is capable of
generating digital twins, which are simulated subjects having the same baseline
data as actual subjects. Digital twins allow for subject-level statistical
analyses of disease progression. The CRBM is trained using data from 2395
subjects enrolled in the placebo arms of clinical trials across the three
primary subtypes of MS. We discuss how CRBMs are trained and show that digital
twins generated by the model are statistically indistinguishable from their
actual subject counterparts along a number of measures.
| stat |
Exponential Random Graph models for Little Networks | Statistical models for social networks have enabled researchers to study
complex social phenomena that give rise to observed patterns of relationships
among social actors and to gain a rich understanding of the interdependent
nature of social ties and actors. Much of this research has focused on social
networks within medium to large social groups. To date, these advances in
statistical models for social networks, and in particular, of
Exponential-Family Random Graph Models (ERGMS), have rarely been applied to the
study of small networks, despite small network data in teams, families, and
personal networks being common in many fields. In this paper, we revisit the
estimation of ERGMs for small networks and propose using exhaustive enumeration
when possible. We developed an R package that implements the estimation of
pooled ERGMs for small networks using Maximum Likelihood Estimation (MLE),
called "ergmito". Based on the results of an extensive simulation study to
assess the properties of the MLE estimator, we conclude that there are several
benefits of direct MLE estimation compared to approximate methods and that this
creates opportunities for valuable methodological innovations that can be
applied to modeling social networks with ERGMs.
| stat |
Non-parametric Archimedean generator estimation with implications for
multiple testing | In multiple testing, the family-wise error rate can be bounded under some
conditions by the copula of the test statistics. Assuming that this copula is
Archimedean, we consider two non-parametric Archimedean generator estimators.
More specifically, we use the non-parametric estimator from Genest et al.
(2011) and a slight modification thereof. In simulations, we compare the
resulting multiple tests with the Bonferroni test and the multiple test derived
from the true generator as baselines.
| stat |
High Dimensional Level Set Estimation with Bayesian Neural Network | Level Set Estimation (LSE) is an important problem with applications in
various fields such as material design, biotechnology, machine operational
testing, etc. Existing techniques suffer from the scalability issue, that is,
these methods do not work well with high dimensional inputs. This paper
proposes novel methods to solve the high dimensional LSE problems using
Bayesian Neural Networks. In particular, we consider two types of LSE problems:
(1) \textit{explicit} LSE problem where the threshold level is a fixed
user-specified value, and, (2) \textit{implicit} LSE problem where the
threshold level is defined as a percentage of the (unknown) maximum of the
objective function. For each problem, we derive the corresponding theoretic
information based acquisition function to sample the data points so as to
maximally increase the level set accuracy. Furthermore, we also analyse the
theoretical time complexity of our proposed acquisition functions, and suggest
a practical methodology to efficiently tune the network hyper-parameters to
achieve high model accuracy. Numerical experiments on both synthetic and
real-world datasets show that our proposed method can achieve better results
compared to existing state-of-the-art approaches.
| stat |
Prediction of Bayesian Intervals for Tropical Storms | Building on recent research for prediction of hurricane trajectories using
recurrent neural networks (RNNs), we have developed improved methods and
generalized the approach to predict Bayesian intervals in addition to simple
point estimates. Tropical storms are capable of causing severe damage, so
accurately predicting their trajectories can bring significant benefits to
cities and lives, especially as they grow more intense due to climate change
effects. By implementing the Bayesian interval using dropout in an RNN, we
improve the actionability of the predictions, for example by estimating the
areas to evacuate in the landfall region. We used an RNN to predict the
trajectory of the storms at 6-hour intervals. We used latitude, longitude,
windspeed, and pressure features from a Statistical Hurricane Intensity
Prediction Scheme (SHIPS) dataset of about 500 tropical storms in the Atlantic
Ocean. Our results show how neural network dropout values affect predictions
and intervals.
| stat |
Recruitment prediction for multi-centre clinical trials based on a
hierarchical Poisson-gamma model: asymptotic analysis and improved intervals | We analyse predictions of future recruitment to a multi-centre clinical trial
based on a maximum-likelihood fitting of a commonly used hierarchical
Poisson-Gamma model for recruitments at individual centres. We consider the
asymptotic accuracy of quantile predictions in the limit as the number of
recruitment centres grows large and find that, in an important sense, the
accuracy of the quantiles does not improve as the number of centres increases.
When predicting the number of further recruits in an additional time period,
the accuracy degrades as the ratio of the additional time to the census time
increases, whereas when predicting the amount of additional time to recruit a
further $n^+_\bullet$ patients, the accuracy degrades as the ratio of
$n^+_\bullet$ to the number recruited up to the census period increases. Our
analysis suggests an improved quantile predictor. Simulation studies verify
that the predicted pattern holds for typical recruitment scenarios in clinical
trials and verify the much improved coverage properties of prediction intervals
obtained from our quantile predictor. In the process of extending the
applicability of our methodology, we show that in terms of the accuracy of all
integer moments it is always better to approximate the sum of independent gamma
random variables by a single gamma random variable matched on the first two
moments than by the moment-matched Gaussian available from the central limit
theorem.
| stat |
A Mean Field Games model for finite mixtures of Bernoulli and
Categorical distributions | Finite mixture models are an important tool in the statistical analysis of
data, for example in data clustering. The optimal parameters of a mixture model
are usually computed by maximizing the log-likelihood functional via the
Expectation-Maximization algorithm. We propose an alternative approach based on
the theory of Mean Field Games, a class of differential games with an infinite
number of agents. We show that the solution of a finite state space
multi-population Mean Field Games system characterizes the critical points of
the log-likelihood functional for a Bernoulli mixture. The approach is then
generalized to mixture models of categorical distributions. Hence, the Mean
Field Games approach provides a method to compute the parameters of the mixture
model, and we show its application to some standard examples in cluster
analysis.
| stat |
Multiple membership multilevel models | Multiple membership multilevel models are an extension of standard multilevel
models for non-hierarchical data that have multiple membership structures.
Traditional multilevel models involve hierarchical data structures whereby
lower-level units such as students are nested within higher-level units such as
schools and where these higher-level units may in turn be nested within further
groupings or clusters such as school districts, regions, and countries. With
hierarchical data structures, there is an exact nesting of each lower-level
unit in one and only one higher-level unit. For example, each student attends
one school, each school is located within one school district, and so on.
However, social reality is more complicated than this, and so social and
behavioural data often do not follow pure or strict hierarchies. Two types of
non-hierarchical data structures which often appear in practice are
cross-classified and multiple membership structures. In this article, we
describe multiple membership data structures and multiple membership models
which can be used to analyse them.
| stat |
Robust Uniform Inference for Quantile Treatment Effects in Regression
Discontinuity Designs | The practical importance of inference with robustness against large
bandwidths for causal effects in regression discontinuity and kink designs is
widely recognized. Existing robust methods cover many cases, but do not handle
uniform inference for CDF and quantile processes in fuzzy designs, despite its
use in the recent literature in empirical microeconomics. In this light, this
paper extends the literature by developing a unified framework of inference
with robustness against large bandwidths that applies to uniform inference for
quantile treatment effects in fuzzy designs, as well as all the other cases of
sharp/fuzzy mean/quantile regression discontinuity/kink designs. We present
Monte Carlo simulation studies and an empirical application for evaluations of
the Oklahoma pre-K program.
| stat |
Dynamic Assortment Selection under the Nested Logit Models | We study a stylized dynamic assortment planning problem during a selling
season of finite length $T$. At each time period, the seller offers an arriving
customer an assortment of substitutable products and the customer makes the
purchase among offered products according to a discrete choice model. The goal
of the seller is to maximize the expected revenue, or equivalently, to minimize
the worst-case expected regret. One key challenge is that utilities of products
are unknown to the seller and need to be learned. Although the dynamic
assortment planning problem has received increasing attention in revenue
management, most existing work is based on the multinomial logit choice models
(MNL). In this paper, we study the problem of dynamic assortment planning under
a more general choice model -- the nested logit model, which models
hierarchical choice behavior and is ``the most widely used member of the GEV
(generalized extreme value) family''. By leveraging the revenue-ordered
structure of the optimal assortment within each nest, we develop a novel upper
confidence bound (UCB) policy with an aggregated estimation scheme. Our policy
simultaneously learns customers' choice behavior and makes dynamic decisions on
assortments based on the current knowledge. It achieves the accumulated regret
at the order of $\tilde{O}(\sqrt{MNT})$, where $M$ is the number of nests and
$N$ is the number of products in each nest. We further provide a lower bound
result of $\Omega(\sqrt{MT})$, which shows the near optimality of the upper
bound when $T$ is much larger than $M$ and $N$. When the number of items per
nest $N$ is large, we further provide a discretization heuristic for better
performance of our algorithm. Numerical results are presented to demonstrate
the empirical performance of our proposed algorithms.
| stat |
Bayesian Nonparametric Estimation for Point Processes with Spatial
Homogeneity: A Spatial Analysis of NBA Shot Locations | Basketball shot location data provide valuable summary information regarding
players to coaches, sports analysts, fans, statisticians, as well as players
themselves. Represented by spatial points, such data are naturally analyzed
with spatial point process models. We present a novel nonparametric Bayesian
method for learning the underlying intensity surface built upon a combination
of Dirichlet process and Markov random field. Our method has the advantage of
effectively encouraging local spatial homogeneity when estimating a globally
heterogeneous intensity surface. Posterior inferences are performed with an
efficient Markov chain Monte Carlo (MCMC) algorithm. Simulation studies show
that the inferences are accurate and that the method is superior compared to
the competing methods. Application to the shot location data of $20$
representative NBA players in the 2017-2018 regular season offers interesting
insights about the shooting patterns of these players. A comparison against the
competing method shows that the proposed method can effectively incorporate
spatial contiguity into the estimation of intensity surfaces.
| stat |
Location-adjusted Wald statistics for scalar parameters | Inference about a scalar parameter of interest is a core statistical task
that has attracted immense research in statistics. The Wald statistic is a
prime candidate for the task, on the grounds of the asymptotic validity of the
standard normal approximation to its finite-sample distribution, simplicity and
low computational cost. It is well known, though, that this normal
approximation can be inadequate, especially when the sample size is small or
moderate relative to the number of parameters. A novel, algebraic adjustment to
the Wald statistic is proposed, delivering significant improvements in
inferential performance with only small implementation and computational
overhead, predominantly due to additional matrix multiplications. The Wald
statistic is viewed as an estimate of a transformation of the model parameters
and is appropriately adjusted, using either maximum likelihood or reduced-bias
estimators, bringing its expectation asymptotically closer to zero. The
location adjustment depends on the expected information, an approximation to
the bias of the estimator, and the derivatives of the transformation, which are
all either readily available or easily obtainable in standard software for a
wealth of models. An algorithm for the implementation of the location-adjusted
Wald statistics in general models is provided, as well as a bootstrap scheme
for the further scale correction of the location-adjusted statistic. Ample
analytical and numerical evidence is presented for the adoption of the
location-adjusted statistic in prominent modelling settings, including
inference about log-odds and binomial proportions, logistic regression in the
presence of nuisance parameters, beta regression, and gamma regression. The
location-adjusted Wald statistics are used for the construction of significance
maps for the analysis of multiple sclerosis lesions from MRI data.
| stat |
Random Vector Functional Link Networks for Function Approximation on
Manifolds | The learning speed of feed-forward neural networks is notoriously slow and
has presented a bottleneck in deep learning applications for several decades.
For instance, gradient-based learning algorithms, which are used extensively to
train neural networks, tend to work slowly when all of the network parameters
must be iteratively tuned. To counter this, both researchers and practitioners
have tried introducing randomness to reduce the learning requirement. Based on
the original construction of Igelnik and Pao, single layer neural-networks with
random input-to-hidden layer weights and biases have seen success in practice,
but the necessary theoretical justification is lacking. In this paper, we begin
to fill this theoretical gap. We provide a (corrected) rigorous proof that the
Igelnik and Pao construction is a universal approximator for continuous
functions on compact domains, with approximation error decaying asymptotically
like $O(1/\sqrt{n})$ for the number $n$ of network nodes. We then extend this
result to the non-asymptotic setting, proving that one can achieve any desired
approximation error with high probability provided $n$ is sufficiently large.
We further adapt this randomized neural network architecture to approximate
functions on smooth, compact submanifolds of Euclidean space, providing
theoretical guarantees in both the asymptotic and non-asymptotic forms.
Finally, we illustrate our results on manifolds with numerical experiments.
| stat |
Extragradient with player sampling for faster Nash equilibrium finding | Data-driven modeling increasingly requires to find a Nash equilibrium in
multi-player games, e.g. when training GANs. In this paper, we analyse a new
extra-gradient method for Nash equilibrium finding, that performs gradient
extrapolations and updates on a random subset of players at each iteration.
This approach provably exhibits a better rate of convergence than full
extra-gradient for non-smooth convex games with noisy gradient oracle. We
propose an additional variance reduction mechanism to obtain speed-ups in
smooth convex games. Our approach makes extrapolation amenable to massive
multiplayer settings, and brings empirical speed-ups, in particular when using
a heuristic cyclic sampling scheme. Most importantly, it allows to train faster
and better GANs and mixtures of GANs.
| stat |
Heterogeneous item populations across individuals: Consequences for the
factor model, item inter-correlations, and scale validity | The paper is devoted to the consequences of blind random selection of items
from different item populations that might be based on completely uncorrelated
factors for item inter-correlations and corresponding factor loadings. Based on
the model of essentially parallel measurements, we explore the consequences of
presenting items from different populations across individuals and items from
identical populations within each individual for the factor model and item
inter-correlations in the total population of individuals. Moreover, we explore
the consequences of presenting items from different as well as identical item
populations across and within individuals. We show that correlations can be
substantial in the total population of individuals even when -- in
subpopulations of individuals -- items are drawn from populations with
uncorrelated factors. In order to address this challenge for the validity of a
scale, we propose a method that helps to detect whether item inter-correlations
result from different item populations in different subpopulations of
individuals and evaluate the method by means of a simulation study. Based on
the analytical results and on results from a simulation study, we provide
recommendations for the detection of subpopulations of individuals responding
to items from different item populations.
| stat |
Accounting for Uncertainty During a Pandemic | We discuss several issues of statistical design, data collection, analysis,
communication, and decision making that have arisen in recent and ongoing
coronavirus studies, focusing on tools for assessment and propagation of
uncertainty. This paper does not purport to be a comprehensive survey of the
research literature; rather, we use examples to illustrate statistical points
that we think are important.
| stat |
Statistical methods research done as science rather than mathematics | This paper is about how we study statistical methods. As an example, it uses
the random regressions model, in which the intercept and slope of
cluster-specific regression lines are modeled as a bivariate random effect.
Maximizing this model's restricted likelihood often gives a boundary value for
the random effect correlation or variances. We argue that this is a problem;
that it is a problem because our discipline has little understanding of how
contemporary models and methods map data to inferential summaries; that we lack
such understanding, even for models as simple as this, because of a
near-exclusive reliance on mathematics as a means of understanding; and that
math alone is no longer sufficient. We then argue that as a discipline, we can
and should break open our black-box methods by mimicking the five steps that
molecular biologists commonly use to break open Nature's black boxes: design a
simple model system, formulate hypotheses using that system, test them in
experiments on that system, iterate as needed to reformulate and test
hypotheses, and finally test the results in an "in vivo" system. We demonstrate
this by identifying conditions under which the random-regressions restricted
likelihood is likely to be maximized at a boundary value. Resistance to this
approach seems to arise from a view that it lacks the certainty or intellectual
heft of mathematics, perhaps because simulation experiments in our literature
rarely do more than measure a new method's operating characteristics in a small
range of situations. We argue that such work can make useful contributions
including, as in molecular biology, the findings themselves and sometimes the
designs used in the five steps; that these contributions have as much practical
value as mathematical results; and that therefore they merit publication as
much as the mathematical results our discipline esteems so highly.
| stat |
The Reduced PC-Algorithm: Improved Causal Structure Learning in Large
Random Networks | We consider the task of estimating a high-dimensional directed acyclic graph,
given observations from a linear structural equation model with arbitrary noise
distribution. By exploiting properties of common random graphs, we develop a
new algorithm that requires conditioning only on small sets of variables. The
proposed algorithm, which is essentially a modified version of the
PC-Algorithm, offers significant gains in both computational complexity and
estimation accuracy. In particular, it results in more efficient and accurate
estimation in large networks containing hub nodes, which are common in
biological systems. We prove the consistency of the proposed algorithm, and
show that it also requires a less stringent faithfulness assumption than the
PC-Algorithm. Simulations in low and high-dimensional settings are used to
illustrate these findings. An application to gene expression data suggests that
the proposed algorithm can identify a greater number of clinically relevant
genes than current methods.
| stat |
Nonparametric Gaussian Mixture Models for the Multi-Armed Contextual
Bandit | We here adopt Bayesian nonparametric mixture models to extend multi-armed
bandits in general, and Thompson sampling in particular, to scenarios where
there is reward model uncertainty. In the stochastic multi-armed bandit, where
an agent must learn a policy that maximizes long term payoff, the reward for
the selected action is generated from an unknown distribution. Thompson
sampling is a generative and interpretable multi-armed bandit algorithm that
has been shown both to perform well in practice, and to enjoy optimality
properties for certain reward functions. Nevertheless, Thompson sampling
requires knowledge of the true reward model, for calculation of expected
rewards and sampling from its parameter posterior. In this work, we extend
Thompson sampling to complex scenarios where there is model uncertainty, by
adopting a very flexible set of reward distributions: Bayesian nonparametric
Gaussian mixture models. The generative process of Bayesian nonparametric
mixtures naturally aligns with the Bayesian modeling of multi-armed bandits:
the nonparametric model autonomously determines its complexity as new rewards
are observed for the played arms. By characterizing each arm's reward
distribution with independent nonparametric mixture models, the proposed method
sequentially learns the model that best approximates the true underlying reward
distribution, achieving successful performance in complex -- not in the
exponential family -- bandits. Our contribution is valuable for practical
scenarios, as it avoids stringent case-by-case model specifications and
hyperparameter tuning, yet attains reduced regret in diverse bandit settings.
| stat |
Sequential Likelihood-Free Inference with Implicit Surrogate Proposal | Bayesian inference without the access of likelihood, or likelihood-free
inference, has been a key research topic in simulations, to yield a more
realistic generation result. Recent likelihood-free inference updates an
approximate posterior sequentially with the dataset of the cumulative
simulation input-output pairs over inference rounds. Therefore, the dataset is
gathered through the iterative simulations with sampled inputs from a proposal
distribution by MCMC, which becomes the key of inference quality in this
sequential framework. This paper introduces a new proposal modeling, named as
Implicit Surrogate Proposal (ISP), to generate a cumulated dataset with further
sample efficiency. ISP constructs the cumulative dataset in the most diverse
way by drawing i.i.d samples via a feed-forward fashion, so the posterior
inference does not suffer from the disadvantages of MCMC caused by its
non-i.i.d nature, such as auto-correlation and slow mixing. We analyze the
convergence property of ISP in both theoretical and empirical aspects to
guarantee that ISP provides an asymptotically exact sampler. We demonstrate
that ISP outperforms the baseline inference algorithms on simulations with
multi-modal posteriors.
| stat |
A Novel Framework for Online Supervised Learning with Feature Selection | Current online learning methods suffer issues such as lower convergence rates
and limited capability to recover the support of the true features compared to
their offline counterparts. In this paper, we present a novel framework for
online learning based on running averages and introduce a series of online
versions of some popular existing offline methods such as Elastic Net, Minimax
Concave Penalty and Feature Selection with Annealing. The framework can handle
an arbitrarily large number of observations with the restriction that the data
dimension is not too large, e.g. p<50,000. We prove the equivalence between our
online methods and their offline counterparts and give theoretical true feature
recovery and convergence guarantees for some of them. In contrast to the
existing online methods, the proposed methods can extract models with any
desired sparsity level at any time. Numerical experiments indicate that our new
methods enjoy high accuracy of true feature recovery and a fast convergence
rate, compared with standard online and offline algorithms. We also show how
the running averages framework can be used for model adaptation in the presence
of model drift. Finally, we present some applications to large datasets where
again the proposed framework shows competitive results compared to popular
online and offline algorithms.
| stat |
Slamming the sham: A Bayesian model for adaptive adjustment with noisy
control data | It is not always clear how to adjust for control data in causal inference,
balancing the goals of reducing bias and variance. We show how, in a setting
with repeated experiments, Bayesian hierarchical modeling yields an adaptive
procedure that uses the data to determine how much adjustment to perform. The
result is a novel analysis with increased statistical efficiency compared to
the default analysis based on difference estimates. We demonstrate this
procedure on two real examples, as well as on a series of simulated datasets.
We show that the increased efficiency can have real-world consequences in terms
of the conclusions that can be drawn from the experiments. We also discuss the
relevance of this work to causal inference and statistical design and analysis
more generally.
| stat |
An exploratory time series analysis of total deaths per month in Brazil
since 2015 | In this article, we investigate the historical series of the total number of
deaths per month in Brazil since 2015 using time series analysis techniques, in
order to assess whether the COVID-19 pandemic caused any change in the series'
generating mechanism. The results obtained so far indicate that there was no
statistical significant impact.
| stat |
On the convergence time of some non-reversible Markov chain Monte Carlo
methods | It is commonly admitted that non-reversible Markov chain Monte Carlo (MCMC)
algorithms usually yield more accurate MCMC estimators than their reversible
counterparts. In this note, we show that in addition to their variance
reduction effect, some non-reversible MCMC algorithms have also the undesirable
property to slow down the convergence of the Markov chain. This point, which
has been overlooked by the literature, has obvious practical implications. We
illustrate this phenomenon for different non-reversible versions of the
Metropolis-Hastings algorithm on several discrete state space examples and
discuss ways to mitigate the risk of a small asymptotic variance/slow
convergence scenario.
| stat |
Parametric bootstrapping in a generalized extreme value regression model
for binary response | Generalized extreme value (GEV) regression is often more adapted when we
investigate a relationship between a binary response variable $Y$ which
represents a rare event and potentiel predictors $\mathbf{X}$. In particular,
we use the quantile function of the GEV distribution as link function.
Bootstrapping assigns measures of accuracy (bias, variance, confidence
intervals, prediction error, test of hypothesis) to sample estimates. This
technique allows estimation of the sampling distribution of almost any
statistic using random sampling methods. Bootstrapping estimates the properties
of an estimator by measuring those properties when sampling from an
approximating distribution. In this paper, we fitted the generalized extreme
value regression model, then we performed parametric bootstrap method for
testing hupthesis, estimating confidence interval of parameters for generalized
extreme value regression model and a real data application.
| stat |
Interpretable machine learning: definitions, methods, and applications | Machine-learning models have demonstrated great success in learning complex
patterns that enable them to make predictions about unobserved data. In
addition to using models for prediction, the ability to interpret what a model
has learned is receiving an increasing amount of attention. However, this
increased focus has led to considerable confusion about the notion of
interpretability. In particular, it is unclear how the wide array of proposed
interpretation methods are related, and what common concepts can be used to
evaluate them.
We aim to address these concerns by defining interpretability in the context
of machine learning and introducing the Predictive, Descriptive, Relevant (PDR)
framework for discussing interpretations. The PDR framework provides three
overarching desiderata for evaluation: predictive accuracy, descriptive
accuracy and relevancy, with relevancy judged relative to a human audience.
Moreover, to help manage the deluge of interpretation methods, we introduce a
categorization of existing techniques into model-based and post-hoc categories,
with sub-groups including sparsity, modularity and simulatability. To
demonstrate how practitioners can use the PDR framework to evaluate and
understand interpretations, we provide numerous real-world examples. These
examples highlight the often under-appreciated role played by human audiences
in discussions of interpretability. Finally, based on our framework, we discuss
limitations of existing methods and directions for future work. We hope that
this work will provide a common vocabulary that will make it easier for both
practitioners and researchers to discuss and choose from the full range of
interpretation methods.
| stat |
CatBoostLSS -- An extension of CatBoost to probabilistic forecasting | We propose a new framework of CatBoost that predicts the entire conditional
distribution of a univariate response variable. In particular, CatBoostLSS
models all moments of a parametric distribution (i.e., mean, location, scale
and shape [LSS]) instead of the conditional mean only. Choosing from a wide
range of continuous, discrete and mixed discrete-continuous distributions,
modelling and predicting the entire conditional distribution greatly enhances
the flexibility of CatBoost, as it allows to gain insight into the data
generating process, as well as to create probabilistic forecasts from which
prediction intervals and quantiles of interest can be derived. We present both
a simulation study and real-world examples that demonstrate the benefits of our
approach.
| stat |
Vecchia-Laplace approximations of generalized Gaussian processes for big
non-Gaussian spatial data | Generalized Gaussian processes (GGPs) are highly flexible models that combine
latent GPs with potentially non-Gaussian likelihoods from the exponential
family. GGPs can be used in a variety of settings, including GP classification,
nonparametric count regression, modeling non-Gaussian spatial data, and
analyzing point patterns. However, inference for GGPs can be analytically
intractable, and large datasets pose computational challenges due to the
inversion of the GP covariance matrix. We propose a Vecchia-Laplace
approximation for GGPs, which combines a Laplace approximation to the
non-Gaussian likelihood with a computationally efficient Vecchia approximation
to the GP, resulting in a simple, general, scalable, and accurate methodology.
We provide numerical studies and comparisons on simulated and real spatial
data. Our methods are implemented in a freely available R package.
| stat |
Evaluation of a length-based method to estimate discard rate and the
effect of sampling size | The common fisheries policy aims at eliminating discarding which has been
part of fisheries for centuries. It is important to monitor the compliance with
the new regulations but estimating the discard rate is a challenging task,
especially where the practise is illegal. The aim of this study was to review a
length-based method that has been used to estimate the discard rate in
Icelandic waters and explore the effects of different monitoring schemes. The
length-based method estimates the minimum discard rate and the method of
bootstrapping can be used to determine the uncertainty of the estimate. This
study showed that the number of ships is the most important factor to consider
in order to decrease the uncertainty.
| stat |
High-Dimensional Inference for Cluster-Based Graphical Models | Motivated by modern applications in which one constructs graphical models
based on a very large number of features, this paper introduces a new class of
cluster-based graphical models, in which variable clustering is applied as an
initial step for reducing the dimension of the feature space. We employ model
assisted clustering, in which the clusters contain features that are similar to
the same unobserved latent variable. Two different cluster-based Gaussian
graphical models are considered: the latent variable graph, corresponding to
the graphical model associated with the unobserved latent variables, and the
cluster-average graph, corresponding to the vector of features averaged over
clusters. Our study reveals that likelihood based inference for the latent
graph, not analyzed previously, is analytically intractable. Our main
contribution is the development and analysis of alternative estimation and
inference strategies, for the precision matrix of an unobservable latent vector
$Z$. We replace the likelihood of the data by an appropriate class of empirical
risk functions, that can be specialized to the latent graphical model and to
the simpler, but under-analyzed, cluster-average graphical model. The
estimators thus derived can be used for inference on the graph structure, for
instance on edge strength or pattern recovery. Inference is based on the
asymptotic limits of the entry-wise estimates of the precision matrices
associated with the conditional independence graphs under consideration. While
taking the uncertainty induced by the clustering step into account, we
establish Berry-Esseen central limit theorems for the proposed estimators. It
is noteworthy that, although the clusters are estimated adaptively from the
data, the central limit theorems regarding the entries of the estimated graphs
are proved under the same conditions one would use if the clusters were
known....
| stat |
Binary Space Partitioning Forests | The Binary Space Partitioning~(BSP)-Tree process is proposed to produce
flexible 2-D partition structures which are originally used as a Bayesian
nonparametric prior for relational modelling. It can hardly be applied to other
learning tasks such as regression trees because extending the BSP-Tree process
to a higher dimensional space is nontrivial. This paper is the first attempt to
extend the BSP-Tree process to a d-dimensional (d>2) space. We propose to
generate a cutting hyperplane, which is assumed to be parallel to d-2
dimensions, to cut each node in the d-dimensional BSP-tree. By designing a
subtle strategy to sample two free dimensions from d dimensions, the extended
BSP-Tree process can inherit the essential self-consistency property from the
original version. Based on the extended BSP-Tree process, an ensemble model,
which is named the BSP-Forest, is further developed for regression tasks.
Thanks to the retained self-consistency property, we can thus significantly
reduce the geometric calculations in the inference stage. Compared to its
counterpart, the Mondrian Forest, the BSP-Forest can achieve similar
performance with fewer cuts due to its flexibility. The BSP-Forest also
outperforms other (Bayesian) regression forests on a number of real-world data
sets.
| stat |
Principal manifold estimation via model complexity selection | We propose a framework of principal manifolds to model high-dimensional data.
This framework is based on Sobolev spaces and designed to model data of any
intrinsic dimension. It includes principal component analysis and principal
curve algorithm as special cases. We propose a novel method for model
complexity selection to avoid overfitting, eliminate the effects of outliers,
and improve the computation speed. Additionally, we propose a method for
identifying the interiors of circle-like curves and cylinder/ball-like
surfaces. The proposed approach is compared to existing methods by simulations
and applied to estimate tumor surfaces and interiors in a lung cancer study.
| stat |
Testing for Association in Multi-View Network Data | In this paper, we consider data consisting of multiple networks, each
comprised of a different edge set on a common set of nodes. Many models have
been proposed for the analysis of such multi-view network data under the
assumption that the data views are closely related. In this paper, we provide
tools for evaluating this assumption. In particular, we ask: given two networks
that each follow a stochastic block model, is there an association between the
latent community memberships of the nodes in the two networks? To answer this
question, we extend the stochastic block model for a single network view to the
two-view setting, and develop a new hypothesis test for the null hypothesis
that the latent community memberships in the two data views are independent. We
apply our test to protein-protein interaction data from the HINT database (Das
and Hint, 2012). We find evidence of a weak association between the latent
community memberships of proteins defined with respect to binary interaction
data and the latent community memberships of proteins defined with respect to
co-complex association data. We also extend this proposal to the setting of a
network with node covariates.
| stat |
Estimating a pressure dependent thermal conductivity coefficient with
applications in food technology | In this paper we introduce a method to estimate a pressure dependent thermal
conductivity coefficient arising in a heat diffusion model with applications in
food technology. To address the known smoothing effect of the direct problem,
we model the uncertainty of the conductivity coefficient as a hierarchical
Gaussian Markov random field (GMRF) restricted to uniqueness conditions for the
solution of the inverse problem established in Fraguela et al. Furthermore, we
propose a Single Variable Exchange Metropolis-Hastings algorithm to sample the
corresponding conditional probability distribution of the conductivity
coefficient given observations of the temperature. Sensitivity analysis of the
direct problem suggests that large integration times are necessary to identify
the thermal conductivity coefficient. Numerical evidence indicates that a
signal to noise ratio of roughly 1000 suffices to reliably retrieve the thermal
conductivity coefficient.
| stat |
Elements and Principles for Characterizing Variation between Data
Analyses | The data revolution has led to an increased interest in the practice of data
analysis. For a given problem, there can be significant or subtle differences
in how a data analyst constructs or creates a data analysis, including
differences in the choice of methods, tooling, and workflow. In addition, data
analysts can prioritize (or not) certain objective characteristics in a data
analysis, leading to differences in the quality or experience of the data
analysis, such as an analysis that is more or less reproducible or an analysis
that is more or less exhaustive. However, data analysts currently lack a formal
mechanism to compare and contrast what makes analyses different from each
other. To address this problem, we introduce a vocabulary to describe and
characterize variation between data analyses. We denote this vocabulary as the
elements and principles of data analysis, and we use them to describe the
fundamental concepts for the practice and teaching of creating a data analysis.
This leads to two insights: it suggests a formal mechanism to evaluate data
analyses based on objective characteristics, and it provides a framework to
teach students how to build data analyses.
| stat |
Modern Subsampling Methods for Large-Scale Least Squares Regression | Subsampling methods aim to select a subsample as a surrogate for the observed
sample. As a powerful technique for large-scale data analysis, various
subsampling methods are developed for more effective coefficient estimation and
model prediction. This review presents some cutting-edge subsampling methods
based on the large-scale least squares estimation. Two major families of
subsampling methods are introduced, respectively, the randomized subsampling
approach and the optimal subsampling approach. The former aims to develop a
more effective data-dependent sampling probability, while the latter aims to
select a deterministic subsample in accordance with certain optimality
criteria. Real data examples are provided to compare these methods empirically,
respecting both the estimation accuracy and the computing time.
| stat |
Multi-outcome trials with a generalised number of efficacious outcomes | Existing multi-outcome designs focus almost entirely on evaluating whether
all outcomes show evidence of efficacy or whether at least one outcome shows
evidence of efficacy. While a small number of authors have provided
multi-outcome designs that evaluate when a general number of outcomes show
promise, these designs have been single-stage in nature only. We therefore
propose two designs, of group-sequential and drop the loser form, that provide
this design characteristic in a multi-stage setting. Previous such
multi-outcome multi-stage designs have allowed only for a maximum of two
outcomes; our designs thus also extend previous related proposals by permitting
any number of outcomes.
| stat |
A Model-free Approach for Testing Association | The question of association between outcome and feature is generally framed
in the context of a model on functional and distributional forms. Our
motivating application is that of identifying serum biomarkers of angiogenesis,
energy metabolism, apoptosis, and inflammation, predictive of recurrence after
lung resection in node-negative non-small cell lung cancer patients with tumor
stage T2a or less. We propose an omnibus approach for testing association that
is free of assumptions on functional forms and distributions and can be used as
a black box method. This proposed maximal permutation test is based on the idea
of thresholding, is readily implementable and is computationally efficient. We
illustrate that the proposed omnibus tests maintain their levels and have
strong power as black box tests for detecting linear, nonlinear and
quantile-based associations, even with outlier-prone and heavy-tailed error
distributions and under nonparametric setting. We additionally illustrate the
use of this approach in model-free feature screening and further examine the
level and power of these tests for binary outcome. We compare the performance
of the proposed omnibus tests with comparator methods in our motivating
application to identify preoperative serum biomarkers associated with non-small
cell lung cancer recurrence in early stage patients.
| stat |
Learning disconnected manifolds: a no GANs land | Typical architectures of Generative AdversarialNetworks make use of a
unimodal latent distribution transformed by a continuous generator.
Consequently, the modeled distribution always has connected support which is
cumbersome when learning a disconnected set of manifolds. We formalize this
problem by establishing a no free lunch theorem for the disconnected manifold
learning stating an upper bound on the precision of the targeted distribution.
This is done by building on the necessary existence of a low-quality region
where the generator continuously samples data between two disconnected modes.
Finally, we derive a rejection sampling method based on the norm of generators
Jacobian and show its efficiency on several generators including BigGAN.
| stat |
Deep learning for Chemometric and non-translational data | We propose a novel method to train deep convolutional neural networks which
learn from multiple data sets of varying input sizes through weight sharing.
This is an advantage in chemometrics where individual measurements represent
exact chemical compounds and thus signals cannot be translated or resized
without disturbing their interpretation. Our approach show superior performance
compared to transfer learning when a medium sized and a small data set are
trained together. While we observe a small improvement compared to individual
training when two medium sized data sets are trained together, in particular
through a reduction in the variance.
| stat |
Sum-Product-Transform Networks: Exploiting Symmetries using Invertible
Transformations | In this work, we propose Sum-Product-Transform Networks (SPTN), an extension
of sum-product networks that uses invertible transformations as additional
internal nodes. The type and placement of transformations determine properties
of the resulting SPTN with many interesting special cases. Importantly, SPTN
with Gaussian leaves and affine transformations pose the same inference task
tractable that can be computed efficiently in SPNs. We propose to store affine
transformations in their SVD decompositions using an efficient parametrization
of unitary matrices by a set of Givens rotations. Last but not least, we
demonstrate that G-SPTNs achieve state-of-the-art results on the density
estimation task and are competitive with state-of-the-art methods for anomaly
detection.
| stat |
Concentration bounds for linear Monge mapping estimation and optimal
transport domain adaptation | This article investigates the quality of the estimator of the linear Monge
mapping between distributions. We provide the first concentration result on the
linear mapping operator and prove a sample complexity of $n^{-1/2}$ when using
empirical estimates of first and second order moments. This result is then used
to derive a generalization bound for domain adaptation with optimal transport.
As a consequence, this method approaches the performance of theoretical Bayes
predictor under mild conditions on the covariance structure of the problem. We
also discuss the computational complexity of the linear mapping estimation and
show that when the source and target are stationary the mapping is a
convolution that can be estimated very efficiently using fast Fourier
transforms. Numerical experiments reproduce the behavior of the proven bounds
on simulated and real data for mapping estimation and domain adaptation on
images.
| stat |
The Use of Gaussian Processes in System Identification | Gaussian processes are used in machine learning to learn input-output
mappings from observed data. Gaussian process regression is based on imposing a
Gaussian process prior on the unknown regressor function and statistically
conditioning it on the observed data. In system identification, Gaussian
processes are used to form time series prediction models such as non-linear
finite-impulse response (NFIR) models as well as non-linear autoregressive
(NARX) models. Gaussian process state-space models (GPSS) can be used to learn
the dynamic and measurement models for a state-space representation of the
input-output data. Temporal and spatio-temporal Gaussian processes can be
directly used to form regressor on the data in the time domain. The aim of this
article is to briefly outline the main directions in system identification
methods using Gaussian processes.
| stat |
PICAR: An Efficient Extendable Approach for Fitting Hierarchical Spatial
Models | Hierarchical spatial models are very flexible and popular for a vast array of
applications in areas such as ecology, social science, public health, and
atmospheric science. It is common to carry out Bayesian inference for these
models via Markov chain Monte Carlo (MCMC). Each iteration of the MCMC
algorithm is computationally expensive due to costly matrix operations. In
addition, the MCMC algorithm needs to be run for more iterations because the
strong cross-correlations among the spatial latent variables result in slow
mixing Markov chains. To address these computational challenges, we propose a
projection-based intrinsic conditional autoregression (PICAR) approach, which
is a discretized and dimension-reduced representation of the underlying spatial
random field using empirical basis functions on a triangular mesh. Our approach
exhibits fast mixing as well as a considerable reduction in computational cost
per iteration. PICAR is computationally efficient and scales well to high
dimensions. It is also automated and easy to implement for a wide array of
user-specified hierarchical spatial models. We show, via simulation studies,
that our approach performs well in terms of parameter inference and prediction.
We provide several examples to illustrate the applicability of our method,
including (i) a parasitic tree infestation data set that showcases its
computational efficiency in high-dimensional applications, (ii) a spatially
varying coefficient model that demonstrates the ease of implementation of PICAR
in the probabilistic programming language stan, and (iii) a watershed survey
example that illustrates how PICAR applies to models that are not amenable to
efficient inference via existing methods.
| stat |
Matching Algorithms for Causal Inference with Multiple Treatments | Randomized clinical trials (RCTs) are ideal for estimating causal effects,
because the distributions of background covariates are similar in expectation
across treatment groups. When estimating causal effects using observational
data, matching is a commonly used method to replicate the covariate balance
achieved in a RCT. Matching algorithms have a rich history dating back to the
mid-1900s, but have been used mostly to estimate causal effects between two
treatment groups. When there are more than two treatments, estimating causal
effects requires additional assumptions and techniques. We propose matching
algorithms that address the drawbacks of the current methods, and we use
simulations to compare current and new methods. All of the methods display
improved covariate balance in the matched sets relative to the pre-matched
cohorts. In addition, we provide advice to investigators on which matching
algorithms are preferred for different covariate distributions.
| stat |
Nonparametric causal mediation analysis for stochastic interventional
(in)direct effects | Causal mediation analysis has historically been limited in two important
ways: (i) a focus has traditionally been placed on binary treatments and static
interventions, and (ii) direct and indirect effect decompositions have been
pursued that are only identifiable in the absence of intermediate confounders
affected by treatment. We present a theoretical study of an (in)direct effect
decomposition of the population intervention effect, defined by stochastic
interventions jointly applied to the treatment and mediators. In contrast to
existing proposals, our causal effects can be evaluated regardless of whether a
treatment is categorical or continuous and remain well-defined even in the
presence of intermediate confounders affected by treatment. Our (in)direct
effects are identifiable without a restrictive assumption on cross-world
counterfactual independencies, allowing for substantive conclusions drawn from
them to be validated in randomized controlled trials. Beyond the novel effects
introduced, we provide a careful study of nonparametric efficiency theory
relevant for the construction of flexible, multiply robust estimators of our
(in)direct effects, while avoiding undue restrictions induced by assuming
parametric models of nuisance parameter functionals. To complement our
nonparametric estimation strategy, we introduce inferential techniques for
constructing confidence intervals and hypothesis tests, and discuss open source
software implementing the proposed methodology.
| stat |
Flexible Clustering with a Sparse Mixture of Generalized Hyperbolic
Distributions | Robust clustering of high-dimensional data is an important topic because, in
many practical situations, real data sets are heavy-tailed and/or asymmetric.
Moreover, traditional model-based clustering often fails for high dimensional
data due to the number of free covariance parameters. A parametrization of the
component scale matrices for the mixture of generalized hyperbolic
distributions is proposed by including a penalty term in the likelihood
constraining the parameters resulting in a flexible model for high dimensional
data and a meaningful interpretation. An analytically feasible EM algorithm is
developed by placing a gamma-Lasso penalty constraining the concentration
matrix. The proposed methodology is investigated through simulation studies and
two real data sets.
| stat |
Threshold Selection in Univariate Extreme Value Analysis | Threshold selection plays a key role for various aspects of statistical
inference of rare events. Most classical approaches tackling this problem for
heavy-tailed distributions crucially depend on tuning parameters or critical
values to be chosen by the practitioner. To simplify the use of automated,
data-driven threshold selection methods, we introduce two new procedures not
requiring the manual choice of any parameters. The first method measures the
deviation of the log-spacings from the exponential distribution and achieves
good performance in simulations for estimating high quantiles. The second
approach smoothly estimates the asymptotic mean square error of the Hill
estimator and performs consistently well over a wide range of distributions.
The methods are compared to existing procedures in an extensive simulation
study and applied to a dataset of financial losses, where the underlying
extreme value index is assumed to vary over time. This application strongly
emphasizes the importance of solid automated threshold selection.
| stat |
Distances with mixed type variables some modified Gower's coefficients | Nearest neighbor methods have become popular in official statistics, mainly
in imputation or in statistical matching problems; they play a key role in
machine learning too, where a high number of variants have been proposed. The
choice of the distance function depends mainly on the type of the selected
variables. Unfortunately, relatively few options permit to handle mixed type
variables, a situation frequently encountered in official statistics. The most
popular distance for mixed type variables is derived as the complement of the
Gower's similarity coefficient; it is appealing because ranges between 0 and 1
and allows to handle missing values. Unfortunately, the unweighted standard
setting the contribution of the single variables to the overall Gower's
distance is unbalanced because of the different nature of the variables
themselves. This article tries to address the main drawbacks that affect the
overall unweighted Gower's distance by suggesting some modifications in
calculating the distance on the interval and ratio scaled variables. Simple
modifications try to attenuate the impact of outliers on the scaled Manhattan
distance; other modifications, relying on the kernel density estimation methods
attempt to reduce the unbalanced contribution of the different types of
variables. The performance of the proposals is evaluated in simulations
mimicking the imputation of missing values through nearest neighbor distance
hotdeck method.
| stat |
Similarity indexing & GIS analysis of air pollution | Pollution has become a major threat in almost all metropolitan cities around
the world. Currently, atmospheric scientists are working on various models that
could help us understand air pollution. In this paper, we have formulated a new
metric tool called Delhi Similarity Index (DSI). The DSI is defined as the
geometrical mean of the trace gases such as ozone, sulfur-dioxide and
carbon-monoxide, which ranges from 0 (dissimilar to Delhi) to 0.9-1 (similar to
Delhi). The limitation of the tool concerning the result of the
nitrous-di-oxide data set is also analyzed. Also, the GIS projections of PM 2.5
role for Indian cities are graphically represented. The DSI results from 2011
to 2014 data show that Bengaluru is in the threshold of becoming as polluted
like Delhi with values varying from 0.8 to 0.9 (i.e. 80-90%) and Jungfraujoch
with a 0.65 to 0.7 (i.e. 65-70%).
| stat |
Instrumental Variable Value Iteration for Causal Offline Reinforcement
Learning | In offline reinforcement learning (RL) an optimal policy is learnt solely
from a priori collected observational data. However, in observational data,
actions are often confounded by unobserved variables. Instrumental variables
(IVs), in the context of RL, are the variables whose influence on the state
variables are all mediated through the action. When a valid instrument is
present, we can recover the confounded transition dynamics through
observational data. We study a confounded Markov decision process where the
transition dynamics admit an additive nonlinear functional form. Using IVs, we
derive a conditional moment restriction (CMR) through which we can identify
transition dynamics based on observational data. We propose a provably
efficient IV-aided Value Iteration (IVVI) algorithm based on a primal-dual
reformulation of CMR. To the best of our knowledge, this is the first provably
efficient algorithm for instrument-aided offline RL.
| stat |
Gradient Regularisation as Approximate Variational Inference | Variational inference in Bayesian neural networks is usually performed using
stochastic sampling which gives very high-variance gradients, and hence slow
learning. Here, we show that it is possible to obtain a deterministic
approximation of the ELBO for a Bayesian neural network by doing a
Taylor-series expansion around the mean of the current variational
distribution. The resulting approximate ELBO is the training-log-likelihood
plus a squared gradient regulariser. In addition to learning the approximate
posterior variance, we also consider a uniform-variance approximate posterior,
inspired by the stationary distribution of SGD. The corresponding approximate
ELBO has a simple form, as the log-likelihood plus a simple squared-gradient
regulariser. We argue that this squared-gradient regularisation may at the root
of the excellent empirical performance of SGD.
| stat |
Statistical methods for biomarker data pooled from multiple nested
case-control studies | Pooling biomarker data across multiple studies allows for examination of a
wider exposure range than generally possible in individual studies, evaluation
of population subgroups and disease subtypes with more statistical power, and
more precise estimation of biomarker-disease associations. However, biomarker
measurements often require calibration to a reference assay prior to pooling
due to assay and laboratory variability across studies. We propose several
methods for calibrating and combining biomarker data from nested case-control
studies when reference assay data are obtained from a subset of controls in
each contributing study. Specifically, we describe a two-stage method and two
aggregated methods, named the internalized and full calibration methods, to
evaluate the main effect of the biomarker exposure on disease risk and whether
that association is modified by a potential covariate. The internalized method
uses the reference laboratory measurement in the analysis when available and
otherwise uses the calibrated measurement. The full calibration method uses
calibrated biomarker measurements for all subjects, even those with reference
laboratory measurements. Our results demonstrate that the full calibration
method is the preferred aggregated approach to minimize bias in point estimates
regardless of the inclusion of an interaction term in the model. We also
observe that the two-stage and full calibration methods provide similar effect
and variance estimates, but that the variance estimates for these two methods
are slightly larger than those from the internalized approach. As an
illustrative example, we apply the methods in a pooling project of nested
case-control studies to evaluate (i) the association between circulating
vitamin D levels and risk of stroke, and (ii) how BMI modifies the association
between circulating vitamin D levels and cardiovascular disease.
| stat |
Robust policy evaluation from large-scale observational studies | Under current policy decision making paradigm, we make or evaluate a policy
decision by intervening different socio-economic parameters and analyzing the
impact of those interventions. This process involves identifying the causal
relation between interventions and outcomes. Matching method is one of the
popular techniques to identify such causal relations. However, in one-to-one
matching, when a treatment or control unit has multiple pair assignment options
with similar match quality, different matching algorithms often assign
different pairs. Since, all the matching algorithms assign pair without
considering the outcomes, it is possible that with same data and same
hypothesis, different experimenters can make different conclusions. This
problem becomes more prominent in case of large-scale observational studies.
Recently, a robust approach is proposed to tackle the uncertainty which uses
discrete optimization techniques to explore all possible assignments. Though
optimization techniques are very efficient in its own way, they are not
scalable to big data. In this work, we consider causal inference testing with
binary outcomes and propose computationally efficient algorithms that are
scalable to large-scale observational studies. By leveraging the structure of
the optimization model, we propose a robustness condition which further reduces
the computational burden. We validate the efficiency of the proposed algorithms
by testing the causal relation between Hospital Readmission Reduction Program
(HRRP) and readmission to different hospital (non-index readmission) on the
State of California Patient Discharge Database from 2010 to 2014. Our result
shows that HRRP has a causal relation with the increase in non-index
readmission and the proposed algorithms proved to be highly scalable in testing
causal relations from large-scale observational studies.
| stat |
A Generalized Heckman Model With Varying Sample Selection Bias and
Dispersion Parameters | Many proposals have emerged as alternatives to the Heckman selection model,
mainly to address the non-robustness of its normal assumption. The 2001 Medical
Expenditure Panel Survey data is often used to illustrate this non-robustness
of the Heckman model. In this paper, we propose a generalization of the Heckman
sample selection model by allowing the sample selection bias and dispersion
parameters to depend on covariates. We show that the non-robustness of the
Heckman model may be due to the assumption of the constant sample selection
bias parameter rather than the normality assumption. Our proposed methodology
allows us to understand which covariates are important to explain the sample
selection bias phenomenon rather than to only form conclusions about its
presence. We explore the inferential aspects of the maximum likelihood
estimators (MLEs) for our proposed generalized Heckman model. More
specifically, we show that this model satisfies some regularity conditions such
that it ensures consistency and asymptotic normality of the MLEs. Proper score
residuals for sample selection models are provided, and model adequacy is
addressed. Simulated results are presented to check the finite-sample behavior
of the estimators and to verify the consequences of not considering varying
sample selection bias and dispersion parameters. We show that the normal
assumption for analyzing medical expenditure data is suitable and that the
conclusions drawn using our approach are coherent with findings from prior
literature. Moreover, we identify which covariates are relevant to explain the
presence of sample selection bias in this important dataset.
| stat |
Rough-Fuzzy CPD: A Gradual Change Point Detection Algorithm | Changepoint detection is the problem of finding abrupt or gradual changes in
time series data when the distribution of the time series changes
significantly. There are many sophisticated statistical algorithms for solving
changepoint detection problem, although there is not much work devoted towards
gradual changepoints as compared to abrupt ones. Here we present a new approach
to solve changepoint detection problem using fuzzy rough set theory which is
able to detect such gradual changepoints. An expression for the rough-fuzzy
estimate of changepoints is derived along with its mathematical properties
concerning fast computation. In a statistical hypothesis testing framework,
asymptotic distribution of the proposed statistic on both single and multiple
changepoints is derived under null hypothesis enabling multiple changepoint
detection. Extensive simulation studies have been performed to investigate how
simple crude statistical measures of disparity can be subjected to improve
their efficiency in estimation of gradual changepoints. Also, the said
rough-fuzzy estimate is robust to signal-to-noise ratio, high degree of
fuzziness in true changepoints and also to hyper parameter values. Simulation
studies reveal that the proposed method beats other fuzzy methods and also
popular crisp methods like WBS, PELT and BOCD in detecting gradual
changepoints. The applicability of the estimate is demonstrated using multiple
real-life datasets including Covid-19. We have developed the python package
"roufcp" for broader dissemination of the methods.
| stat |
Latent Simplex Position Model: High Dimensional Multi-view Clustering
with Uncertainty Quantification | High dimensional data often contain multiple facets, and several clustering
patterns can co-exist under different variable subspaces, also known as the
views. While multi-view clustering algorithms were proposed, the uncertainty
quantification remains difficult --- a particular challenge is in the high
complexity of estimating the cluster assignment probability under each view,
and sharing information among views. In this article, we propose an approximate
Bayes approach --- treating the similarity matrices generated over the views as
rough first-stage estimates for the co-assignment probabilities; in its
Kullback-Leibler neighborhood, we obtain a refined low-rank matrix, formed by
the pairwise product of simplex coordinates. Interestingly, each simplex
coordinate directly encodes the cluster assignment uncertainty. For multi-view
clustering, we let each view draw a parameterization from a few candidates,
leading to dimension reduction. With high model flexibility, the estimation can
be efficiently carried out as a continuous optimization problem, hence enjoys
gradient-based computation. The theory establishes the connection of this model
to a random partition distribution under multiple views. Compared to
single-view clustering approaches, substantially more interpretable results are
obtained when clustering brains from a human traumatic brain injury study,
using high-dimensional gene expression data.
KEY WORDS: Co-regularized Clustering, Consensus, PAC-Bayes, Random Cluster
Graph, Variable Selection
| stat |
Counting Defiers | The LATE monotonicity assumption of Imbens and Angrist (1994) precludes
"defiers," individuals whose treatment always runs counter to the instrument,
in the terminology of Balke and Pearl (1993) and Angrist et al. (1996). I allow
for defiers in a model with a binary instrument and a binary treatment. The
model is explicit about the randomization process that gives rise to the
instrument. I use the model to develop estimators of the counts of defiers,
always takers, compliers, and never takers. I propose separate versions of the
estimators for contexts in which the parameter of the randomization process is
unspecified, which I intend for use with natural experiments with virtual
random assignment. I present an empirical application that revisits Angrist and
Evans (1998), which examines the impact of virtual random assignment of the sex
of the first two children on subsequent fertility. I find that subsequent
fertility is much more responsive to the sex mix of the first two children when
defiers are allowed.
| stat |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.