title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
An Assumption-Free Exact Test For Fixed-Design Linear Models With
Exchangeable Errors | We propose the Cyclic Permutation Test (CPT) to test general linear
hypotheses for linear models. This test is non-randomized and valid in finite
samples with exact Type I error $\alpha$ for an arbitrary fixed design matrix
and arbitrary exchangeable errors, whenever $1 / \alpha$ is an integer and $n /
p \ge 1 / \alpha - 1$. The test involves applying the marginal rank test to $1
/ \alpha$ linear statistics of the outcome vector, where the coefficient
vectors are determined by solving a linear system such that the joint
distribution of the linear statistics is invariant with respect to a
non-standard cyclic permutation group under the null hypothesis.The power can
be further enhanced by solving a secondary non-linear travelling salesman
problem, for which the genetic algorithm can find a reasonably good solution.
Extensive simulation studies show that the CPT has comparable power to existing
tests. When testing for a single contrast of coefficients, an exact confidence
interval can be obtained by inverting the test. Furthermore, we provide a
selective yet extensive literature review of the century-long efforts on this
problem, highlighting the novelty of our test.
| stat |
Multi-fidelity Bayesian Optimization with Max-value Entropy Search and
its parallelization | In a standard setting of Bayesian optimization (BO), the objective function
evaluation is assumed to be highly expensive. Multi-fidelity Bayesian
optimization (MFBO) accelerates BO by incorporating lower fidelity observations
available with a lower sampling cost. In this paper, we focus on the
information-based approach, which is a popular and empirically successful
approach in BO. For MFBO, however, existing information-based methods are
plagued by difficulty in estimating the information gain. We propose an
approach based on max-value entropy search (MES), which greatly facilitates
computations by considering the entropy of the optimal function value instead
of the optimal input point. We show that, in our multi-fidelity MES (MF-MES),
most of additional computations, compared with usual MES, is reduced to
analytical computations. Although an additional numerical integration is
necessary for the information across different fidelities, this is only in one
dimensional space, which can be performed efficiently and accurately. Further,
we also propose parallelization of MF-MES. Since there exist a variety of
different sampling costs, queries typically occur asynchronously in MFBO. We
show that similar simple computations can be derived for asynchronous parallel
MFBO. We demonstrate effectiveness of our approach by using benchmark datasets
and a real-world application to materials science data.
| stat |
Simultaneous Preference and Metric Learning from Paired Comparisons | A popular model of preference in the context of recommendation systems is the
so-called \emph{ideal point} model. In this model, a user is represented as a
vector $\mathbf{u}$ together with a collection of items $\mathbf{x_1}, \ldots,
\mathbf{x_N}$ in a common low-dimensional space. The vector $\mathbf{u}$
represents the user's "ideal point," or the ideal combination of features that
represents a hypothesized most preferred item. The underlying assumption in
this model is that a smaller distance between $\mathbf{u}$ and an item
$\mathbf{x_j}$ indicates a stronger preference for $\mathbf{x_j}$. In the vast
majority of the existing work on learning ideal point models, the underlying
distance has been assumed to be Euclidean. However, this eliminates any
possibility of interactions between features and a user's underlying
preferences. In this paper, we consider the problem of learning an ideal point
representation of a user's preferences when the distance metric is an unknown
Mahalanobis metric. Specifically, we present a novel approach to estimate the
user's ideal point $\mathbf{u}$ and the Mahalanobis metric from paired
comparisons of the form "item $\mathbf{x_i}$ is preferred to item
$\mathbf{x_j}$." This can be viewed as a special case of a more general metric
learning problem where the location of some points are unknown a priori. We
conduct extensive experiments on synthetic and real-world datasets to exhibit
the effectiveness of our algorithm.
| stat |
Adaptive Batching for Gaussian Process Surrogates with Application in
Noisy Level Set Estimation | We develop adaptive replicated designs for Gaussian process metamodels of
stochastic experiments. Adaptive batching is a natural extension of sequential
design heuristics with the benefit of replication growing as response features
are learned, inputs concentrate, and the metamodeling overhead rises. Motivated
by the problem of learning the level set of the mean simulator response we
develop four novel schemes: Multi-Level Batching (MLB), Ratchet Batching (RB),
Adaptive Batched Stepwise Uncertainty Reduction (ABSUR), Adaptive Design with
Stepwise Allocation (ADSA) and Deterministic Design with Stepwise Allocation
(DDSA). Our algorithms simultaneously (MLB, RB and ABSUR) or sequentially (ADSA
and DDSA) determine the sequential design inputs and the respective number of
replicates. Illustrations using synthetic examples and an application in
quantitative finance (Bermudan option pricing via Regression Monte Carlo) show
that adaptive batching brings significant computational speed-ups with minimal
loss of modeling fidelity.
| stat |
Improving Offline Contextual Bandits with Distributional Robustness | This paper extends the Distributionally Robust Optimization (DRO) approach
for offline contextual bandits. Specifically, we leverage this framework to
introduce a convex reformulation of the Counterfactual Risk Minimization
principle. Besides relying on convex programs, our approach is compatible with
stochastic optimization, and can therefore be readily adapted tothe large data
regime. Our approach relies on the construction of asymptotic confidence
intervals for offline contextual bandits through the DRO framework. By
leveraging known asymptotic results of robust estimators, we also show how to
automatically calibrate such confidence intervals, which in turn removes the
burden of hyper-parameter selection for policy optimization. We present
preliminary empirical results supporting the effectiveness of our approach.
| stat |
From univariate to multivariate coupling between continuous signals and
point processes: a mathematical framework | Time series datasets often contain heterogeneous signals, composed of both
continuously changing quantities and discretely occurring events. The coupling
between these measurements may provide insights into key underlying mechanisms
of the systems under study. To better extract this information, we investigate
the asymptotic statistical properties of coupling measures between continuous
signals and point processes. We first introduce martingale stochastic
integration theory as a mathematical model for a family of statistical
quantities that include the Phase Locking Value, a classical coupling measure
to characterize complex dynamics. Based on the martingale Central Limit
Theorem, we can then derive the asymptotic Gaussian distribution of estimates
of such coupling measure, that can be exploited for statistical testing.
Second, based on multivariate extensions of this result and Random Matrix
Theory, we establish a principled way to analyze the low rank coupling between
a large number of point processes and continuous signals. For a null hypothesis
of no coupling, we establish sufficient conditions for the empirical
distribution of squared singular values of the matrix to converge, as the
number of measured signals increases, to the well-known Marchenko-Pastur (MP)
law, and the largest squared singular value converges to the upper end of the
MPs support. This justifies a simple thresholding approach to assess the
significance of multivariate coupling. Finally, we illustrate with simulations
the relevance of our univariate and multivariate results in the context of
neural time series, addressing how to reliably quantify the interplay between
multi channel Local Field Potential signals and the spiking activity of a large
population of neurons.
| stat |
A Review on Modern Computational Optimal Transport Methods with
Applications in Biomedical Research | Optimal transport has been one of the most exciting subjects in mathematics,
starting from the 18th century. As a powerful tool to transport between two
probability measures, optimal transport methods have been reinvigorated
nowadays in a remarkable proliferation of modern data science applications. To
meet the big data challenges, various computational tools have been developed
in the recent decade to accelerate the computation for optimal transport
methods. In this review, we present some cutting-edge computational optimal
transport methods with a focus on the regularization-based methods and the
projection-based methods. We discuss their real-world applications in
biomedical research.
| stat |
Non-asymptotic error bounds for scaled underdamped Langevin MCMC | Recent works have derived non-asymptotic upper bounds for convergence of
underdamped Langevin MCMC. We revisit these bound and consider introducing
scaling terms in the underlying underdamped Langevin equation. In particular,
we provide conditions under which an appropriate scaling allows to improve the
error bounds in terms of the condition number of the underlying density of
interest.
| stat |
Cross-Validated Loss-Based Covariance Matrix Estimator Selection in High
Dimensions | The covariance matrix plays a fundamental role in many modern exploratory and
inferential statistical procedures, including dimensionality reduction,
hypothesis testing, and regression. In low-dimensional regimes, where the
number of observations far exceeds the number of variables, the optimality of
the sample covariance matrix as an estimator of this parameter is
well-established. High-dimensional regimes do not admit such a convenience,
however. As such, a variety of estimators have been derived to overcome the
shortcomings of the sample covariance matrix in these settings. Yet, the
question of selecting an optimal estimator from among the plethora available
remains largely unaddressed. Using the framework of cross-validated loss-based
estimation, we develop the theoretical underpinnings of just such an estimator
selection procedure. In particular, we propose a general class of loss
functions for covariance matrix estimation and establish finite-sample risk
bounds and conditions for the asymptotic optimality of the cross-validated
estimator selector with respect to these loss functions. We evaluate our
proposed approach via a comprehensive set of simulation experiments and
demonstrate its practical benefits by application in the exploratory analysis
of two single-cell transcriptome sequencing datasets. A free and open-source
software implementation of the proposed methodology, the cvCovEst R package, is
briefly introduced.
| stat |
Ensemble Conditional Variance Estimator for Sufficient Dimension
Reduction | Ensemble Conditional Variance Estimation (ECVE) is a novel sufficient
dimension reduction (SDR) method in regressions with continuous response and
predictors. ECVE applies to general non-additive error regression models. It
operates under the assumption that the predictors can be replaced by a lower
dimensional projection without loss of information. It is a semiparametric
forward regression model based exhaustive sufficient dimension reduction
estimation method that is shown to be consistent under mild assumptions. It is
shown to outperform central subspace mean average variance estimation (csMAVE),
its main competitor, under several simulation settings and in a benchmark data
set analysis.
| stat |
Stochastic gradient descent with noise of machine learning type. Part I:
Discrete time analysis | Stochastic gradient descent (SGD) is one of the most popular algorithms in
modern machine learning. The noise encountered in these applications is
different from that in many theoretical analyses of stochastic gradient
algorithms. In this article, we discuss some of the common properties of energy
landscapes and stochastic noise encountered in machine learning problems, and
how they affect SGD-based optimization.
In particular, we show that the learning rate in SGD with machine learning
noise can be chosen to be small, but uniformly positive for all times if the
energy landscape resembles that of overparametrized deep learning problems. If
the objective function satisfies a Lojasiewicz inequality, SGD converges to the
global minimum exponentially fast, and even for functions which may have local
minima, we establish almost sure convergence to the global minimum at an
exponential rate from any finite energy initialization. The assumptions that we
make in this result concern the behavior where the objective function is either
small or large and the nature of the gradient noise, but the energy landscape
is fairly unconstrained on the domain where the objective function takes values
in an intermediate regime.
| stat |
A Likelihood-based Alternative to Null Hypothesis Significance Testing | The logical and practical difficulties associated with research
interpretation using P values and null hypothesis significance testing have
been extensively documented. This paper describes an alternative,
likelihood-based approach to P-value interpretation. The P-value and sample
size of a research study are used to derive a likelihood function with a single
parameter, the estimated population effect size, and the method of maximum
likelihood estimation is used to calculate the most likely effect size.
Comparison of the likelihood of the most likely effect size and the likelihood
of the minimum clinically significant effect size using the likelihood ratio
test yields the clinical significance support level (or S-value), a logical and
easily understood metric of research evidence. This clinical significance
likelihood approach has distinct advantages over null hypothesis significance
testing. As motivating examples we demonstrate the calculation and
interpretation of S-values applied to two recent widely publicised trials,
WOMAN from the Lancet and RELIEF from the New England Journal of Medicine.
| stat |
Information Criterion for Boltzmann Approximation Problems | This paper considers the problem of approximating a density when it can be
evaluated up to a normalizing constant at a limited number of points. We call
this problem the Boltzmann approximation (BA) problem. The BA problem is
ubiquitous in statistics, such as approximating a posterior density for
Bayesian inference and estimating an optimal density for importance sampling.
Approximating the density with a parametric model can be cast as a model
selection problem. This problem cannot be addressed with traditional approaches
that maximize the (marginal) likelihood of a model, for example, using the
Akaike information criterion (AIC) or Bayesian information criterion (BIC). We
instead aim to minimize the cross-entropy that gauges the deviation of a
parametric model from the target density. We propose a novel information
criterion called the cross-entropy information criterion (CIC) and prove that
the CIC is an asymptotically unbiased estimator of the cross-entropy (up to a
multiplicative constant) under some regularity conditions. We propose an
iterative method to approximate the target density by minimizing the CIC. We
demonstrate that the proposed method selects a parametric model that well
approximates the target density.
| stat |
A spatial dependence graph model for multivariate spatial hybrid
processes | This paper is concerned with the joint analysis of multivariate mixed-type
spatial data, where some components are point processes and some are of
lattice-type by nature. After a survey of statistical methods for marked
spatial point and lattice processes, the class of multivariate spatial hybrid
processes is defined and embedded within the framework of spatial dependence
graph models. In this model, the point and lattice sub-processes are identified
with nodes of a graph whereas missing edges represent conditional independence
among the components. This finally leads to a general framework for any type of
spatial data in a multivariate setting. We demonstrate the application of our
method in the analysis of a multivariate point-lattice pattern on crime and
ambulance service call-out incidents recorded in London, where the points are
the locations of different pre-classified crime events and the lattice
components report different aggregated incident rates at ward level.
| stat |
On the dimensional indeterminacy of one-wave factor analysis under
causal effects | It is shown, with two sets of survey items that separately load on two
distinct factors, independent of one another conditional on the past, that if
it is the case that at least one of the factors causally affects the other,
then, in many settings, the process will converge to a factor model in which a
single factor will suffice to capture the structure of the associations among
the items. Factor analysis with one wave of data can then not distinguish
between factor models with a single factor versus those with two factors that
are causally related. Therefore, unless causal relations between factors can be
ruled out a priori, alleged empirical evidence from one-wave factor analysis
for a single factor still leaves open the possibilities of a single factor or
of two factors that causally affect one another. The implications for
interpreting the factor structure of psychological scales, such as self-report
scales for anxiety and depression, are discussed. Some further generalizations
to an arbitrary number of underlying factors are noted.
| stat |
Real-time detection of a change-point in a linear expectile model | In the present paper we address the real-time detection problem of a
change-point in the coefficients of a linear model with the possibility that
the model errors are asymmetrical and that the explanatory variables number is
large. We build test statistics based on the cumulative sum (CUSUM) of the
expectile function derivatives calculated on the residuals obtained by the
expectile and adaptive LASSO expectile estimation methods. The asymptotic
distribution of these statistics are obtained under the hypothesis that the
model does not change. Moreover, we prove that they diverge when the model
changes at an unknown observation. The asymptotic study of the test statistics
under these two hypotheses allows us to find the asymptotic critical region and
the stopping time, that is the observation where the model will change. The
empirical performance is investigated by a comparative simulation study with
other statistics of CUSUM type. Two examples on real data are also presented to
demonstrate its interest in practice.
| stat |
Conditionally-additive-noise Models for Structure Learning | Constraint-based structure learning algorithms infer the causal structure of
multivariate systems from observational data by determining an equivalent class
of causal structures compatible with the conditional independencies in the
data. Methods based on additive-noise (AN) models have been proposed to further
discriminate between causal structures that are equivalent in terms of
conditional independencies. These methods rely on a particular form of the
generative functional equations, with an additive noise structure, which allows
inferring the directionality of causation by testing the independence between
the residuals of a nonlinear regression and the predictors
(nrr-independencies). Full causal structure identifiability has been proven for
systems that contain only additive-noise equations and have no hidden
variables. We extend the AN framework in several ways. We introduce alternative
regression-free tests of independence based on conditional variances
(cv-independencies). We consider conditionally-additive-noise (CAN) models, in
which the equations may have the AN form only after conditioning. We exploit
asymmetries in nrr-independencies or cv-independencies resulting from the CAN
form to derive a criterion that infers the causal relation between a pair of
variables in a multivariate system without any assumption about the form of the
equations or the presence of hidden variables.
| stat |
Time-varying Gaussian Process Bandit Optimization with Non-constant
Evaluation Time | The Gaussian process bandit is a problem in which we want to find a maximizer
of a black-box function with the minimum number of function evaluations. If the
black-box function varies with time, then time-varying Bayesian optimization is
a promising framework. However, a drawback with current methods is in the
assumption that the evaluation time for every observation is constant, which
can be unrealistic for many practical applications, e.g., recommender systems
and environmental monitoring. As a result, the performance of current methods
can be degraded when this assumption is violated. To cope with this problem, we
propose a novel time-varying Bayesian optimization algorithm that can
effectively handle the non-constant evaluation time. Furthermore, we
theoretically establish a regret bound of our algorithm. Our bound elucidates
that a pattern of the evaluation time sequence can hugely affect the difficulty
of the problem. We also provide experimental results to validate the practical
effectiveness of the proposed method.
| stat |
A Tukey type trend test for repeated carcinogenicity bioassays,
motivated by multiple glyphosate studies | In the last two decades, significant methodological progress to the
simultaneous inference of simple and complex randomized designs, particularly
proportions as endpoints, occurred. This includes: i) the new Tukey trend test
approach, ii) multiple contrast tests for binomial proportions, iii) multiple
contrast tests for poly-k estimates, and Add-1 approximation for one-sided
inference. This report focus on a new Tukey type trend test to evaluate
repeated long-term carcinogenicity bioassays which was motivated by multiple
glyphosate studies. Notice, it is not the aim here to contribute to the
evaluation of Glyphosate and its controversies. By means of the CRAN-packages
tukeytrend, MCPAN, multcomp the real data analysis is straightforward possible.
| stat |
Improving multilevel regression and poststratification with structured
priors | A central theme in the field of survey statistics is estimating
population-level quantities through data coming from potentially
non-representative samples of the population. Multilevel Regression and
Poststratification (MRP), a model-based approach, is gaining traction against
the traditional weighted approach for survey estimates. MRP estimates are
susceptible to bias if there is an underlying structure that the methodology
does not capture. This work aims to provide a new framework for specifying
structured prior distributions that lead to bias reduction in MRP estimates. We
use simulation studies to explore the benefit of these prior distributions and
demonstrate their efficacy on non-representative US survey data. We show that
structured prior distributions offer absolute bias reduction and variance
reduction for posterior MRP estimates in a large variety of data regimes.
| stat |
High-Performance Statistical Computing in the Computing Environments of
the 2020s | Technological advances in the past decade, hardware and software alike, have
made access to high-performance computing (HPC) easier than ever. We review
these advances from a statistical computing perspective. Cloud computing allows
access to supercomputers affordable. Deep learning software libraries make
programming statistical algorithms easy, and enable users to write code once
and run it anywhere from a laptop to a workstation with multiple graphics
processing units (GPUs) or a supercomputer in a cloud. To promote statisticians
to benefit from these developments, we review recent optimization algorithms
that are useful for high-dimensional models and can harness the power of HPC.
Code snippets are provided for the readers to grasp the ease of programming. We
also provide an easy-to-use distributed matrix data structure suitable for HPC.
Employing this data structure, we illustrate various statistical applications
including large-scale nonnegative matrix factorization, positron emission
tomography, multidimensional scaling, and $\ell_1$-regularized Cox regression.
Our examples easily scale up to an 8-GPU workstation and a 720-CPU-core cluster
in a cloud. As a case in point, we analyze the on-set of type-2 diabetes from
the UK Biobank with 200,000 subjects and about 500,000 single nucleotide
polymorphisms using the HPC $\ell_1$-regularized Cox regression. Fitting a
half-million-variate model takes less than 45 minutes, reconfirming known
associations. To our knowledge, the feasibility of jointly genome-wide
association analysis of survival outcomes at this scale is first demonstrated.
| stat |
Support vector comparison machines | In ranking problems, the goal is to learn a ranking function from labeled
pairs of input points. In this paper, we consider the related comparison
problem, where the label indicates which element of the pair is better, or if
there is no significant difference. We cast the learning problem as a margin
maximization, and show that it can be solved by converting it to a standard
SVM. We use simulated nonlinear patterns, a real learning to rank sushi data
set, and a chess data set to show that our proposed SVMcompare algorithm
outperforms SVMrank when there are equality pairs.
| stat |
Data-dependent compression of random features for large-scale kernel
approximation | Kernel methods offer the flexibility to learn complex relationships in
modern, large data sets while enjoying strong theoretical guarantees on
quality. Unfortunately, these methods typically require cubic running time in
the data set size, a prohibitive cost in the large-data setting. Random feature
maps (RFMs) and the Nystrom method both consider low-rank approximations to the
kernel matrix as a potential solution. But, in order to achieve desirable
theoretical guarantees, the former may require a prohibitively large number of
features J+, and the latter may be prohibitively expensive for high-dimensional
problems. We propose to combine the simplicity and generality of RFMs with a
data-dependent feature selection scheme to achieve desirable theoretical
approximation properties of Nystrom with just O(log J+) features. Our key
insight is to begin with a large set of random features, then reduce them to a
small number of weighted features in a data-dependent, computationally
efficient way, while preserving the statistical guarantees of using the
original large set of features. We demonstrate the efficacy of our method with
theory and experiments--including on a data set with over 50 million
observations. In particular, we show that our method achieves small kernel
matrix approximation error and better test set accuracy with provably fewer
random features than state-of-the-art methods.
| stat |
Simulating space-time random fields with nonseparable Gneiting-type
covariance functions | Two algorithms are proposed to simulate space-time Gaussian random fields
with a covariance function belonging to an extended Gneiting class, the
definition of which depends on a completely monotone function associated with
the spatial structure and a conditionally negative definite function associated
with the temporal structure. In both cases, the simulated random field is
constructed as a weighted sum of cosine waves, with a Gaussian spatial
frequency vector and a uniform phase. The difference lies in the way to handle
the temporal component. The first algorithm relies on a spectral decomposition
in order to simulate a temporal frequency conditional upon the spatial one,
while in the second algorithm the temporal frequency is replaced by an
intrinsic random field whose variogram is proportional to the conditionally
negative definite function associated with the temporal structure. Both
algorithms are scalable as their computational cost is proportional to the
number of space-time locations, which may be unevenly spaced in space and/or in
time. They are illustrated and validated through synthetic examples.
| stat |
Learning Fair Scoring Functions: Bipartite Ranking under ROC-based
Fairness Constraints | Many applications of AI involve scoring individuals using a learned function
of their attributes. These predictive risk scores are then used to take
decisions based on whether the score exceeds a certain threshold, which may
vary depending on the context. The level of delegation granted to such systems
in critical applications like credit lending and medical diagnosis will heavily
depend on how questions of fairness can be answered. In this paper, we study
fairness for the problem of learning scoring functions from binary labeled
data, a classic learning task known as bipartite ranking. We argue that the
functional nature of the ROC curve, the gold standard measure of ranking
accuracy in this context, leads to several ways of formulating fairness
constraints. We introduce general families of fairness definitions based on the
AUC and on ROC curves, and show that our ROC-based constraints can be
instantiated such that classifiers obtained by thresholding the scoring
function satisfy classification fairness for a desired range of thresholds. We
establish generalization bounds for scoring functions learned under such
constraints, design practical learning algorithms and show the relevance our
approach with numerical experiments on real and synthetic data.
| stat |
Bayesian analysis of extreme values in economic indexes and climate
data: Simulation and application | Mixed modeling of extreme values and random effects is relatively unexplored
topic. Computational difficulties in using the maximum likelihood method for
mixed models and the fact that maximum likelihood method uses available data
and does not use the prior information motivate us to use Bayesian method. Our
simulation studies indicate that random effects modeling produces more reliable
estimates when heterogeneity is present. The application of the proposed model
to the climate data and return values of some economic indexes reveals the same
pattern as the simulation results and confirms the usefulness of mixed modeling
of random effects and extremes. As the nature of climate and economic data are
massive and there is always a possibility of missing a considerable part of
data, saving the information included in past data is useful. Our simulation
studies and applications show the benefit of Bayesian method to save the
information from the past data into the posterior distributions of the
parameters to be used as informative prior distributions to fit the future
data. We show that informative prior distributions obtained from the past data
help to estimate the return level in Block Maxima method and Value-at-Risk and
Expected Shortfall in Peak Over Threshold method with less bias than using
uninformative prior distributions.
| stat |
Gene-based Association Analysis for Bivariate Time-to-event Data through
Functional Regression with Copula Models | Several gene-based association tests for time-to-event traits have been
proposed recently, to detect whether a gene region (containing multiple
variants), as a set, is associated with the survival outcome. However, for
bivariate survival outcomes, to the best of our knowledge, there is no
statistical method that can be directly applied for gene-based association
analysis. Motivated by a genetic study to discover gene regions associated with
the progression of a bilateral eye disease, Age-related Macular Degeneration
(AMD), we implement a novel functional regression method under the copula
framework. Specifically, the effects of variants within a gene region are
modeled through a functional linear model, which then contributes to the
marginal survival functions within the copula. Generalized score test and
likelihood ratio test statistics are derived to test for the association
between bivariate survival traits and the genetic region. Extensive simulation
studies are conducted to evaluate the type-I error control and power
performance of the proposed approach, with comparisons to several existing
methods for a single survival trait, as well as the marginal Cox functional
regression model using the robust sandwich estimator for bivariate survival
traits. Finally, we apply our method to a large AMD study, the Age-related Eye
Disease Study (AREDS), to identify gene regions that are associated with AMD
progression.
| stat |
Towards better understanding of meta-features contributions | Meta learning is a difficult problem as the expected performance of a model
is affected by various aspects such as selected hyperparameters, dataset
properties and landmarkers. Existing approaches are focused on searching for
the best model but do not explain how these different aspects contribute to the
performance of a model. To build a new generation of meta-models we need a
deeper understanding of relative importance of meta-features and construction
of better meta-features. In this paper we (1) introduce a method that can be
used to understand how meta-features influence a model performance, (2) discuss
the relative importance of different groups of meta-features, (3) analyse in
detail the most informative hyperparameters that may result in insights for
selection of empirical priors of engineering for hyperparameters. To our
knowledge this is the first paper that uses techniques developed for
eXplainable Artificial Intelligence (XAI) to examine the behaviour of a meta
model.
| stat |
Efficient Causal Inference from Combined Observational and
Interventional Data through Causal Reductions | Unobserved confounding is one of the main challenges when estimating causal
effects. We propose a novel causal reduction method that replaces an arbitrary
number of possibly high-dimensional latent confounders with a single latent
confounder that lives in the same space as the treatment variable without
changing the observational and interventional distributions entailed by the
causal model. After the reduction, we parameterize the reduced causal model
using a flexible class of transformations, so-called normalizing flows. We
propose a learning algorithm to estimate the parameterized reduced model
jointly from observational and interventional data. This allows us to estimate
the causal effect in a principled way from combined data. We perform a series
of experiments on data simulated using nonlinear causal mechanisms and find
that we can often substantially reduce the number of interventional samples
when adding observational training samples without sacrificing accuracy. Thus,
adding observational data may help to more accurately estimate causal effects
even in the presence of unobserved confounders.
| stat |
Multiscale change point detection via gradual bandwidth adjustment in
moving sum processes | A method for the detection of changes in the expectation in univariate
sequences is provided. Moving sum processes are studied. These rely on the
selection of a tuning bandwidth. Here, a framework to overcome bandwidth
selection is presented - the bandwidth adjusts gradually. For that, moving sum
processes are made dependent on both time and the bandwidth: the domain becomes
a triangle. On the triangle, paths are constructed which systematically lead to
change points. An algorithm is provided that estimates change points by
subsequent consideration of paths. Strong consistency for the number and
location of change points is shown. Simulations support estimation precision. A
companion R-package mscp is made available on CRAN.
| stat |
Non-Asymptotic Pure Exploration by Solving Games | Pure exploration (aka active testing) is the fundamental task of sequentially
gathering information to answer a query about a stochastic environment. Good
algorithms make few mistakes and take few samples. Lower bounds (for
multi-armed bandit models with arms in an exponential family) reveal that the
sample complexity is determined by the solution to an optimisation problem. The
existing state of the art algorithms achieve asymptotic optimality by solving a
plug-in estimate of that optimisation problem at each step. We interpret the
optimisation problem as an unknown game, and propose sampling rules based on
iterative strategies to estimate and converge to its saddle point. We apply
no-regret learners to obtain the first finite confidence guarantees that are
adapted to the exponential family and which apply to any pure exploration query
and bandit structure. Moreover, our algorithms only use a best response oracle
instead of fully solving the optimisation problem.
| stat |
Approximate Bayesian inference for multivariate point pattern analysis
in disease mapping | We present a novel approach for the analysis of multivariate case-control
georeferenced data using Bayesian inference in the context of disease mapping,
where the spatial distribution of different types of cancers is analyzed.
Extending other methodology in point pattern analysis, we propose a
log-Gaussian Cox process for point pattern of cases and the controls, which
accounts for risk factors, such as exposure to pollution sources, and includes
a term to measure spatial residual variation.
For each disease, its intensity is modeled on a baseline spatial effect
(estimated from both controls and cases), a disease-specific spatial term and
the effects on covariates that account for risk factors. By fitting these
models the effect of the covariates on the set of cases can be assessed, and
the residual spatial terms can be easily compared to detect areas of high risk
not explained by the covariates.
Three different types of effects to model exposure to pollution sources are
considered. First of all, a fixed effect on the distance to the source. Next,
smooth terms on the distance are used to model non-linear effects by means of a
discrete random walk of order one and a Gaussian process in one dimension with
a Mat\'ern covariance.
Models are fit using the integrated nested Laplace approximation (INLA) so
that the spatial terms are approximated using an approach based on solving
Stochastic Partial Differential Equations (SPDE). Finally, this new framework
is applied to a dataset of three different types of cancer and a set of
controls from Alcal\'a de Henares (Madrid, Spain). Covariates available include
the distance to several polluting industries and socioeconomic indicators. Our
findings point to a possible risk increase due to the proximity to some of
these industries.
| stat |
Low Rank Forecasting | We consider the problem of forecasting multiple values of the future of a
vector time series, using some past values. This problem, and related ones such
as one-step-ahead prediction, have a very long history, and there are a number
of well-known methods for it, including vector auto-regressive models,
state-space methods, multi-task regression, and others. Our focus is on low
rank forecasters, which break forecasting up into two steps: estimating a
vector that can be interpreted as a latent state, given the past, and then
estimating the future values of the time series, given the latent state
estimate. We introduce the concept of forecast consistency, which means that
the estimates of the same value made at different times are consistent. We
formulate the forecasting problem in general form, and focus on linear
forecasters, for which we propose a formulation that can be solved via convex
optimization. We describe a number of extensions and variations, including
nonlinear forecasters, data weighting, the inclusion of auxiliary data, and
additional objective terms. We illustrate our methods with several examples.
| stat |
Multiple Pretext-Task for Self-Supervised Learning via Mixing Multiple
Image Transformations | Self-supervised learning is one of the most promising approaches to learn
representations capturing semantic features in images without any manual
annotation cost. To learn useful representations, a self-supervised model
solves a pretext-task, which is defined by data itself. Among a number of
pretext-tasks, the rotation prediction task (Rotation) achieves better
representations for solving various target tasks despite its simplicity of the
implementation. However, we found that Rotation can fail to capture semantic
features related to image textures and colors. To tackle this problem, we
introduce a learning technique called multiple pretext-task for self-supervised
learning (MP-SSL), which solves multiple pretext-task in addition to Rotation
simultaneously. In order to capture features of textures and colors, we employ
the transformations of image enhancements (e.g., sharpening and solarizing) as
the additional pretext-tasks. MP-SSL efficiently trains a model by leveraging a
Frank-Wolfe based multi-task training algorithm. Our experimental results show
MP-SSL models outperform Rotation on multiple standard benchmarks and achieve
state-of-the-art performance on Places-205.
| stat |
Embeddings and Attention in Predictive Modeling | We explore in depth how categorical data can be processed with embeddings in
the context of claim severity modeling. We develop several models that range in
complexity from simple neural networks to state-of-the-art attention based
architectures that utilize embeddings. We illustrate the utility of learned
embeddings from neural networks as pretrained features in generalized linear
models, and discuss methods for visualizing and interpreting embeddings.
Finally, we explore how attention based models can contextually augment
embeddings, leading to enhanced predictive performance.
| stat |
Survival Analysis of Organizational Networks -- An Exploratory Study | Organizations interact with the environment and with other organizations, and
these interactions constitute an important way of learning and evolution. To
overcome the problems that they face during their existence, organizations must
certainly adopt survival strategies, both individually and in group. The aim of
this study is to evaluate the effect of a set of prognostic factors
(organizational, size, collaborate strategies, etc.) in the survival of
organizational networks. Statistical methods for time to event data were used
to analyze the data. We have used the Kaplan-Meier product-limit method to
compute and plot estimates of survival, while hypothesis tests were used to
compare survival times across several groups. Regression models were used to
study the effect of continuous predictors as well as to test multiple
predictors at once. Since violations of the proportional hazards were found for
several predictors, accelerated failure time models were used to study the
effect of explanatory variables on network survival.
| stat |
A Convex Programming Solution Based Debiased Estimator for Quantile with
Missing Response and High-dimensional Covariables | This paper is concerned with the estimating problem of response quantile with
high dimensional covariates when response is missing at random. Some existing
methods define root-n consistent estimators for the response quantile. But
these methods require correct specifications of both the conditional
distribution of response given covariates and the selection probability
function. In this paper, a debiased method is proposed by solving a convex
programming. The estimator obtained by the proposed method is asymptotically
normal given a correctly specified parametric model for the condition
distribution function, without the requirement to specify and estimate the
selection probability function. Moreover, the proposed estimator is
asymptotically more efficient than the existing estimators. The proposed method
is evaluated by a simulation study and is illustrated by a real data example.
| stat |
Multi-stage models for the failure of complex systems, cascading
disasters, and the onset of disease | Complex systems can fail through different routes, often progressing through
a series of (rate-limiting) steps and modified by environmental exposures. The
onset of disease, cancer in particular, is no different. Multi-stage models
provide a simple but very general mathematical framework for studying the
failure of complex systems, or equivalently, the onset of disease. They include
the Armitage Doll multi-stage cancer model as a particular case, and have
potential to provide new insights into how failures and disease, arise and
progress. A method described by E.T. Jaynes is developed to provide an
analytical solution for a large class of these models, and highlights
connections between the convolution of Laplace transforms, sums of random
variables, and Schwinger/Feynman parameterisations. Examples include: exact
solutions to the Armitage-Doll model, the sum of Gamma-distributed variables
with integer-valued shape parameters, a clonal-growth cancer model, and a model
for cascading disasters. Applications and limitations of the approach are
discussed in the context of recent cancer research. The model is sufficiently
general to be used in many contexts, such as engineering, project management,
disease progression, and disaster risk for example, allowing the estimation of
failure rates in complex systems and projects. The intended result is a
mathematical toolkit for applying multi-stage models to the study of failure
rates in complex systems and to the onset of disease, cancer in particular.
| stat |
A Self-supervised Approach to Hierarchical Forecasting with Applications
to Groupwise Synthetic Controls | When forecasting time series with a hierarchical structure, the existing
state of the art is to forecast each time series independently, and, in a
post-treatment step, to reconcile the time series in a way that respects the
hierarchy (Hyndman et al., 2011; Wickramasuriya et al., 2018). We propose a new
loss function that can be incorporated into any maximum likelihood objective
with hierarchical data, resulting in reconciled estimates with confidence
intervals that correctly account for additional uncertainty due to imperfect
reconciliation. We evaluate our method using a non-linear model and synthetic
data on a counterfactual forecasting problem, where we have access to the
ground truth and contemporaneous covariates, and show that we largely improve
over the existing state-of-the-art method.
| stat |
Nearly Optimal Variational Inference for High Dimensional Regression
with Shrinkage Priors | We propose a variational Bayesian (VB) procedure for high-dimensional linear
model inferences with heavy tail shrinkage priors, such as student-t prior.
Theoretically, we establish the consistency of the proposed VB method and prove
that under the proper choice of prior specifications, the contraction rate of
the VB posterior is nearly optimal. It justifies the validity of VB inference
as an alternative of Markov Chain Monte Carlo (MCMC) sampling. Meanwhile,
comparing to conventional MCMC methods, the VB procedure achieves much higher
computational efficiency, which greatly alleviates the computing burden for
modern machine learning applications such as massive data analysis. Through
numerical studies, we demonstrate that the proposed VB method leads to shorter
computing time, higher estimation accuracy, and lower variable selection error
than competitive sparse Bayesian methods.
| stat |
Critical Risk Indicators (CRIs) for the electric power grid: A survey
and discussion of interconnected effects | The electric power grid is a critical societal resource connecting multiple
infrastructural domains such as agriculture, transportation, and manufacturing.
The electrical grid as an infrastructure is shaped by human activity and public
policy in terms of demand and supply requirements. Further, the grid is subject
to changes and stresses due to solar weather, climate, hydrology, and ecology.
The emerging interconnected and complex network dependencies make such
interactions increasingly dynamic causing potentially large swings, thus
presenting new challenges to manage the coupled human-natural system. This
paper provides a survey of models and methods that seek to explore the
significant interconnected impact of the electric power grid and interdependent
domains. We also provide relevant critical risk indicators (CRIs) across
diverse domains that may influence electric power grid risks, including
climate, ecology, hydrology, finance, space weather, and agriculture. We
discuss the convergence of indicators from individual domains to explore
possible systemic risk, i.e., holistic risk arising from cross-domains
interconnections. Our study provides an important first step towards
data-driven analysis and predictive modeling of risks in the coupled
interconnected systems. Further, we propose a compositional approach to risk
assessment that incorporates diverse domain expertise and information, data
science, and computer science to identify domain-specific CRIs and their union
in systemic risk indicators.
| stat |
Learning Decomposed Representation for Counterfactual Inference | One fundamental problem in the learning treatment effect from observational
data is confounder identification and balancing. Most of the previous methods
realized confounder balancing by treating all observed variables as
confounders, ignoring the identification of confounders and non-confounders. In
general, not all the observed variables are confounders which are the common
causes of both the treatment and the outcome, some variables only contribute to
the treatment and some contribute to the outcome. Balancing those
non-confounders would generate additional bias for treatment effect estimation.
By modeling the different relations among variables, treatment and outcome, we
propose a synergistic learning framework to 1) identify and balance confounders
by learning decomposed representation of confounders and non-confounders, and
simultaneously 2) estimate the treatment effect in observational studies via
counterfactual inference. Our empirical results demonstrate that the proposed
method can precisely identify and balance confounders, while the estimation of
the treatment effect performs better than the state-of-the-art methods on both
synthetic and real-world datasets.
| stat |
Improved log-Gaussian approximation for over-dispersed Poisson
regression: application to spatial analysis of COVID-19 | In the era of open data, Poisson and other count regression models are
increasingly important. Provided this, we develop a closed-form inference for
an over-dispersed Poisson regression, especially for (over-dispersed) Bayesian
Poisson wherein the exact inference is unobtainable. The approach is derived
via mode-based log-Gaussian approximation. Unlike closed-form alternatives, it
remains accurate even for zero-inflated count data. Besides, our approach has
no arbitrary parameter that must be determined a priori. Monte Carlo
experiments demonstrate that the estimation error of the proposed method is a
considerably smaller estimation error than the closed-form alternatives and as
small as the usual Poisson regressions. We obtained similar results in the case
of Poisson additive mixed modeling considering spatial or group effects. The
developed method was applied for analyzing COVID-19 data in Japan. This result
suggests that influences of pedestrian density, age, and other factors on the
number of cases change over periods.
| stat |
Nonlinear ICA Using Auxiliary Variables and Generalized Contrastive
Learning | Nonlinear ICA is a fundamental problem for unsupervised representation
learning, emphasizing the capacity to recover the underlying latent variables
generating the data (i.e., identifiability). Recently, the very first
identifiability proofs for nonlinear ICA have been proposed, leveraging the
temporal structure of the independent components. Here, we propose a general
framework for nonlinear ICA, which, as a special case, can make use of temporal
structure. It is based on augmenting the data by an auxiliary variable, such as
the time index, the history of the time series, or any other available
information. We propose to learn nonlinear ICA by discriminating between true
augmented data, or data in which the auxiliary variable has been randomized.
This enables the framework to be implemented algorithmically through logistic
regression, possibly in a neural network. We provide a comprehensive proof of
the identifiability of the model as well as the consistency of our estimation
method. The approach not only provides a general theoretical framework
combining and generalizing previously proposed nonlinear ICA models and
algorithms, but also brings practical advantages.
| stat |
Extending the scope of the small-ball method | The small-ball method was introduced as a way of obtaining a high
probability, isomorphic lower bound on the quadratic empirical process, under
weak assumptions on the indexing class. The key assumption was that class
members satisfy a uniform small-ball estimate: that $Pr(|f| \geq
\kappa\|f\|_{L_2}) \geq \delta$ for given constants $\kappa$ and $\delta$.
Here we extend the small-ball method and obtain a high probability,
almost-isometric (rather than isomorphic) lower bound on the quadratic
empirical process. The scope of the result is considerably wider than the
small-ball method: there is no need for class members to satisfy a uniform
small-ball condition, and moreover, motivated by the notion of tournament
learning procedures, the result is stable under a `majority vote'.
| stat |
Handling Hard Affine SDP Shape Constraints in RKHSs | Shape constraints, such as non-negativity, monotonicity, convexity or
supermodularity, play a key role in various applications of machine learning
and statistics. However, incorporating this side information into predictive
models in a hard way (for example at all points of an interval) for rich
function classes is a notoriously challenging problem. We propose a unified and
modular convex optimization framework, relying on second-order cone (SOC)
tightening, to encode hard affine SDP constraints on function derivatives, for
models belonging to vector-valued reproducing kernel Hilbert spaces (vRKHSs).
The modular nature of the proposed approach allows to simultaneously handle
multiple shape constraints, and to tighten an infinite number of constraints
into finitely many. We prove the consistency of the proposed scheme and that of
its adaptive variant, leveraging geometric properties of vRKHSs. The efficiency
of the approach is illustrated in the context of shape optimization,
safety-critical control and econometrics.
| stat |
Generalized Evolutionary Point Processes: Model Specifications and Model
Comparison | Generalized evolutionary point processes offer a class of point process
models that allows for either excitation or inhibition based upon the history
of the process. In this regard, we propose modeling which comprises
generalization of the nonlinear Hawkes process. Working within a Bayesian
framework, model fitting is implemented through Markov chain Monte Carlo. This
entails discussion of computation of the likelihood for such point patterns.
Furthermore, for this class of models, we discuss strategies for model
comparison. Using simulation, we illustrate how well we can distinguish these
models from point pattern specifications with conditionally independent event
times, e.g., Poisson processes. Specifically, we demonstrate that these models
can correctly identify true relationships (i.e., excitation or
inhibition/control). Then, we consider a novel extension of the log Gaussian
Cox process that incorporates evolutionary behavior and illustrate that our
model comparison approach prefers the evolutionary log Gaussian Cox process
compared to simpler models. We also examine a real dataset consisting of
violent crime events from the 11th police district in Chicago from the year
2018. This data exhibits strong daily seasonality and changes across the year.
After we account for these data attributes, we find significant but mild
self-excitation, implying that event occurrence increases the intensity of
future events.
| stat |
Kernel learning approaches for summarising and combining posterior
similarity matrices | When using Markov chain Monte Carlo (MCMC) algorithms to perform inference
for Bayesian clustering models, such as mixture models, the output is typically
a sample of clusterings (partitions) drawn from the posterior distribution. In
practice, a key challenge is how to summarise this output. Here we build upon
the notion of the posterior similarity matrix (PSM) in order to suggest new
approaches for summarising the output of MCMC algorithms for Bayesian
clustering models. A key contribution of our work is the observation that PSMs
are positive semi-definite, and hence can be used to define
probabilistically-motivated kernel matrices that capture the clustering
structure present in the data. This observation enables us to employ a range of
kernel methods to obtain summary clusterings, and otherwise exploit the
information summarised by PSMs. For example, if we have multiple PSMs, each
corresponding to a different dataset on a common set of statistical units, we
may use standard methods for combining kernels in order to perform integrative
clustering. We may moreover embed PSMs within predictive kernel models in order
to perform outcome-guided data integration. We demonstrate the performances of
the proposed methods through a range of simulation studies as well as two real
data applications. R code is available at
https://github.com/acabassi/combine-psms.
| stat |
Routine Hospital-based SARS-CoV-2 Testing Outperforms State-based Data
in Predicting Clinical Burden | Throughout the COVID-19 pandemic, government policy and healthcare
implementation responses have been guided by reported positivity rates and
counts of positive cases in the community. The selection bias of these data
calls into question their validity as measures of the actual viral incidence in
the community and as predictors of clinical burden. In the absence of any
successful public or academic campaign for comprehensive or random testing, we
have developed a proxy method for synthetic random sampling, based on viral RNA
testing of patients who present for elective procedures within a hospital
system. We present here an approach under multilevel regression and
poststratification (MRP) to collecting and analyzing data on viral exposure
among patients in a hospital system and performing statistical adjustment that
has been made publicly available to estimate true viral incidence and trends in
the community. We apply our MRP method to track viral behavior in a mixed
urban-suburban-rural setting in Indiana. This method can be easily implemented
in a wide variety of hospital settings. Finally, we provide evidence that this
model predicts the clinical burden of SARS-CoV-2 earlier and more accurately
than currently accepted metrics.
| stat |
Extended Sensitivity Analysis for Heterogeneous Unmeasured Confounding
with An Application to Sibling Studies of Returns to Education | The conventional model for assessing insensitivity to hidden bias in paired
observational studies constructs a worst-case distribution for treatment
assignments subject to bounds on the maximal bias to which any given pair is
subjected. In studies where rare cases of extreme hidden bias are suspected,
the maximal bias may be substantially larger than the typical bias across
pairs, such that a correctly specified bound on the maximal bias would yield an
unduly pessimistic perception of the study's robustness to hidden bias. We
present an extended sensitivity analysis which allows researchers to
simultaneously bound the maximal and typical bias perturbing the pairs under
investigation while maintaining the desired Type I error rate. We motivate and
illustrate our method with two sibling studies on the impact of schooling on
earnings, one containing information of cognitive ability of siblings and the
other not. Cognitive ability, clearly influential of both earnings and degree
of schooling, is likely similar between members of most sibling pairs yet
could, conceivably, vary drastically for some siblings. The method is
straightforward to implement, simply requiring the solution to a quadratic
program.
| stat |
Gibbsian T-tessellation model for agricultural landscape
characterization | A new class of planar tessellations, named T-tessellations, was introduced in
([10]). A model was proposed to be considered as a completely random
T-tessellation model (CRTT) and its Gibbsian variants were discussed. A general
simulation algorithm of Metropolis-Hastings-Green type was derived for model
simulation, involving three local transformations of T-tessellations. The
current paper focuses on statistical inference for Gibbs models of
T-tessellations. Statistical methods originated from point pattern analysis are
implemented on the example of three agricultural landscapes approximated by
T-tessellations. The choice of model statistics is guided by their capacity to
highlight the differences between the landscape patterns. Model parameters are
estimated by Monte Carlo Maximum Likelihood method, yielding a baseline for
landscapes comparison. In the last part of the paper a global envelope test
based on the empty-space function is proposed for assessing the goodness-of-fit
of the model.
| stat |
Contraction of a quasi-Bayesian model with shrinkage priors in precision
matrix estimation | Currently several Bayesian approaches are available to estimate large sparse
precision matrices, including Bayesian graphical Lasso (Wang, 2012), Bayesian
structure learning (Banerjee and Ghosal, 2015), and graphical horseshoe (Li et
al., 2019). Although these methods have exhibited nice empirical performances,
in general they are computationally expensive. Moreover, we have limited
knowledge about the theoretical properties, e.g., posterior contraction rate,
of graphical Bayesian Lasso and graphical horseshoe. In this paper, we propose
a new method that integrates some commonly used continuous shrinkage priors
into a quasi-Bayesian framework featured by a pseudo-likelihood. Under mild
conditions, we establish an optimal posterior contraction rate for the proposed
method. Compared to existing approaches, our method has two main advantages.
First, our method is computationally more efficient while achieving similar
error rate; second, our framework is more amenable to theoretical analysis.
Extensive simulation experiments and the analysis on a real data set are
supportive of our theoretical results.
| stat |
Covid-19 risk factors: Statistical learning from German healthcare
claims data | We analyse prior risk factors for severe, critical or fatal courses of
Covid-19 based on a retrospective cohort using claims data of the AOK Bayern.
As our main methodological contribution, we avoid prior grouping and
pre-selection of candidate risk factors. Instead, fine-grained hierarchical
information from medical classification systems for diagnoses, pharmaceuticals
and procedures are used, resulting in more than 33,000 covariates. Our approach
has better predictive ability than well-specified morbidity groups but does not
need prior subject-matter knowledge. The methodology and estimated coefficients
are made available to decision makers to prioritize protective measures towards
vulnerable subpopulations and to researchers who like to adjust for a large set
of confounders in studies of individual risk factors.
| stat |
Significance tests of feature relevance for a blackbox learner | An exciting recent development is the uptake of deep learning in many
scientific fields, where the objective is seeking novel scientific insights and
discoveries. To interpret a learning outcome, researchers perform hypothesis
testing for explainable features to advance scientific domain knowledge. In
such a situation, testing for a blackbox learner poses a severe challenge
because of intractable models, unknown limiting distributions of parameter
estimates, and high computational constraints. In this article, we derive two
consistent tests for the feature relevance of a blackbox learner. The first one
evaluates a loss difference with perturbation on an inference sample, which is
independent of an estimation sample used for parameter estimation in model
fitting. The second further splits the inference sample into two but does not
require data perturbation. Also, we develop their combined versions by
aggregating the order statistics of the $p$-values based on repeated sample
splitting. To estimate the splitting ratio and the perturbation size, we
develop adaptive splitting schemes for suitably controlling the Type \rom{1}
error subject to computational constraints. By deflating the
\textit{bias-sd-ratio}, we establish asymptotic null distributions of the test
statistics and their consistency in terms of statistical power. Our theoretical
power analysis and simulations indicate that the one-split test is more
powerful than the two-split test, though the latter is easier to apply for
large datasets. Moreover, the combined tests are more stable while compensating
for a power loss by repeated sample splitting. Numerically, we demonstrate the
utility of the proposed tests on two benchmark examples. Accompanying this
paper is our Python library {\tt dnn-inference}
https://dnn-inference.readthedocs.io/en/latest/ that implements the proposed
tests.
| stat |
Generalized Bayes Quantification Learning under Dataset Shift | Quantification learning is the task of prevalence estimation for a test
population using predictions from a classifier trained on a different
population. Quantification methods assume that the sensitivities and
specificities of the classifier are either perfect or transportable from the
training to the test population. These assumptions are inappropriate in the
presence of dataset shift, when the misclassification rates in the training
population are not representative of those for the test population.
Quantification under dataset shift has been addressed only for single-class
(categorical) predictions and assuming perfect knowledge of the true labels on
a small subset of the test population. We propose generalized Bayes
quantification learning (GBQL) that uses the entire compositional predictions
from probabilistic classifiers and allows for uncertainty in true class labels
for the limited labeled test data. Instead of positing a full model, we use a
model-free Bayesian estimating equation approach to compositional data based
only on a first-moment assumption. The idea will be useful in Bayesian
compositional data analysis in general as it is robust to different generating
mechanisms for compositional data and includes categorical outputs as a special
case. We show how our method yields existing quantification approaches as
special cases. Extension to an ensemble GBQL that uses predictions from
multiple classifiers yielding inference robust to inclusion of a poor
classifier is discussed. We outline a fast and efficient Gibbs sampler using a
rounding and coarsening approximation to the loss functions. We also establish
posterior consistency, asymptotic normality and valid coverage of interval
estimates from GBQL, as well as finite sample posterior concentration rate.
Empirical performance of GBQL is demonstrated through simulations and analysis
of real data with evident dataset shift.
| stat |
Hybrid Machine Learning Forecasts for the FIFA Women's World Cup 2019 | In this work, we combine two different ranking methods together with several
other predictors in a joint random forest approach for the scores of soccer
matches. The first ranking method is based on the bookmaker consensus, the
second ranking method estimates adequate ability parameters that reflect the
current strength of the teams best. The proposed combined approach is then
applied to the data from the two previous FIFA Women's World Cups 2011 and
2015. Finally, based on the resulting estimates, the FIFA Women's World Cup
2019 is simulated repeatedly and winning probabilities are obtained for all
teams. The model clearly favors the defending champion USA before the host
France.
| stat |
A Bayesian Approach to Modelling Longitudinal Data in Electronic Health
Records | Analyzing electronic health records (EHR) poses significant challenges
because often few samples are available describing a patient's health and, when
available, their information content is highly diverse. The problem we consider
is how to integrate sparsely sampled longitudinal data, missing measurements
informative of the underlying health status and fixed demographic information
to produce estimated survival distributions updated through a patient's follow
up. We propose a nonparametric probabilistic model that generates survival
trajectories from an ensemble of Bayesian trees that learns variable
interactions over time without specifying beforehand the longitudinal process.
We show performance improvements on Primary Biliary Cirrhosis patient data.
| stat |
Comprehensive decision-strategy space exploration for efficient
territorial planning strategies | GIS-based Multi-Criteria Decision Analysis is a well-known decision support
tool that can be used in a wide variety of contexts. It is particularly useful
for territorial planning in situations where several actors with different, and
sometimes contradictory, point of views have to take a decision regarding land
use development. While the impact of the weights used to represent the relative
importance of criteria has been widely studied in the recent literature, the
impact of the order weights used to combine the criteria have rarely been
investigated. This paper presents a spatial sensitivity analysis to assess the
impact of order weights determination in GIS-based Multi-Criteria Analysis by
Ordered Weighted Averaging. We propose a methodology based on an efficient
exploration of the decision-strategy space defined by the level of risk and
trade-off in the decision process. We illustrate our approach with a land use
planning process in the South of France. The objective is to find suitable
areas for urban development while preserving green areas and their associated
ecosystem services. The ecosystem service approach has indeed the potential to
widen the scope of traditional landscape-ecological planning by including
ecosystem-based benefits, including social and economic benefits, green
infrastructures and biophysical parameters in urban and territorial planning.
We show that in this particular case the decision-strategy space can be divided
into four clusters. Each of them is associated with a map summarizing the
average spatial suitability distribution used to identify potential areas for
urban development. We also demonstrate the pertinence of a spatial variance
within-cluster analysis to disentangle the relationship between risk and
trade-off values. At the end, we perform a site suitability ranking analysis to
assess the relationship between the four detected clusters.
| stat |
F-measure Maximizing Logistic Regression | Logistic regression is a widely used method in several fields. When applying
logistic regression to imbalanced data, for which majority classes dominate
over minority classes, all class labels are estimated as `majority class.' In
this article, we use an F-measure optimization method to improve the
performance of logistic regression applied to imbalanced data. While many
F-measure optimization methods adopt a ratio of the estimators to approximate
the F-measure, the ratio of the estimators tends to have more bias than when
the ratio is directly approximated. Therefore, we employ an approximate
F-measure for estimating the relative density ratio. In addition, we define a
relative F-measure and approximate the relative F-measure. We show an algorithm
for a logistic regression weighted approximated relative to the F-measure. The
experimental results using real world data demonstrated that our proposed
method is an efficient algorithm to improve the performance of logistic
regression applied to imbalanced data.
| stat |
Model-based clustering for identifying disease-associated SNPs in
case-control genome-wide association studies | Genome-wide association studies (GWASs) aim to detect genetic risk factors
for complex human diseases by identifying disease-associated single-nucleotide
polymorphisms (SNPs). The traditional SNP-wise approach along with multiple
testing adjustment is over-conservative and lack of power in many GWASs. In
this article, we proposed a model-based clustering method that transforms the
challenging high-dimension-small-sample-size problem to
low-dimension-large-sample-size problem and borrows information across SNPs by
grouping SNPs into three clusters. We pre-specify the patterns of clusters by
minor allele frequencies of SNPs between cases and controls, and enforce the
patterns with prior distributions. In the simulation studies our proposed novel
model outperform traditional SNP-wise approach by showing better controls of
false discovery rate (FDR) and higher sensitivity. We re-analyzed two real
studies to identifying SNPs associated with severe bortezomib-induced
peripheral neuropathy (BiPN) in patients with multiple myeloma (MM). The
original analysis in the literature failed to identify SNPs after FDR
adjustment. Our proposed method not only detected the reported SNPs after FDR
adjustment but also discovered a novel BiPN-associated SNP rs4351714 that has
been reported to be related to MM in another study.
| stat |
Some Cautionary Comments on Principal Component Analysis for Time Series
Data | Principal component analysis (PCA) is a most frequently used statistical tool
in almost all branches of data science. However, like many other statistical
tools, there is sometimes the risk of misuse or even abuse. In this short note,
we highlight possible pitfalls in using the theoretical results of PCA based on
the assumption of independent data when the data are time series. For the
latter, we state a central limit theorem of the eigenvalues and eigenvectors,
give analytical and bootstrap methods to estimate the covariance, and assess
their efficacy via simulation. An empirical example is given to illustrate the
pitfalls of a common misuse of PCA. We conclude that while the conventional
scree plot continues to be useful for time series data, the interpretation of
the principal component loadings requires careful attention.
| stat |
Clarifying causal mediation analysis for the applied researcher: Effect
identification via three assumptions and five potential outcomes | Causal mediation analysis is complicated with multiple effect definitions
that require different sets of assumptions for identification. This paper
provides a systematic explanation of such assumptions. We define five potential
outcome types whose means are involved in various effect definitions. We tackle
their mean/distribution's identification, starting with the one that requires
the weakest assumptions and gradually building up to the one that requires the
strongest assumptions. This presentation shows clearly why an assumption is
required for one estimand and not another, and provides a succinct table from
which an applied researcher could pick out the assumptions required for
identifying the causal effects they target. Using a running example, the paper
illustrates the assembling and consideration of identifying assumptions for a
range of causal contrasts. For several that are commonly encountered in the
literature, this exercise clarifies that identification requires weaker
assumptions than those often stated in the literature. This attention to the
details also draws attention to the differences in the positivity assumption
for different estimands, with practical implications. Clarity on the
identifying assumptions of these various estimands will help researchers
conduct appropriate mediation analyses and interpret the results with
appropriate caution given the plausibility of the assumptions.
| stat |
Explicit Estimation of Derivatives from Data and Differential Equations
by Gaussian Process Regression | In this work, we employ the Bayesian inference framework to solve the problem
of estimating the solution and particularly, its derivatives, which satisfy a
known differential equation, from the given noisy and scarce observations of
the solution data only. To address the key issue of accuracy and robustness of
derivative estimation, we use the Gaussian processes to jointly model the
solution, the derivatives, and the differential equation. By regarding the
linear differential equation as a linear constraint, a Gaussian process
regression with constraint method (GPRC) is developed to improve the accuracy
of prediction of derivatives. For nonlinear differential equations, we propose
a Picard-iteration-like approximation of linearization around the Gaussian
process obtained only from data so that our GPRC can be still iteratively
applicable. Besides, a product of experts method is applied to ensure the
initial or boundary condition is considered to further enhance the prediction
accuracy of the derivatives. We present several numerical results to illustrate
the advantages of our new method in comparison to the standard data-driven
Gaussian process regression.
| stat |
Detecting abnormal connectivity in schizophrenia via a joint directed
acyclic graph estimation model | Functional connectivity (FC) has been widely used to study brain network
interactions underlying the emerging cognition and behavior of an individual.
FC is usually defined as the correlation or partial correlation between brain
regions. Although FC is proved to be a good starting point to understand the
brain organization, it fails to tell the causal relationship or the direction
of interactions. Many directed acyclic graph (DAG) based methods were applied
to study the directed interactions using functional magnetic resonance imaging
(fMRI) data but the performance was severely limited by the small sample size
and high dimensionality, hindering its applications. To overcome the obstacles,
we propose a score based joint directed acyclic graph model to estimate the
directed FC in fMRI data. Instead of using a combinatorial optimization
framework, the structure of DAG is characterized with an algebra equation and
further regularized with sparsity and group similarity terms. The simulation
results have demonstrated the improved accuracy of the proposed model in
detecting causality as compared to other existing methods. In our case-control
study of the MIND Clinical Imaging Consortium (MCIC) data, we have successfully
identified decreased functional integration, disrupted hub structures and
characteristic edges (CtEs) in schizophrenia (SZ) patients. Further comparison
between the results from directed FC and undirected FC illustrated the their
different emphasis on selected features. We speculate that combining the
features from undirected graphical model and directed graphical model might be
a promising way to do FC analysis.
| stat |
Reverse KL-Divergence Training of Prior Networks: Improved Uncertainty
and Adversarial Robustness | Ensemble approaches for uncertainty estimation have recently been applied to
the tasks of misclassification detection, out-of-distribution input detection
and adversarial attack detection. Prior Networks have been proposed as an
approach to efficiently \emph{emulate} an ensemble of models for classification
by parameterising a Dirichlet prior distribution over output distributions.
These models have been shown to outperform alternative ensemble approaches,
such as Monte-Carlo Dropout, on the task of out-of-distribution input
detection. However, scaling Prior Networks to complex datasets with many
classes is difficult using the training criteria originally proposed. This
paper makes two contributions. First, we show that the appropriate training
criterion for Prior Networks is the \emph{reverse} KL-divergence between
Dirichlet distributions. This addresses issues in the nature of the training
data target distributions, enabling prior networks to be successfully trained
on classification tasks with arbitrarily many classes, as well as improving
out-of-distribution detection performance. Second, taking advantage of this new
training criterion, this paper investigates using Prior Networks to detect
adversarial attacks and proposes a generalized form of adversarial training. It
is shown that the construction of successful \emph{adaptive} whitebox attacks,
which affect the prediction and evade detection, against Prior Networks trained
on CIFAR-10 and CIFAR-100 using the proposed approach requires a greater amount
of computational effort than against networks defended using standard
adversarial training or MC-dropout.
| stat |
Robust and Efficient Semi-Supervised Estimation of Average Treatment
Effects with Application to Electronic Health Records Data | We consider the problem of estimating the average treatment effect (ATE) in a
semi-supervised learning setting, where a very small proportion of the entire
set of observations are labeled with the true outcome but features predictive
of the outcome are available among all observations. This problem arises, for
example, when estimating treatment effects in electronic health records (EHR)
data because gold-standard outcomes are often not directly observable from the
records but are observed for a limited number of patients through small-scale
manual chart review. We develop an imputation-based approach for estimating the
ATE that is robust to misspecification of the imputation model. This
effectively allows information from the predictive features to be safely
leveraged to improve efficiency in estimating the ATE. The estimator is
additionally doubly-robust in that it is consistent under correct specification
of either an initial propensity score model or a baseline outcome model. It is
also locally semiparametric efficient under an ideal semi-supervised model
where the distribution of the unlabeled data is known. Simulations exhibit the
efficiency and robustness of the proposed method compared to existing
approaches in finite samples.We illustrate the method by comparing rates of
treatment response to two biologic agents for treatment inflammatory bowel
disease using EHR data from Partner's Healthcare.
| stat |
Identify More, Observe Less: Mediation Analysis Synthetic Control | The synthetic control method (SCM) allows estimation of the causal effect of
an intervention in settings where panel data on just a few treated units and
control units are available. We show that the existing SCM as well as its
extensions can be easily modified to estimate how much of the "total" effect
goes through observed causal channels. The additional assumptions needed are
arguably very mild in many settings. Furthermore, in an illustrative empirical
application we estimate the effects of adopting the euro on labor productivity
in several countries and show that a reduction in the Economic Complexity Index
helped to mitigate the negative short run effects of adopting the new currency
in some countries and boosted the positive effects in others.
| stat |
missSBM: An R Package for Handling Missing Values in the Stochastic
Block Model | The Stochastic Block Model (SBM) is a popular probabilistic model for random
graphs. It is commonly used to perform clustering on network data by
aggregating nodes that share similar connectivity patterns into blocks. When
fitting an SBM to a network which is partially observed, it is important to
take into account the underlying process that originates the missing values,
otherwise the inference may be biased. This paper introduces missSBM, an
R-package fitting the SBM when the network is partially observed, i.e., the
adjacency matrix contains not only 1 or 0 encoding presence or absence of edges
but also NA encoding missing information between pairs of nodes. It implements
a series of algorithms for fitting the binary SBM, possibly in the presence of
external covariates, by performing variational inference for several
observation processes. Our implementation automatically explores different
block numbers to select the most relevant according to the Integrated
Classification Likelihood (ICL) criterion. The ICL criterion can also help
determine which observation process better corresponds to a given dataset.
Finally, missSBM can be used to perform imputation of missing entries in the
adjacency matrix. We illustrate the package on a network data set consisting in
interactions between political blogs sampled during the French presidential
election in 2007.
| stat |
Epidemic changepoint detection in the presence of nuisance changes | Many time series problems feature epidemic changes - segments where a
parameter deviates from a background baseline. The number and location of such
changes can be estimated in a principled way by existing detection methods,
providing that the background level is stable and known. However, practical
data often contains nuisance changes in background level, which interfere with
standard estimation techniques. Furthermore, such changes often differ from the
target segments only in duration, and appear as false alarms in the detection
results. To solve these issues, we propose a two-level detector that models and
separates nuisance and signal changes. As part of this method, we developed a
new, efficient approach to simultaneously estimate unknown, but fixed,
background level and detect epidemic changes. The analytic and computational
properties of the proposed methods are established, including consistency and
convergence. We demonstrate via simulations that our two-level detector
provides accurate estimation of changepoints under a nuisance process, while
other state-of-the-art detectors fail. Using real-world genomic and demographic
datasets, we demonstrate that our method can identify and localise target
events while separating out seasonal variations and experimental artefacts.
| stat |
A New Algorithm for Convex Biclustering and Its Extension to the
Compositional Data | Biclustering is a powerful data mining technique that allows simultaneously
clustering rows (observations) and columns (features) in a matrix-format data
set, which can provide results in a checkerboard-like pattern for visualization
and exploratory analysis in a wide array of domains. Multiple biclustering
algorithms have been developed in the past two decades, among which the convex
biclustering can guarantee a global optimum by formulating in as a convex
optimization problem. On the other hand, the application of biclustering has
not progressed in parallel with the algorithm design. For example, biclustering
for increasingly popular microbiome research data is under-applied, and one
reason may be its compositional constraints. In this manuscript, we propose a
new convex biclustering algorithm, called the bi-ADMM, under general setups
based on the ADMM algorithm, which is free of extra smoothing steps to obtain
informative biclusters required by existing convex biclustering algorithms.
Furthermore, we tailor it to the algorithm named biC-ADMM specifically to
tackle compositional constraints confronted in microbiome data. The key step of
our methods utilizes the Sylvester Equation, which is new to the clustering
research. The effectiveness of the proposed methods is examined through a
variety of numerical experiments and a microbiome data application.
| stat |
A quadratic linear-parabolic model-based classification to detect
epileptic EEG seizures | The two-point central difference is a common algorithm in biological signal
processing and is particularly useful in analyzing physiological signals. In
this paper, we develop a model-based classification method to detect epileptic
seizures that relies on this algorithm to filter EEG signals. The underlying
idea is to design an EEG filter that enhances the waveform of epileptic
signals. The filtered signal is fitted to a quadratic linear-parabolic model
using the curve fitting technique. The model fitting is assessed using four
statistical parameters, which are used as classification features with a random
forest algorithm to discriminate seizure and non-seizure events. The proposed
method was applied to 66 epochs from the Children Hospital Boston database.
Results show that the method achieves fast and accurate detection of epileptic
seizures, with a 92% sensitivity, 96% specificity, and 94.1% accuracy.
| stat |
Perfect Sampling of Multivariate Hawkes Process | As an extension of self-exciting Hawkes process, the multivariate Hawkes
process models counting processes of different types of random events with
mutual excitement. In this paper, we present a perfect sampling algorithm that
can generate i.i.d. stationary sample paths of multivariate Hawkes process
without any transient bias. In addition, we provide an explicit expression of
algorithm complexity in model and algorithm parameters and provide numerical
schemes to find the optimal parameter set that minimizes the complexity of the
perfect sampling algorithm.
| stat |
M-type penalized splines with auxiliary scale estimation | Penalized spline smoothing is a popular and flexible method of obtaining
estimates in nonparametric regression but the classical least-squares criterion
is highly susceptible to model deviations and atypical observations. Penalized
spline estimation with a resistant loss function is a natural remedy, yet to
this day the asymptotic properties of M-type penalized spline estimators have
not been studied. We show in this paper that M-type penalized spline estimators
achieve the same rates of convergence as their least-squares counterparts, even
with auxiliary scale estimation. We further find theoretical justification for
the use of a small number of knots relative to the sample size. We illustrate
the benefits of M-type penalized splines in a Monte-Carlo study and two
real-data examples, which contain atypical observations.
| stat |
Statistical Adaptive Stochastic Gradient Methods | We propose a statistical adaptive procedure called SALSA for automatically
scheduling the learning rate (step size) in stochastic gradient methods. SALSA
first uses a smoothed stochastic line-search procedure to gradually increase
the learning rate, then automatically switches to a statistical method to
decrease the learning rate. The line search procedure ``warms up'' the
optimization process, reducing the need for expensive trial and error in
setting an initial learning rate. The method for decreasing the learning rate
is based on a new statistical test for detecting stationarity when using a
constant step size. Unlike in prior work, our test applies to a broad class of
stochastic gradient algorithms without modification. The combined method is
highly robust and autonomous, and it matches the performance of the best
hand-tuned learning rate schedules in our experiments on several deep learning
tasks.
| stat |
VEST: Automatic Feature Engineering for Forecasting | Time series forecasting is a challenging task with applications in a wide
range of domains. Auto-regression is one of the most common approaches to
address these problems. Accordingly, observations are modelled by multiple
regression using their past lags as predictor variables. We investigate the
extension of auto-regressive processes using statistics which summarise the
recent past dynamics of time series. The result of our research is a novel
framework called VEST, designed to perform feature engineering using univariate
and numeric time series automatically. The proposed approach works in three
main steps. First, recent observations are mapped onto different
representations. Second, each representation is summarised by statistical
functions. Finally, a filter is applied for feature selection. We discovered
that combining the features generated by VEST with auto-regression
significantly improves forecasting performance. We provide evidence using 90
time series with high sampling frequency. VEST is publicly available online.
| stat |
High-Dimensional Feature Selection for Sample Efficient Treatment Effect
Estimation | The estimation of causal treatment effects from observational data is a
fundamental problem in causal inference. To avoid bias, the effect estimator
must control for all confounders. Hence practitioners often collect data for as
many covariates as possible to raise the chances of including the relevant
confounders. While this addresses the bias, this has the side effect of
significantly increasing the number of data samples required to accurately
estimate the effect due to the increased dimensionality. In this work, we
consider the setting where out of a large number of covariates $X$ that satisfy
strong ignorability, an unknown sparse subset $S$ is sufficient to include to
achieve zero bias, i.e. $c$-equivalent to $X$. We propose a common objective
function involving outcomes across treatment cohorts with nonconvex joint
sparsity regularization that is guaranteed to recover $S$ with high probability
under a linear outcome model for $Y$ and subgaussian covariates for each of the
treatment cohort. This improves the effect estimation sample complexity so that
it scales with the cardinality of the sparse subset $S$ and $\log |X|$, as
opposed to the cardinality of the full set $X$. We validate our approach with
experiments on treatment effect estimation.
| stat |
Life After Bootstrap: Residual Randomization Inference in Regression
Models | We develop a randomization-based method for inference in regression models.
The basis of inference is an invariance assumption on the regression errors,
such as invariance to permutations or random signs. To test significance, the
randomization method repeatedly calculates a suitable test statistic over
transformations of the regression residuals according to the invariant.
Inversion of the test can produce confidence intervals. We prove general
conditions for asymptotic validity of this residual randomization test and
illustrate in many models, including clustered errors with one-way or two-way
clustering structure. We also show that finite-sample validity is possible
under a suitable construction, and illustrate with an exact test for a case of
the Behrens-Fisher problem. The proposed method offers four main advantages
over the bootstrap: (1) it addresses the inference problem in a unified way,
while bootstrap typically needs to be adapted to the task; (2) it can be more
powerful by exploiting a richer and more flexible set of invariances than
exchangeability; (3) it does not rely on asymptotic normality; and (4) it can
be valid in finite samples. In extensive empirical evaluations, including high
dimensional regression and autocorrelated errors, the proposed method performs
favorably against many alternatives, including bootstrap variants and
asymptotic robust error methods.
| stat |
Machine learning based digital twin for stochastic nonlinear
multi-degree of freedom dynamical system | The potential of digital twin technology is immense, specifically in the
infrastructure, aerospace, and automotive sector. However, practical
implementation of this technology is not at an expected speed, specifically
because of lack of application-specific details. In this paper, we propose a
novel digital twin framework for stochastic nonlinear multi-degree of freedom
(MDOF) dynamical systems. The approach proposed in this paper strategically
decouples the problem into two time-scales -- (a) a fast time-scale governing
the system dynamics and (b) a slow time-scale governing the degradation in the
system. The proposed digital twin has four components - (a) a physics-based
nominal model (low-fidelity), (b) a Bayesian filtering algorithm a (c) a
supervised machine learning algorithm and (d) a high-fidelity model for
predicting future responses. The physics-based nominal model combined with
Bayesian filtering is used combined parameter state estimation and the
supervised machine learning algorithm is used for learning the temporal
evolution of the parameters. While the proposed framework can be used with any
choice of Bayesian filtering and machine learning algorithm, we propose to use
unscented Kalman filter and Gaussian process. Performance of the proposed
approach is illustrated using two examples. Results obtained indicate the
applicability and excellent performance of the proposed digital twin framework.
| stat |
Robust Design and Analysis of Clinical Trials With Non-proportional
Hazards: A Straw Man Guidance from a Cross-pharma Working Group | Loss of power and clear description of treatment differences are key issues
in designing and analyzing a clinical trial where non-proportional hazard is a
possibility. A log-rank test may be very inefficient and interpretation of the
hazard ratio estimated using Cox regression is potentially problematic. In this
case, the current ICH E9 (R1) addendum would suggest designing a trial with a
clinically relevant estimand, e.g., expected life gain. This approach considers
appropriate analysis methods for supporting the chosen estimand. However, such
an approach is case specific and may suffer lack of power for important choices
of the underlying alternate hypothesis distribution. On the other hand, there
may be a desire to have robust power under different deviations from
proportional hazards. Also, we would contend that no single number adequately
describes treatment effect under non-proportional hazards scenarios. The
cross-pharma working group has proposed a combination test to provide robust
power under a variety of alternative hypotheses. These can be specified for
primary analysis at the design stage and methods appropriately accounting for
combination test correlations are efficient for a variety of scenarios. We have
provided design and analysis considerations based on a combination test under
different non-proportional hazard types and present a straw man proposal for
practitioners. The proposals are illustrated with real life example and
simulation.
| stat |
Importance subsampling for power system planning under multi-year demand
and weather uncertainty | This paper introduces a generalised version of importance subsampling for
time series reduction/aggregation in optimisation-based power system planning
models. Recent studies indicate that reliably determining optimal electricity
(investment) strategy under climate variability requires the consideration of
multiple years of demand and weather data. However, solving planning models
over long simulation lengths is typically computationally unfeasible, and
established time series reduction approaches induce significant errors. The
importance subsampling method reliably estimates long-term planning model
outputs at greatly reduced computational cost, allowing the consideration of
multi-decadal samples. The key innovation is a systematic identification and
preservation of relevant extreme events in modeling subsamples. Simulation
studies on generation and transmission expansion planning models illustrate the
method's enhanced performance over established "representative days" clustering
approaches. The models, data and sample code are made available as open-source
software.
| stat |
Increased risk of hospitalisation for COVID-19 patients infected with
SARS-CoV-2 variant B.1.1.7 | OBJECTIVE: To evaluate the association between coronavirus disease 2019
(COVID-19) diagnosis with SARS-CoV-2 variant B.1.1.7 (also known as Variant of
Concern 202012/01) and the risk of hospitalisation compared to diagnosis with
wildtype SARS-CoV-2 variants.
DESIGN: Retrospective cohort, analysed using a matched nested case-control
study design.
SETTING: Community-based SARS-CoV-2 testing in England, individually linked
with hospitalisation data from the Secondary Uses Service and the Emergency
Care Data Set.
PARTICIPANTS: 493,053 laboratory-confirmed COVID-19 patients, of whom 19,482
had been hospitalised, tested between 23 November 2020 and 4 January 2021 and
analysed at a laboratory with an available TaqPath assay that enables
assessment of S-gene target failure (SGTF). SGTF is a proxy test for the
B.1.1.7 variant. Nested case-control analysis of 18,814 pairs of
hospitalised/non-hospitalised patients, one-to-one matched on age, sex,
ethnicity, deprivation, region of residence, and date of positive test (matches
found for 96.6%).
MAIN OUTCOME MEASURES: Hospitalisation between 1 and 14 days after the first
positive SARS-CoV-2 test.
RESULTS: The matching-adjusted conditional odds ratio of hospitalisation was
1.58 (95% confidence interval 1.50 to 1.67) for COVID-19 patients infected with
a SGTF-associated variant, compared to those infected with non-SGTF-associated
variants. Secondary cohort analyses yielded similar estimates. The effect was
modified by age (p<0.001), with ORs of 0.96-1.13 below age 20 years and
1.57-1.67 in age groups 40 years or older.
CONCLUSIONS: The results suggest that the risk of hospitalisation is higher
for individuals infected with the B.1.1.7 variant compared to wildtype
SARS-CoV-2, likely reflecting a more severe disease. The higher severity may be
specific to older age groups.
| stat |
Ranking of average treatment effects with generalized random forests for
time-to-event outcomes | In this paper we present a data-adaptive estimation procedure for estimation
of average treatment effects in a time-to-event setting based on generalized
random forests. In these kinds of settings, the definition of causal effect
parameters are complicated by competing risks; here we distinguish between
treatment effects on the crude and the net probabilities, respectively. To
handle right-censoring, and to switch between crude and net probabilities, we
propose a two-step procedure for estimation, applying inverse probability
weighting to construct time-point specific weighted outcomes as input for the
forest. The forest adaptively handles confounding of the treatment assigned by
applying a splitting rule that targets a causal parameter. We demonstrate that
our method is effective for a causal search through a list of treatments to be
ranked according to the magnitude of their effect. We further apply our method
to a dataset from the Danish health registries where it is of interest to
discover drugs with an unexpected protective effect against relapse of severe
depression.
| stat |
Beyond Greedy Ranking: Slate Optimization via List-CVAE | The conventional solution to the recommendation problem greedily ranks
individual document candidates by prediction scores. However, this method fails
to optimize the slate as a whole, and hence, often struggles to capture biases
caused by the page layout and document interdepedencies. The slate
recommendation problem aims to directly find the optimally ordered subset of
documents (i.e. slates) that best serve users' interests. Solving this problem
is hard due to the combinatorial explosion in all combinations of document
candidates and their display positions on the page. Therefore we propose a
paradigm shift from the traditional viewpoint of solving a ranking problem to a
direct slate generation framework. In this paper, we introduce List Conditional
Variational Auto-Encoders (List-CVAE), which learns the joint distribution of
documents on the slate conditioned on user responses, and directly generates
full slates. Experiments on simulated and real-world data show that List-CVAE
outperforms popular comparable ranking methods consistently on various scales
of documents corpora.
| stat |
Probabilistic Count Matrix Factorization for Single Cell Expression Data
Analysis | The development of high throughput single-cell sequencing technologies now
allows the investigation of the population level diversity of cellular
transcriptomes. This diversity has shown two faces. First, the expression
dynamics (gene to gene variability) can be quantified more accurately, thanks
to the measurement of lowly-expressed genes. Second, the cell-to-cell
variability is high, with a low proportion of cells expressing the same gene at
the same time/level. Those emerging patterns appear to be very challenging from
the statistical point of view, especially to represent and to provide a
summarized view of single-cell expression data. PCA is one of the most powerful
framework to provide a suitable representation of high dimensional datasets, by
searching for latent directions catching the most variability in the data.
Unfortunately, classical PCA is based on Euclidean distances and projections
that work poorly in presence of over-dispersed counts that show drop-out events
(zero-inflation) like single-cell expression data We propose a probabilistic
Count Matrix Factorization (pCMF) approach for single-cell expression data
analysis, that relies on a sparse Gamma-Poisson factor model. This hierarchical
model is inferred using a variational EM algorithm. It is able to jointly build
a low dimensional representation of cells and genes. We show how this
probabilistic framework induces a geometry that is suitable for single-cell
data visualization, and produces a compression of the data that is very
powerful for clustering purposes. Our method is competed against other standard
representation methods like t-SNE, and we illustrate its performance for the
representation of single-cell data. We especially focus on publicly available
single-cell RNA-seq datasets.
| stat |
Representative endowments and uniform Gini orderings of multi-attribute
inequality | For the comparison of inequality in multiple attributes the use of
generalized Gini indices is proposed. Spectral social evaluation functions are
used in the multivariate setting, and Gini dominance orderings are introduced
that are uniform in attribute weights. Classes of spectral evaluators are
considered that are parameterized by their aversion to inequality. Then a
set-valued representative endowment is defined that characterizes
$d$-dimensioned inequality. It consists of all points above the lower border of
a convex compact in $R^d$, while the pointwise ordering of such endowments
corresponds to uniform Gini dominance. Properties of uniform Gini dominance are
derived, including relations to other orderings of $d$-variate distributions
such as usual multivariate stochastic order and convex order. The
multi-dimensioned representative endowment can be efficiently calculated from
data; in a sampling context, it consistently estimates its population version.
| stat |
The use of restricted mean time lost under competing risks data | Background: Under competing risks, the commonly used sub-distribution hazard
ratio (SHR) is not easy to interpret clinically and is valid only under the
proportional sub-distribution hazard (SDH) assumption. This paper introduces an
alternative statistical measure: the restricted mean time lost (RMTL). Methods:
First, the definition and estimation methods of the measures are introduced.
Second, based on the differences in RMTLs, a basic difference test (Diff) and a
supremum difference test (sDiff) are constructed. Then, the corresponding
sample size estimation method is proposed. The statistical properties of the
methods and the estimated sample size are evaluated using Monte Carlo
simulations, and these methods are also applied to two real examples. Results:
The simulation results show that sDiff performs well and has relatively high
test efficiency in most situations. Regarding sample size calculation, sDiff
exhibits good performance in various situations. The methods are illustrated
using two examples. Conclusions: RMTL can meaningfully summarize treatment
effects for clinical decision making, which can then be reported with the SDH
ratio for competing risks data. The proposed sDiff test and the two calculated
sample size formulas have wide applicability and can be considered in real data
analysis and trial design.
| stat |
Nonparametric Trend Estimation in Functional Time Series with
Application to Annual Mortality Rates | Here, we address the problem of trend estimation for functional time series.
Existing contributions either deal with detecting a functional trend or
assuming a simple model. They consider neither the estimation of a general
functional trend nor the analysis of functional time series with a functional
trend component. Similarly to univariate time series, we propose an alternative
methodology to analyze functional time series, taking into account a functional
trend component. We propose to estimate the functional trend by using a tensor
product surface that is easy to implement, to interpret, and allows to control
the smoothness properties of the estimator. Through a Monte Carlo study, we
simulate different scenarios of functional processes to show that our estimator
accurately identifies the functional trend component. We also show that the
dependency structure of the estimated stationary time series component is not
significantly affected by the error approximation of the functional trend
component. We apply our methodology to annual mortality rates in France.
| stat |
Teaching Design of Experiments using Hasse diagrams | Hasse diagrams provide a principled means for visualizing the structure of
statistical designs constructed by crossing and nesting of experimental
factors. They have long been applied for automated construction of linear
models and their associated linear subspaces for complex designs. Here, we
argue that they could also provide a central component for planning and
teaching introductory or service courses in experimental design.
Specifically, we show how Hasse diagrams allow constructing most elementary
designs and finding many of their properties, such as degrees of freedom, error
strata, experimental units and denominators for F-tests. Linear (mixed) models
for analysis directly correspond to the diagrams, which facilitates both
defining a model and specifying it in statistical software. We demonstrate how
instructors can seamlessly use Hasse diagrams to construct designs by combining
simple unit- and treatment structures, identify pseudo-replication, and discuss
a design's randomization, unit-treatment versus treatment-treatment
interactions, or complete confounding. These features commend Hasse diagrams as
a powerful tool for unifying ideas and concepts.
| stat |
Is there Anisotropy in Structural Bias? | Structural Bias (SB) is an important type of algorithmic deficiency within
iterative optimisation heuristics. However, methods for detecting structural
bias have not yet fully matured, and recent studies have uncovered many
interesting questions. One of these is the question of how structural bias can
be related to anisotropy. Intuitively, an algorithm that is not isotropic would
be considered structurally biased. However, there have been cases where
algorithms appear to only show SB in some dimensions. As such, we investigate
whether these algorithms actually exhibit anisotropy, and how this impacts the
detection of SB. We find that anisotropy is very rare, and even in cases where
it is present, there are clear tests for SB which do not rely on any
assumptions of isotropy, so we can safely expand the suite of SB tests to
encompass these kinds of deficiencies not found by the original tests.
We propose several additional testing procedures for SB detection and aim to
motivate further research into the creation of a robust portfolio of tests.
This is crucial since no single test will be able to work effectively with all
types of SB we identify.
| stat |
Copula-based Sensitivity Analysis for Multi-Treatment Causal Inference
with Unobserved Confounding | Recent work has focused on the potential and pitfalls of causal
identification in observational studies with multiple simultaneous treatments.
On the one hand, a latent variable model fit to the observed treatments can
identify essential aspects of the distribution of unobserved confounders. On
the other hand, it has been shown that even when the latent confounder
distribution is known exactly, causal effects are still not point identifiable.
Thus, the practical benefits of latent variable modeling in multi-treatment
settings remain unclear. We clarify these issues with a sensitivity analysis
method that can be used to characterize the range of causal effects that are
compatible with the observed data. Our method is based on a copula
factorization of the joint distribution of outcomes, treatments, and
confounders, and can be layered on top of arbitrary observed data models. We
propose a practical implementation of this approach making use of the Gaussian
copula, and establish conditions under which causal effects can be bounded. We
also describe approaches for reasoning about effects, including calibrating
sensitivity parameters, quantifying robustness of effect estimates, and
selecting models which are most consistent with prior hypotheses.
| stat |
Exact Langevin Dynamics with Stochastic Gradients | Stochastic gradient Markov Chain Monte Carlo algorithms are popular samplers
for approximate inference, but they are generally biased. We show that many
recent versions of these methods (e.g. Chen et al. (2014)) cannot be corrected
using Metropolis-Hastings rejection sampling, because their acceptance
probability is always zero. We can fix this by employing a sampler with
realizable backwards trajectories, such as Gradient-Guided Monte Carlo
(Horowitz, 1991), which generalizes stochastic gradient Langevin dynamics
(Welling and Teh, 2011) and Hamiltonian Monte Carlo. We show that this sampler
can be used with stochastic gradients, yielding nonzero acceptance
probabilities, which can be computed even across multiple steps.
| stat |
Interaction Pursuit Biconvex Optimization | Multivariate regression models are widely used in various fields such as
biology and finance. In this paper, we focus on two key challenges: (a) When
should we favor a multivariate model over a series of univariate models; (b) If
the numbers of responses and predictors are allowed to greatly exceed the
sample size, how to reduce the computational cost and provide precise
estimation. The proposed method, Interaction Pursuit Biconvex Optimization
(IPBO), explores the regression relationship allowing the predictors and
responses derived from different multivariate normal distributions with general
covariance matrices. In practice, the correlation structures within are complex
and interact on each other based on the regression function. The proposed
method solves this problem by building a structured sparsity penalty to
encourages the shared structure between the network and the regression
coefficients. We prove theoretical results under interpretable conditions, and
provide an efficient algorithm to compute the estimator. Simulation studies and
real data examples compare the proposed method with several existing methods,
indicating that IPBO works well.
| stat |
Multi-Objective Automatic Machine Learning with AutoxgboostMC | AutoML systems are currently rising in popularity, as they can build powerful
models without human oversight. They often combine techniques from many
different sub-fields of machine learning in order to find a model or set of
models that optimize a user-supplied criterion, such as predictive performance.
The ultimate goal of such systems is to reduce the amount of time spent on
menial tasks, or tasks that can be solved better by algorithms while leaving
decisions that require human intelligence to the end-user. In recent years, the
importance of other criteria, such as fairness and interpretability, and many
others have become more and more apparent. Current AutoML frameworks either do
not allow to optimize such secondary criteria or only do so by limiting the
system's choice of models and preprocessing steps. We propose to optimize
additional criteria defined by the user directly to guide the search towards an
optimal machine learning pipeline. In order to demonstrate the need and
usefulness of our approach, we provide a simple multi-criteria AutoML system
and showcase an exemplary application.
| stat |
Understanding complex predictive models with Ghost Variables | We propose a procedure for assigning a relevance measure to each explanatory
variable in a complex predictive model. We assume that we have a training set
to fit the model and a test set to check the out of sample performance. First,
the individual relevance of each variable is computed by comparing the
predictions in the test set, given by the model that includes all the variables
with those of another model in which the variable of interest is substituted by
its ghost variable, defined as the prediction of this variable by using the
rest of explanatory variables. Second, we check the joint effects among the
variables by using the eigenvalues of a relevance matrix that is the covariance
matrix of the vectors of individual effects. It is shown that in simple models,
as linear or additive models, the proposed measures are related to standard
measures of significance of the variables and in neural networks models (and in
other algorithmic prediction models) the procedure provides information about
the joint and individual effects of the variables that is not usually available
by other methods. The procedure is illustrated with simulated examples and the
analysis of a large real data set.
| stat |
Simulation study of estimating between-study variance and overall effect
in meta-analysis of odds-ratios | Random-effects meta-analysis requires an estimate of the between-study
variance, $\tau^2$. We study methods of estimation of $\tau^2$ and its
confidence interval in meta-analysis of odds ratio, and also the performance of
related estimators of the overall effect. We provide results of extensive
simulations on five point estimators of $\tau^2$ (the popular methods of
DerSimonian-Laird, restricted maximum likelihood, and Mandel and Paule; the
less-familiar method of Jackson; and the new method (KD) based on the improved
approximation to the distribution of the Q statistic by Kulinskaya and
Dollinger (2015)); five interval estimators for $\tau^2$ (profile likelihood,
Q-profile, Biggerstaff and Jackson, Jackson, and KD), six point estimators of
the overall effect (the five inverse-variance estimators related to the point
estimators of $\tau^2$ and an estimator (SSW) 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 KD-based modification of HKSJ;
and an interval based on the sample-size-weighted estimator). Results of our
simulations show that none of the point estimators of $\tau^2$ can be
recommended, however the new KD estimator provides a reliable coverage of
$\tau^2$. Inverse-variance estimators of the overall effect are substantially
biased. The SSW estimator of the overall effect and the related confidence
interval provide the reliable point and interval estimation of log-odds-ratio.
| stat |
Sparse Polynomial Chaos expansions using Variational Relevance Vector
Machines | The challenges for non-intrusive methods for Polynomial Chaos modeling lie in
the computational efficiency and accuracy under a limited number of model
simulations. These challenges can be addressed by enforcing sparsity in the
series representation through retaining only the most important basis terms. In
this work, we present a novel sparse Bayesian learning technique for obtaining
sparse Polynomial Chaos expansions which is based on a Relevance Vector Machine
model and is trained using Variational Inference. The methodology shows great
potential in high-dimensional data-driven settings using relatively few data
points and achieves user-controlled sparse levels that are comparable to other
methods such as compressive sensing. The proposed approach is illustrated on
two numerical examples, a synthetic response function that is explored for
validation purposes and a low-carbon steel plate with random Young's modulus
and random loading, which is modeled by stochastic finite element with 38 input
random variables.
| stat |
Another look at the Lady Tasting Tea and differences between permutation
tests and randomization tests | The statistical literature is known to be inconsistent in the use of the
terms "permutation test" and "randomization test". Several authors succesfully
argue that these terms should be used to refer to two distinct classes of tests
and that there are major conceptual differences between these classes. The
present paper explains an important difference in mathematical reasoning
between these classes: a permutation test fundamentally requires that the set
of permutations has a group structure, in the algebraic sense; the reasoning
behind a randomization test is not based on such a group structure and it is
possible to use an experimental design that does not correspond to a group. In
particular, we can use a randomization scheme where the number of possible
treatment patterns is larger than in standard experimental designs. This leads
to exact \emph{p}-values of improved resolution, providing increased power for
very small significance levels, at the cost of decreased power for larger
significance levels. We discuss applications in randomized trials and
elsewhere. Further, we explain that Fisher's famous Lady Tasting Tea
experiment, which is commonly referred to as the first permutation test, is in
fact a randomization test. This distinction is important to avoid confusion and
invalid tests.
| stat |
Bayesian Median Autoregression for Robust Time Series Forecasting | We develop a Bayesian median autoregressive (BayesMAR) model for time series
forecasting. The proposed method utilizes time-varying quantile regression at
the median, favorably inheriting the robustness of median regression in
contrast to the widely used mean-based methods. Motivated by a working Laplace
likelihood approach in Bayesian quantile regression, BayesMAR adopts a
parametric model bearing the same structure as autoregressive models by
altering the Gaussian error to Laplace, leading to a simple, robust, and
interpretable modeling strategy for time series forecasting. We estimate model
parameters by Markov chain Monte Carlo. Bayesian model averaging is used to
account for model uncertainty, including the uncertainty in the autoregressive
order, in addition to a Bayesian model selection approach. The proposed methods
are illustrated using simulations and real data applications. An application to
U.S. macroeconomic data forecasting shows that BayesMAR leads to favorable and
often superior predictive performance compared to the selected mean-based
alternatives under various loss functions that encompass both point and
probabilistic forecasts. The proposed methods are generic and can be used to
complement a rich class of methods that build on autoregressive models.
| stat |
Developing a large scale population screening tool for the assessment of
Parkinson's disease using telephone-quality voice | Recent studies have demonstrated that analysis of laboratory-quality voice
recordings can be used to accurately differentiate people diagnosed with
Parkinson's disease (PD) from healthy controls (HC). These findings could help
facilitate the development of remote screening and monitoring tools for PD. In
this study, we analyzed 2759 telephone-quality voice recordings from 1483 PD
and 15321 recordings from 8300 HC participants. To account for variations in
phonetic backgrounds, we acquired data from seven countries. We developed a
statistical framework for analyzing voice, whereby we computed 307 dysphonia
measures that quantify different properties of voice impairment, such as,
breathiness, roughness, monopitch, hoarse voice quality, and exaggerated vocal
tremor. We used feature selection algorithms to identify robust parsimonious
feature subsets, which were used in combination with a Random Forests (RF)
classifier to accurately distinguish PD from HC. The best 10-fold
cross-validation performance was obtained using Gram-Schmidt Orthogonalization
(GSO) and RF, leading to mean sensitivity of 64.90% (standard deviation, SD
2.90%) and mean specificity of 67.96% (SD 2.90%). This large-scale study is a
step forward towards assessing the development of a reliable, cost-effective
and practical clinical decision support tool for screening the population at
large for PD using telephone-quality voice.
| stat |
Decision Support for COVID-19 Policy Including Poverty Drivers | In this paper we present a multi-attribute decision support framework for
choosing between countermeasure strategies designed to mitigate the effects of
COVID-19 in the UK. Such an analysis can evaluate both the short term and long
term efficacy of various candidate countermeasures.The expected utility scores
of a countermeasure strategy captures the expected impact of the policies on
health outcomes and other measures of population well-being. The broad
methodologies we use here have been established for some time. However, this
application has many unusual elements to it: the pervasive uncertainty of the
science; the necessary dynamic shifts between states within each candidate
suite of counter measures; the fast moving stochastic development of the
underlying threat all present new challenges to this domain. We incorporate
these within our framework. Further, we demonstrate our framework through an
example where we evaluate several strategies by considering short- and
long-term attributes that impact on the health of our population.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.