title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Is VARS more intuitive and efficient than Sobol' indices? | The Variogram Analysis of Response Surfaces (VARS) has been proposed by
Razavi and Gupta as a new comprehensive framework in sensitivity analysis.
According to these authors, VARS provides a more intuitive notion of
sensitivity and it is much more computationally efficient than Sobol' indices.
Here we review these arguments and critically compare the performance of
VARS-TO, for total-order index, against the total-order Jansen estimator. We
argue that, unlike classic variance-based methods, VARS lacks a clear
definition of what an "important" factor is, and show that the alleged
computational superiority of VARS does not withstand scrutiny. We conclude that
while VARS enriches the spectrum of existing methods for sensitivity analysis,
especially for a diagnostic use of mathematical models, it complements rather
than substitutes classic estimators used in variance-based sensitivity
analysis.
| stat |
Parallel Tempering via Simulated Tempering Without Normalizing Constants | In this paper we develop a new general Bayesian methodology that
simultaneously estimates parameters of interest and the marginal likelihood of
the model. The proposed methodology builds on Simulated Tempering, which is a
powerful algorithm that enables sampling from multi-modal distributions.
However, Simulated Tempering comes with the practical limitation of needing to
specify a prior for the temperature along a chosen discretization schedule that
will allow calculation of normalizing constants at each temperature. Our
proposed model defines the prior for the temperature so as to remove the need
for calculating normalizing constants at each temperature and thereby enables a
continuous temperature schedule, while preserving the sampling efficiency of
the Simulated Tempering algorithm. The resulting algorithm simultaneously
estimates parameters while estimating marginal likelihoods through
thermodynamic integration. We illustrate the applicability of the new algorithm
to different examples involving mixture models of Gaussian distributions and
ordinary differential equation models.
| stat |
A prior-based approximate latent Riemannian metric | Stochastic generative models enable us to capture the geometric structure of
a data manifold lying in a high dimensional space through a Riemannian metric
in the latent space. However, its practical use is rather limited mainly due to
inevitable complexity. In this work we propose a surrogate conformal Riemannian
metric in the latent space of a generative model that is simple, efficient and
robust. This metric is based on a learnable prior that we propose to learn
using a basic energy-based model. We theoretically analyze the behavior of the
proposed metric and show that it is sensible to use in practice. We demonstrate
experimentally the efficiency and robustness, as well as the behavior of the
new approximate metric. Also, we show the applicability of the proposed
methodology for data analysis in the life sciences.
| stat |
Multidataset Independent Subspace Analysis with Application to
Multimodal Fusion | In the last two decades, unsupervised latent variable models---blind source
separation (BSS) especially---have enjoyed a strong reputation for the
interpretable features they produce. Seldom do these models combine the rich
diversity of information available in multiple datasets. Multidatasets, on the
other hand, yield joint solutions otherwise unavailable in isolation, with a
potential for pivotal insights into complex systems.
To take advantage of the complex multidimensional subspace structures that
capture underlying modes of shared and unique variability across and within
datasets, we present a direct, principled approach to multidataset combination.
We design a new method called multidataset independent subspace analysis (MISA)
that leverages joint information from multiple heterogeneous datasets in a
flexible and synergistic fashion.
Methodological innovations exploiting the Kotz distribution for subspace
modeling in conjunction with a novel combinatorial optimization for evasion of
local minima enable MISA to produce a robust generalization of independent
component analysis (ICA), independent vector analysis (IVA), and independent
subspace analysis (ISA) in a single unified model.
We highlight the utility of MISA for multimodal information fusion, including
sample-poor regimes and low signal-to-noise ratio scenarios, promoting novel
applications in both unimodal and multimodal brain imaging data.
| stat |
Latent likelihood ratio tests for assessing spatial kernels in epidemic
models | One of the most important issues in the critical assessment of
spatio-temporal stochastic models for epidemics is the selection of the
transmission kernel used to represent the relationship between infectious
challenge and spatial separation of infected and susceptible hosts. As the
design of control strategies is often based on an assessment of the distance
over which transmission can realistically occur and estimation of this distance
is very sensitive to the choice of kernel function, it is important that models
used to inform control strategies can be scrutinised in the light of
observation in order to elicit possible evidence against the selected kernel
function. While a range of approaches to model criticism are in existence, the
field remains one in which the need for further research is recognised. In this
paper, building on earlier contributions by the authors, we introduce a new
approach to assessing the validity of spatial kernels - the latent likelihood
ratio tests - and compare its capacity to detect model misspecification with
that of tests based on the use of infection-link residuals. We demonstrate that
the new approach, which combines Bayesian and frequentist ideas by treating the
statistical decision maker as a complex entity, can be used to formulate tests
with greater power than infection-link residuals to detect kernel
misspecification particularly when the degree of misspecification is modest.
This new approach avoids the use of a fully Bayesian approach which may
introduce undesirable complications related to computational complexity and
prior sensitivity.
| stat |
Statistical inference for ARTFIMA time series with stable innovations | Autoregressive tempered fractionally integrated moving average with stable
innovations modifies the power-law kernel of the fractionally integrated time
series model by adding an exponential tempering factor. The tempered time
series is a stationary model that can exhibits semi-long-range dependence. This
paper develops the basic theory of the tempered time series model, including
dependence structure and parameter estimation.
| stat |
Quantitative system risk assessment from incomplete data with belief
networks and pairwise comparison elicitation | A method for conducting Bayesian elicitation and learning in risk assessment
is presented. It assumes that the risk process can be described as a fault
tree. This is viewed as a belief network, for which prior distributions on
primary event probabilities are elicited by means of a pairwise comparison
approach. A Bayesian updating procedure, following observation of some or all
of the events in the fault tree, is described. The application is illustrated
through the motivating example of risk assessment of spacecraft explosion
during controlled re-entry.
| stat |
Sparse Gaussian Processes Revisited: Bayesian Approaches to
Inducing-Variable Approximations | Variational inference techniques based on inducing variables provide an
elegant framework for scalable posterior estimation in Gaussian process (GP)
models. Besides enabling scalability, one of their main advantages over sparse
approximations using direct marginal likelihood maximization is that they
provide a robust alternative for point estimation of the inducing inputs, i.e.
the location of the inducing variables. In this work we challenge the common
wisdom that optimizing the inducing inputs in the variational framework yields
optimal performance. We show that, by revisiting old model approximations such
as the fully-independent training conditionals endowed with powerful
sampling-based inference methods, treating both inducing locations and GP
hyper-parameters in a Bayesian way can improve performance significantly. Based
on stochastic gradient Hamiltonian Monte Carlo, we develop a fully Bayesian
approach to scalable GP and deep GP models, and demonstrate its
state-of-the-art performance through an extensive experimental campaign across
several regression and classification problems.
| stat |
Decision Tree Learning for Uncertain Clinical Measurements | Clinical decision requires reasoning in the presence of imperfect data. DTs
are a well-known decision support tool, owing to their interpretability,
fundamental in safety-critical contexts such as medical diagnosis. However,
learning DTs from uncertain data leads to poor generalization, and generating
predictions for uncertain data hinders prediction accuracy. Several methods
have suggested the potential of probabilistic decisions at the internal nodes
in making DTs robust to uncertainty. Some approaches only employ probabilistic
thresholds during evaluation. Others also consider the uncertainty in the
learning phase, at the expense of increased computational complexity or reduced
interpretability. The existing methods have not clarified the merit of a
probabilistic approach in the distinct phases of DT learning, nor when the
uncertainty is present in the training or the test data. We present a
probabilistic DT approach that models measurement uncertainty as a noise
distribution, independently realized: (1) when searching for the split
thresholds, (2) when splitting the training instances, and (3) when generating
predictions for unseen data. The soft training approaches (1, 2) achieved a
regularizing effect, leading to significant reductions in DT size, while
maintaining accuracy, for increased noise. Soft evaluation (3) showed no
benefit in handling noise.
| stat |
Improved Fixed-Rank Nystr\"om Approximation via QR Decomposition:
Practical and Theoretical Aspects | The Nystrom method is a popular technique that uses a small number of
landmark points to compute a fixed-rank approximation of large kernel matrices
that arise in machine learning problems. In practice, to ensure high quality
approximations, the number of landmark points is chosen to be greater than the
target rank. However, for simplicity the standard Nystrom method uses a
sub-optimal procedure for rank reduction. In this paper, we examine the
drawbacks of the standard Nystrom method in terms of poor performance and lack
of theoretical guarantees. To address these issues, we present an efficient
modification for generating improved fixed-rank Nystrom approximations.
Theoretical analysis and numerical experiments are provided to demonstrate the
advantages of the modified method over the standard Nystrom method. Overall,
the aim of this paper is to convince researchers to use the modified method, as
it has nearly identical computational complexity, is easy to code, has greatly
improved accuracy in many cases, and is optimal in a sense that we make
precise.
| stat |
Hierarchical Hidden Markov Jump Processes for Cancer Screening Modeling | Hidden Markov jump processes are an attractive approach for modeling clinical
disease progression data because they are explainable and capable of handling
both irregularly sampled and noisy data. Most applications in this context
consider time-homogeneous models due to their relative computational
simplicity. However, the time homogeneous assumption is too strong to
accurately model the natural history of many diseases. Moreover, the population
at risk is not homogeneous either, since disease exposure and susceptibility
can vary considerably. In this paper, we propose a piece-wise stationary
transition matrix to explain the heterogeneity in time. We propose a
hierarchical structure for the heterogeneity in population, where prior
information is considered to deal with unbalanced data. Moreover, an efficient,
scalable EM algorithm is proposed for inference. We demonstrate the feasibility
and superiority of our model on a cervical cancer screening dataset from the
Cancer Registry of Norway. Experiments show that our model outperforms
state-of-the-art recurrent neural network models in terms of prediction
accuracy and significantly outperforms a standard hidden Markov jump process in
generating Kaplan-Meier estimators.
| stat |
An efficient and robust approach to Mendelian randomization with
measured pleiotropic effects in a high-dimensional setting | Valid estimation of a causal effect using instrumental variables requires
that all of the instruments are independent of the outcome conditional on the
risk factor of interest and any confounders. In Mendelian randomization studies
with large numbers of genetic variants used as instruments, it is unlikely that
this condition will be met. Any given genetic variant could be associated with
a large number of traits, all of which represent potential pathways to the
outcome which bypass the risk factor of interest. Such pleiotropy can be
accounted for using standard multivariable Mendelian randomization with all
possible pleiotropic traits included as covariates. However, the estimator
obtained in this way will be inefficient if some of the covariates do not truly
sit on pleiotropic pathways to the outcome. We present a method which uses
regularization to identify which out of a set of potential covariates need to
be accounted for in a Mendelian randomization analysis in order to produce an
efficient and robust estimator of a causal effect. The method can be used in
the case where individual-level data are not available and the analysis must
rely on summary-level data only. It can also be used in the case where there
are more covariates under consideration than instruments, which is not possible
using standard multivariable Mendelian randomization. We show the results of
simulation studies which demonstrate the performance of the proposed
regularization method in realistic settings. We also illustrate the method in
an applied example which looks at the causal effect of urate plasma
concentration on coronary heart disease.
| stat |
Bootstrap Confidence Intervals Using the Likelihood Ratio Test in
Changepoint Detection | This study aims to evaluate the performance of power in the likelihood ratio
test for changepoint detection by bootstrap sampling, and proposes a hypothesis
test based on bootstrapped confidence interval lengths. Assuming i.i.d normally
distributed errors, and using the bootstrap method, the changepoint sampling
distribution is estimated. Furthermore, this study describes a method to
estimate a data set with no changepoint to form the null sampling distribution.
With the null sampling distribution, and the distribution of the estimated
changepoint, critical values and power calculations can be made, over the
lengths of confidence intervals.
| stat |
Monotonic Gaussian Process for Spatio-Temporal Disease Progression
Modeling in Brain Imaging Data | We introduce a probabilistic generative model for disentangling
spatio-temporal disease trajectories from series of high-dimensional brain
images. The model is based on spatio-temporal matrix factorization, where
inference on the sources is constrained by anatomically plausible statistical
priors. To model realistic trajectories, the temporal sources are defined as
monotonic and time-reparametrized Gaussian Processes. To account for the
non-stationarity of brain images, we model the spatial sources as sparse codes
convolved at multiple scales. The method was tested on synthetic data
favourably comparing with standard blind source separation approaches. The
application on large-scale imaging data from a clinical study allows to
disentangle differential temporal progression patterns mapping brain regions
key to neurodegeneration, while revealing a disease-specific time scale
associated to the clinical diagnosis.
| stat |
Stochastic Security: Adversarial Defense Using Long-Run Dynamics of
Energy-Based Models | The vulnerability of deep networks to adversarial attacks is a central
problem for deep learning from the perspective of both cognition and security.
The current most successful defense method is to train a classifier using
adversarial images created during learning. Another defense approach involves
transformation or purification of the original input to remove adversarial
signals before the image is classified. We focus on defending naturally-trained
classifiers using Markov Chain Monte Carlo (MCMC) sampling with an Energy-Based
Model (EBM) for adversarial purification. In contrast to adversarial training,
our approach is intended to secure pre-existing and highly vulnerable
classifiers.
The memoryless behavior of long-run MCMC sampling will eventually remove
adversarial signals, while metastable behavior preserves consistent appearance
of MCMC samples after many steps to allow accurate long-run prediction.
Balancing these factors can lead to effective purification and robust
classification. We evaluate adversarial defense with an EBM using the strongest
known attacks against purification. Our contributions are 1) an improved method
for training EBM's with realistic long-run MCMC samples, 2) an
Expectation-Over-Transformation (EOT) defense that resolves theoretical
ambiguities for stochastic defenses and from which the EOT attack naturally
follows, and 3) state-of-the-art adversarial defense for naturally-trained
classifiers and competitive defense compared to adversarially-trained
classifiers on Cifar-10, SVHN, and Cifar-100. Code and pre-trained models are
available at https://github.com/point0bar1/ebm-defense.
| stat |
Hilbert Space Methods for Reduced-Rank Gaussian Process Regression | This paper proposes a novel scheme for reduced-rank Gaussian process
regression. The method is based on an approximate series expansion of the
covariance function in terms of an eigenfunction expansion of the Laplace
operator in a compact subset of $\mathbb{R}^d$. On this approximate eigenbasis
the eigenvalues of the covariance function can be expressed as simple functions
of the spectral density of the Gaussian process, which allows the GP inference
to be solved under a computational cost scaling as $\mathcal{O}(nm^2)$
(initial) and $\mathcal{O}(m^3)$ (hyperparameter learning) with $m$ basis
functions and $n$ data points. Furthermore, the basis functions are independent
of the parameters of the covariance function, which allows for very fast
hyperparameter learning. The approach also allows for rigorous error analysis
with Hilbert space theory, and we show that the approximation becomes exact
when the size of the compact subset and the number of eigenfunctions go to
infinity. We also show that the convergence rate of the truncation error is
independent of the input dimensionality provided that the differentiability
order of the covariance function is increases appropriately, and for the
squared exponential covariance function it is always bounded by ${\sim}1/m$
regardless of the input dimensionality. The expansion generalizes to Hilbert
spaces with an inner product which is defined as an integral over a specified
input density. The method is compared to previously proposed methods
theoretically and through empirical tests with simulated and real data.
| stat |
A robust multivariate linear non-parametric maximum likelihood model for
ties | Statistical analysis in applied research, across almost every field (e.g.,
biomedical, economics, computer science, and psychological) makes use of
samples upon which the explicit error distribution of the dependent variable is
unknown or, at best, difficult to linearly model. Yet, these assumptions are
extremely common. Unknown distributions are of course biased when incorrectly
specified, compromising the generalisability of our interpretations -- the
linearly unbiased Euclidean distance is very difficult to correctly identify
upon finite samples and therefore results in an estimator which is neither
unbiased nor maximally informative when incorrectly applied. The alternative
common solution to the problem however, the use of non-parametric statistics,
has its own fundamental flaws. In particular, these flaws revolve around the
problem of order-statistics and the estimation in the presence of ties, which
often removes the introduction of multiple independent variables and the
estimation of interactions. We introduce a competitor to the Euclidean norm,
the Kemeny norm, which we prove to be a valid Banach space, and construct a
multivariate linear expansion of the Kendall-Theil-Sen estimator, which
performs without compromising the parameter space extensibility, and establish
its linear maximum likelihood properties. Empirical demonstrations upon both
simulated and empirical data shall be used to demonstrate these properties,
such that the new estimator is nearly equivalent in power for the glm upon
Gaussian data, but grossly superior for a vast array of analytic scenarios,
including finite ordinal sum-score analysis, thereby aiding in the resolution
of replication in the Applied Sciences.
| stat |
A Hitchhiker's Guide to Statistical Comparisons of Reinforcement
Learning Algorithms | Consistently checking the statistical significance of experimental results is
the first mandatory step towards reproducible science. This paper presents a
hitchhiker's guide to rigorous comparisons of reinforcement learning
algorithms. After introducing the concepts of statistical testing, we review
the relevant statistical tests and compare them empirically in terms of false
positive rate and statistical power as a function of the sample size (number of
seeds) and effect size. We further investigate the robustness of these tests to
violations of the most common hypotheses (normal distributions, same
distributions, equal variances). Beside simulations, we compare empirical
distributions obtained by running Soft-Actor Critic and Twin-Delayed Deep
Deterministic Policy Gradient on Half-Cheetah. We conclude by providing
guidelines and code to perform rigorous comparisons of RL algorithm
performances.
| stat |
Detecting Out-of-Distribution Inputs to Deep Generative Models Using
Typicality | Recent work has shown that deep generative models can assign higher
likelihood to out-of-distribution data sets than to their training data
(Nalisnick et al., 2019; Choi et al., 2019). We posit that this phenomenon is
caused by a mismatch between the model's typical set and its areas of high
probability density. In-distribution inputs should reside in the former but not
necessarily in the latter, as previous work has presumed. To determine whether
or not inputs reside in the typical set, we propose a statistically principled,
easy-to-implement test using the empirical distribution of model likelihoods.
The test is model agnostic and widely applicable, only requiring that the
likelihood can be computed or closely approximated. We report experiments
showing that our procedure can successfully detect the out-of-distribution sets
in several of the challenging cases reported by Nalisnick et al. (2019).
| stat |
Comments on "correspondence analysis makes you blind" | Collins' (2002) statement "correspondence analysis makes you blind" followed
after his seriation like description of a brand attribute count data set
analyzed by Whitlark and Smith (2001), who applied correspondence analysis. In
this essay we comment on Collins' statement within taxicab correspondence
analysis framework by simultaneously decomposing the covariance matrix and its
associated density matrix, thus interpreting two interrelated maps for
contingency tables : TCov map and TCA map.
| stat |
MALA-within-Gibbs samplers for high-dimensional distributions with
sparse conditional structure | Markov chain Monte Carlo (MCMC) samplers are numerical methods for drawing
samples from a given target probability distribution. We discuss one particular
MCMC sampler, the MALA-within-Gibbs sampler, from the theoretical and practical
perspectives. We first show that the acceptance ratio and step size of this
sampler are independent of the overall problem dimension when (i) the target
distribution has sparse conditional structure, and (ii) this structure is
reflected in the partial updating strategy of MALA-within-Gibbs. If, in
addition, the target density is block-wise log-concave, then the sampler's
convergence rate is independent of dimension. From a practical perspective, we
expect that MALA-within-Gibbs is useful for solving high-dimensional Bayesian
inference problems where the posterior exhibits sparse conditional structure at
least approximately. In this context, a partitioning of the state that
correctly reflects the sparse conditional structure must be found, and we
illustrate this process in two numerical examples. We also discuss trade-offs
between the block size used for partial updating and computational requirements
that may increase with the number of blocks.
| stat |
Towards Testing of Deep Learning Systems with Training Set Reduction | Testing the implementation of deep learning systems and their training
routines is crucial to maintain a reliable code base. Modern software
development employs processes, such as Continuous Integration, in which changes
to the software are frequently integrated and tested. However, testing the
training routines requires running them and fully training a deep learning
model can be resource-intensive, when using the full data set. Using only a
subset of the training data can improve test run time, but can also reduce its
effectiveness. We evaluate different ways for training set reduction and their
ability to mimic the characteristics of model training with the original full
data set. Our results underline the usefulness of training set reduction,
especially in resource-constrained environments.
| stat |
Robust Linear Regression: Optimal Rates in Polynomial Time | We obtain robust and computationally efficient estimators for learning
several linear models that achieve statistically optimal convergence rate under
minimal distributional assumptions. Concretely, we assume our data is drawn
from a $k$-hypercontractive distribution and an $\epsilon$-fraction is
adversarially corrupted. We then describe an estimator that converges to the
optimal least-squares minimizer for the true distribution at a rate
proportional to $\epsilon^{2-2/k}$, when the noise is independent of the
covariates. We note that no such estimator was known prior to our work, even
with access to unbounded computation. The rate we achieve is
information-theoretically optimal and thus we resolve the main open question in
Klivans, Kothari and Meka [COLT'18].
Our key insight is to identify an analytic condition that serves as a
polynomial relaxation of independence of random variables. In particular, we
show that when the moments of the noise and covariates are
negatively-correlated, we obtain the same rate as independent noise. Further,
when the condition is not satisfied, we obtain a rate proportional to
$\epsilon^{2-4/k}$, and again match the information-theoretic lower bound. Our
central technical contribution is to algorithmically exploit independence of
random variables in the "sum-of-squares" framework by formulating it as the
aforementioned polynomial inequality.
| stat |
Distributional Sliced-Wasserstein and Applications to Generative
Modeling | Sliced-Wasserstein distance (SW) and its variant, Max Sliced-Wasserstein
distance (Max-SW), have been used widely in the recent years due to their fast
computation and scalability even when the probability measures lie in a very
high dimensional space. However, SW requires many unnecessary projection
samples to approximate its value while Max-SW only uses the most important
projection, which ignores the information of other useful directions. In order
to account for these weaknesses, we propose a novel distance, named
Distributional Sliced-Wasserstein distance (DSW), that finds an optimal
distribution over projections that can balance between exploring distinctive
projecting directions and the informativeness of projections themselves. We
show that the DSW is a generalization of Max-SW, and it can be computed
efficiently by searching for the optimal push-forward measure over a set of
probability measures over the unit sphere satisfying certain regularizing
constraints that favor distinct directions. Finally, we conduct extensive
experiments with large-scale datasets to demonstrate the favorable performances
of the proposed distances over the previous sliced-based distances in
generative modeling applications.
| stat |
Efficient Quantile Tracking Using an Oracle | For incremental quantile estimators the step size and possibly other tuning
parameters must be carefully set. However, little attention has been given on
how to set these values in an online manner. In this article we suggest two
novel procedures that address this issue.
The core part of the procedures is to estimate the current tracking mean
squared error (MSE). The MSE is decomposed in tracking variance and bias and
novel and efficient procedures to estimate these quantities are presented. It
is shown that estimation bias can be tracked by associating it with the portion
of observations below the quantile estimates.
The first procedure runs an ensemble of $L$ quantile estimators for wide
range of values of the tuning parameters and typically around $L = 100$. In
each iteration an oracle selects the best estimate by the guidance of the
estimated MSEs. The second method only runs an ensemble of $L = 3$ estimators
and thus the values of the tuning parameters need from time to time to be
adjusted for the running estimators. The procedures have a low memory foot
print of $8L$ and a computational complexity of $8L$ per iteration.
The experiments show that the procedures are highly efficient and track
quantiles with an error close to the theoretical optimum. The Oracle approach
performs best, but comes with higher computational cost. The procedures were
further applied to a massive real-life data stream of tweets and proofed real
world applicability of them.
| stat |
Rapid Acceleration of the Permutation Test via Slow Random Walks in the
Permutation Group | The permutation test is an often used test procedure in brain imaging.
Unfortunately, generating every possible permutation for large-scale brain
image datasets such as HCP and ADNI with hundreds images is not practical. Many
previous attempts at speeding up the permutation test rely on various
approximation strategies such as estimating the tail distribution with known
parametric distributions. In this study, we show how to rapidly accelerate the
permutation test without any type of approximate strategies by exploiting the
underlying algebraic structure of the permutation group. The method is applied
to large number of MRIs in two applications: (1) localizing the male and female
differences and (2) localizing the regions of high genetic heritability in the
sulcal and gyral pattern of the human cortical brain.
| stat |
Using Neural Architecture Search for Improving Software Flaw Detection
in Multimodal Deep Learning Models | Software flaw detection using multimodal deep learning models has been
demonstrated as a very competitive approach on benchmark problems. In this
work, we demonstrate that even better performance can be achieved using neural
architecture search (NAS) combined with multimodal learning models. We adapt a
NAS framework aimed at investigating image classification to the problem of
software flaw detection and demonstrate improved results on the Juliet Test
Suite, a popular benchmarking data set for measuring performance of machine
learning models in this problem domain.
| stat |
The False Positive Control Lasso | In high dimensional settings where a small number of regressors are expected
to be important, the Lasso estimator can be used to obtain a sparse solution
vector with the expectation that most of the non-zero coefficients are
associated with true signals. While several approaches have been developed to
control the inclusion of false predictors with the Lasso, these approaches are
limited by relying on asymptotic theory, having to empirically estimate terms
based on theoretical quantities, assuming a continuous response class with
Gaussian noise and design matrices, or high computation costs. In this paper we
show how: (1) an existing model (the SQRT-Lasso) can be recast as a method of
controlling the number of expected false positives, (2) how a similar estimator
can used for all other generalized linear model classes, and (3) this approach
can be fit with existing fast Lasso optimization solvers. Our justification for
false positive control using randomly weighted self-normalized sum theory is to
our knowledge novel. Moreover, our estimator's properties hold in finite
samples up to some approximation error which we find in practical settings to
be negligible under a strict mutual incoherence condition.
| stat |
Synthetic Control Group Methods in the Presence of Interference: The
Direct and Spillover Effects of Light Rail on Neighborhood Retail Activity | In recent years, Synthetic Control Group (SCG) methods have received great
attention from scholars and have been subject to extensions and comparisons
with alternative approaches for program evaluation. However, the existing
methodological literature mainly relies on the assumption of non-interference.
We investigate the use of the SCG method in panel comparative case studies
where interference between the treated and the untreated units is plausible. We
frame our discussion in the potential outcomes approach. Under a partial
interference assumption, we formally define relevant direct and spillover
effects. We also consider the "unrealized" spillover effect on the treated unit
in the hypothetical scenario that another unit in the treated unit's
neighborhood had been assigned to the intervention. Then we investigate the
assumptions under which we can identify and estimate the causal effects of
interest, and show how they can be estimated using the SCG method. We apply our
approach to the analysis of an observational study, where the focus is on
assessing direct and spillover causal effects of a new light rail line recently
built in Florence (Italy) on the retail density of the street where it was
built and of the streets in the treated street's neighborhood.
| stat |
Quantifying the generalization error in deep learning in terms of data
distribution and neural network smoothness | The accuracy of deep learning, i.e., deep neural networks, can be
characterized by dividing the total error into three main types: approximation
error, optimization error, and generalization error. Whereas there are some
satisfactory answers to the problems of approximation and optimization, much
less is known about the theory of generalization. Most existing theoretical
works for generalization fail to explain the performance of neural networks in
practice. To derive a meaningful bound, we study the generalization error of
neural networks for classification problems in terms of data distribution and
neural network smoothness. We introduce the cover complexity (CC) to measure
the difficulty of learning a data set and the inverse of the modulus of
continuity to quantify neural network smoothness. A quantitative bound for
expected accuracy/error is derived by considering both the CC and neural
network smoothness. Although most of the analysis is general and not specific
to neural networks, we validate our theoretical assumptions and results
numerically for neural networks by several data sets of images. The numerical
results confirm that the expected error of trained networks scaled with the
square root of the number of classes has a linear relationship with respect to
the CC. We also observe a clear consistency between test loss and neural
network smoothness during the training process. In addition, we demonstrate
empirically that the neural network smoothness decreases when the network size
increases whereas the smoothness is insensitive to training dataset size.
| stat |
New clustering approach for symbolic polygonal data: application to the
clustering of entrepreneurial regimes | Entrepreneurial regimes are topic, receiving ever more research attention.
Existing studies on entrepreneurial regimes mainly use common methods from
multivariate analysis and some type of institutional related analysis. In our
analysis, the entrepreneurial regimes is analyzed by applying a novel polygonal
symbolic data cluster analysis approach. Considering the diversity of data
structures in Symbolic Data Analysis (SDA), interval-valued data is the most
popular. Yet, this approach requires assuming equidistribution hypothesis. We
use a novel polygonal cluster analysis approach to address this limitation with
additional advantages: to store more information, to significantly reduce large
data sets preserving the classical variability through polygon radius, and to
open new possibilities in symbolic data analysis. We construct a dynamic
cluster analysis algorithm for this type of data with proving main theorems and
lemmata to justify its usage. In the empirical part we use dataset of Global
Entrepreneurship Monitor (GEM) for year 2015 to construct typologies of
countries based on responses to main entrepreneurial questions. The article
presents a novel approach to clustering in statistical theory (with novel type
of variables never accounted for) and application to a pressing issue in
entrepreneurship with novel results.
| stat |
Universal Statistics of Fisher Information in Deep Neural Networks: Mean
Field Approach | The Fisher information matrix (FIM) is a fundamental quantity to represent
the characteristics of a stochastic model, including deep neural networks
(DNNs). The present study reveals novel statistics of FIM that are universal
among a wide class of DNNs. To this end, we use random weights and large width
limits, which enables us to utilize mean field theories. We investigate the
asymptotic statistics of the FIM's eigenvalues and reveal that most of them are
close to zero while the maximum eigenvalue takes a huge value. Because the
landscape of the parameter space is defined by the FIM, it is locally flat in
most dimensions, but strongly distorted in others. Moreover, we demonstrate the
potential usage of the derived statistics in learning strategies. First, small
eigenvalues that induce flatness can be connected to a norm-based capacity
measure of generalization ability. Second, the maximum eigenvalue that induces
the distortion enables us to quantitatively estimate an appropriately sized
learning rate for gradient methods to converge.
| stat |
A penalized likelihood approach for efficiently estimating a partially
linear additive transformation model with current status data | Current status data are commonly encountered in medical and epidemiological
studies in which the failure time for study units is the outcome variable of
interest. Data of this form are characterized by the fact that the failure time
is not directly observed but rather is known relative to an observation time;
i.e., the failure times are either left- or right-censored. Due to its
structure, the analysis of such data can be challenging. To circumvent these
challenges and to provide for a flexible modeling construct which can be used
to analyze current status data, herein, a partially linear additive
transformation model is proposed. In the formulation of this model, constrained
$B$-splines are employed to model the monotone transformation function and
nonlinear covariate effects. To provide for more efficient estimates, a
penalization technique is used to regularize the estimation of all unknown
functions. An easy to implement hybrid algorithm is developed for model fitting
and a simple estimator of the large-sample variance-covariance matrix is
proposed. It is shown theoretically that the proposed estimators of the
finite-dimensional regression coefficients are root-$n$ consistent,
asymptotically normal, and achieve the semi-parametric information bound while
the estimators of the nonparametric components attain the optimal rate of
convergence. The finite-sample performance of the proposed methodology is
evaluated through extensive numerical studies and is further demonstrated
through the analysis of uterine leiomyomata data.
| stat |
High-dimensional principal component analysis with heterogeneous
missingness | We study the problem of high-dimensional Principal Component Analysis (PCA)
with missing observations. In simple, homogeneous missingness settings with a
noise level of constant order, we show that an existing inverse-probability
weighted (IPW) estimator of the leading principal components can (nearly)
attain the minimax optimal rate of convergence. However, deeper investigation
reveals both that, particularly in more realistic settings where the
missingness mechanism is heterogeneous, the empirical performance of the IPW
estimator can be unsatisfactory, and moreover that, in the noiseless case, it
fails to provide exact recovery of the principal components. Our main
contribution, then, is to introduce a new method for high-dimensional PCA,
called `primePCA', that is designed to cope with situations where observations
may be missing in a heterogeneous manner. Starting from the IPW estimator,
primePCA iteratively projects the observed entries of the data matrix onto the
column space of our current estimate to impute the missing entries, and then
updates our estimate by computing the leading right singular space of the
imputed data matrix. It turns out that the interaction between the
heterogeneity of missingness and the low-dimensional structure is crucial in
determining the feasibility of the problem. We therefore introduce an
incoherence condition on the principal components and prove that in the
noiseless case, the error of primePCA converges to zero at a geometric rate
when the signal strength is not too small. An important feature of our
theoretical guarantees is that they depend on average, as opposed to
worst-case, properties of the missingness mechanism. Our numerical studies on
both simulated and real data reveal that primePCA exhibits very encouraging
performance across a wide range of scenarios.
| stat |
Gamification of Pure Exploration for Linear Bandits | We investigate an active pure-exploration setting, that includes best-arm
identification, in the context of linear stochastic bandits. While
asymptotically optimal algorithms exist for standard multi-arm bandits, the
existence of such algorithms for the best-arm identification in linear bandits
has been elusive despite several attempts to address it. First, we provide a
thorough comparison and new insight over different notions of optimality in the
linear case, including G-optimality, transductive optimality from optimal
experimental design and asymptotic optimality. Second, we design the first
asymptotically optimal algorithm for fixed-confidence pure exploration in
linear bandits. As a consequence, our algorithm naturally bypasses the pitfall
caused by a simple but difficult instance, that most prior algorithms had to be
engineered to deal with explicitly. Finally, we avoid the need to fully solve
an optimal design problem by providing an approach that entails an efficient
implementation.
| stat |
Functional Linear Regression: Dependence and Error Contamination | Functional linear regression is an important topic in functional data
analysis. It is commonly assumed that samples of the functional predictor are
independent realizations of an underlying stochastic process, and are observed
over a grid of points contaminated by i.i.d. measurement errors. In practice,
however, the dynamical dependence across different curves may exist and the
parametric assumption on the error covariance structure could be unrealistic.
In this paper, we consider functional linear regression with serially dependent
observations of the functional predictor, when the contamination of the
predictor by the white noise is genuinely functional with fully nonparametric
covariance structure. Inspired by the fact that the autocovariance function of
observed functional predictors automatically filters out the impact from the
unobservable noise term, we propose a novel autocovariance-based generalized
method-of-moments estimate of the slope function. We also develop a
nonparametric smoothing approach to handle the scenario of partially observed
functional predictors. The asymptotic properties of the resulting estimators
under different scenarios are established. Finally, we demonstrate that our
proposed method significantly outperforms possible competing methods through an
extensive set of simulations and an analysis of a public financial dataset.
| stat |
Linearly Constrained Neural Networks | We present a novel approach to modelling and learning vector fields from
physical systems using neural networks that explicitly satisfy known linear
operator constraints. To achieve this, the target function is modelled as a
linear transformation of an underlying potential field, which is in turn
modelled by a neural network. This transformation is chosen such that any
prediction of the target function is guaranteed to satisfy the constraints. The
approach is demonstrated on both simulated and real data examples.
| stat |
Efficient Pure Exploration for Combinatorial Bandits with Semi-Bandit
Feedback | Combinatorial bandits with semi-bandit feedback generalize multi-armed
bandits, where the agent chooses sets of arms and observes a noisy reward for
each arm contained in the chosen set. The action set satisfies a given
structure such as forming a base of a matroid or a path in a graph. We focus on
the pure-exploration problem of identifying the best arm with fixed confidence,
as well as a more general setting, where the structure of the answer set
differs from the one of the action set. Using the recently popularized game
framework, we interpret this problem as a sequential zero-sum game and develop
a CombGame meta-algorithm whose instances are asymptotically optimal algorithms
with finite time guarantees. In addition to comparing two families of learners
to instantiate our meta-algorithm, the main contribution of our work is a
specific oracle efficient instance for best-arm identification with
combinatorial actions. Based on a projection-free online learning algorithm for
convex polytopes, it is the first computationally efficient algorithm which is
asymptotically optimal and has competitive empirical performance.
| stat |
Solving Bayesian Network Structure Learning Problem with Integer Linear
Programming | This dissertation investigates integer linear programming (ILP) formulation
of Bayesian Network structure learning problem. We review the definition and
key properties of Bayesian network and explain score metrics used to measure
how well certain Bayesian network structure fits the dataset. We outline the
integer linear programming formulation based on the decomposability of score
metrics. In order to ensure acyclicity of the structure, we add ``cluster
constraints'' developed specifically for Bayesian network, in addition to cycle
constraints applicable to directed acyclic graphs in general. Since there would
be exponential number of these constraints if we specify them fully, we explain
the methods to add them as cutting planes without declaring them all in the
initial model. Also, we develop a heuristic algorithm that finds a feasible
solution based on the idea of sink node on directed acyclic graphs. We
implemented the ILP formulation and cutting planes as a \textsf{Python}
package, and present the results of experiments with different settings on
reference datasets.
| stat |
Robust approach for variable selection with high dimensional Logitudinal
data analysis | This paper proposes a new robust smooth-threshold estimating equation to
select important variables and automatically estimate parameters for high
dimensional longitudinal data. Our proposed procedure works well when the
number of covariates p increases as the number of subjects n increases and even
when p exceeds n. A novel working correlation matrix is proposed to capture
correlations within the same subject. The proposed estimates are competitive
with the estimates obtained with true correlation structure, especially when
the data are contaminated. Moreover, the proposed method is robust against
outliers in the response variables and/or covariates. Furthermore, the oracle
properties for robust smooth-threshold estimating equations under "large n,
diverging p" are first established under some regularity conditions. Extensive
simulation studies and a yeast cell cycle data are used to evaluate the
performance of the proposed method, and results show that our proposed method
is competitive with existing robust variable selection procedures.
| stat |
Explainable outlier detection through decision tree conditioning | This work describes an outlier detection procedure (named "OutlierTree")
loosely based on the GritBot software developed by RuleQuest research, which
works by evaluating and following supervised decision tree splits on variables,
in whose branches 1-d confidence intervals are constructed for the target
variable and potential outliers flagged according to these confidence
intervals. Under this logic, it's possible to produce human-readable
explanations for why a given value of a variable in an observation can be
considered as outlier, by considering the decision tree branch conditions along
with general distribution statistics among the non-outlier observations that
fell into the same branch, which can then be contrasted against the value which
lies outside the CI. The supervised splits help to ensure that the generated
conditions are not spurious, but rather related to the target variable and
having logical breakpoints.
| stat |
Sparse Quantile Regression | We consider both $\ell_{0}$-penalized and $\ell_{0}$-constrained quantile
regression estimators. For the $\ell_{0}$-penalized estimator, we derive an
exponential inequality on the tail probability of excess quantile prediction
risk and apply it to obtain non-asymptotic upper bounds on the mean-square
parameter and regression function estimation errors. We also derive analogous
results for the $\ell_{0}$-constrained estimator. The resulting rates of
convergence are nearly minimax-optimal and the same as those for
$\ell_{1}$-penalized estimators. Further, we characterize expected Hamming loss
for the $\ell_{0}$-penalized estimator. We implement the proposed procedure via
mixed integer linear programming and also a more scalable first-order
approximation algorithm. We illustrate the finite-sample performance of our
approach in Monte Carlo experiments and its usefulness in a real data
application concerning conformal prediction of infant birth weights (with
$n\approx 10^{3}$ and up to $p>10^{3}$). In sum, our $\ell_{0}$-based method
produces a much sparser estimator than the $\ell_{1}$-penalized approach
without compromising precision.
| stat |
Two-sample Test using Projected Wasserstein Distance: Breaking the Curse
of Dimensionality | We develop a projected Wasserstein distance for the two-sample test, a
fundamental problem in statistics and machine learning: given two sets of
samples, to determine whether they are from the same distribution. In
particular, we aim to circumvent the curse of dimensionality in Wasserstein
distance: when the dimension is high, it has diminishing testing power, which
is inherently due to the slow concentration property of Wasserstein metrics in
the high dimension space. A key contribution is to couple optimal projection to
find the low dimensional linear mapping to maximize the Wasserstein distance
between projected probability distributions. We characterize the theoretical
property of the finite-sample convergence rate on IPMs and present practical
algorithms for computing this metric. Numerical examples validate our
theoretical results.
| stat |
Individual Treatment Prescription Effect Estimation in a Low Compliance
Setting | Individual Treatment Effect (ITE) estimation is an extensively researched
problem, with applications in various domains. We model the case where there
exists heterogeneous non-compliance to a randomly assigned treatment, a typical
situation in health (because of non-compliance to prescription) or digital
advertising (because of competition and ad blockers for instance). The lower
the compliance, the more the effect of treatment prescription, or individual
prescription effect (IPE), signal fades away and becomes hard to estimate. We
propose a new approach for the estimation of the IPE that takes advantage of
observed compliance information to prevent signal fading. Using the Structural
Causal Model framework and do-calculus, we define a general mediated causal
effect setting and propose a corresponding estimator which consistently
recovers the IPE with asymptotic variance guarantees. Finally, we conduct
experiments on both synthetic and real-world datasets that highlight the
benefit of the approach, which consistently improves state-of-the-art in low
compliance settings
| stat |
Robust regression based on shrinkage estimators | A robust estimator is proposed for the parameters that characterize the
linear regression problem. It is based on the notion of shrinkages, often used
in Finance and previously studied for outlier detection in multivariate data. A
thorough simulation study is conducted to investigate: the efficiency with
normal and heavy-tailed errors, the robustness under contamination, the
computational times, the affine equivariance and breakdown value of the
regression estimator. Two classical data-sets often used in the literature and
a real socio-economic data-set about the Living Environment Deprivation of
areas in Liverpool (UK), are studied. The results from the simulations and the
real data examples show the advantages of the proposed robust estimator in
regression.
| stat |
Self-Tuning Bandits over Unknown Covariate-Shifts | Bandits with covariates, a.k.a. contextual bandits, address situations where
optimal actions (or arms) at a given time $t$, depend on a context $x_t$, e.g.,
a new patient's medical history, a consumer's past purchases. While it is
understood that the distribution of contexts might change over time, e.g., due
to seasonalities, or deployment to new environments, the bulk of studies
concern the most adversarial such changes, resulting in regret bounds that are
often worst-case in nature.
Covariate-shift on the other hand has been considered in classification as a
middle-ground formalism that can capture mild to relatively severe changes in
distributions. We consider nonparametric bandits under such middle-ground
scenarios, and derive new regret bounds that tightly capture a continuum of
changes in context distribution. Furthermore, we show that these rates can be
adaptively attained without knowledge of the time of shift nor the amount of
shift.
| stat |
Equivalence testing for standardized effect sizes in linear regression | In this paper, we introduce equivalence testing procedures for standardized
effect sizes in a linear regression. We show how to define valid hypotheses and
calculate p-values for these tests. Such tests are necessary to confirm the
lack of a meaningful association between an outcome and predictors. A
simulation study is conducted to examine type I error rates and statistical
power. We also compare using equivalence testing as part of a frequentist
testing scheme with an alternative Bayesian testing approach. The results
indicate that the proposed equivalence test is a potentially useful tool for
"testing the null."
| stat |
Adversarial Reinforcement Learning under Partial Observability in
Autonomous Computer Network Defence | Recent studies have demonstrated that reinforcement learning (RL) agents are
susceptible to adversarial manipulation, similar to vulnerabilities previously
demonstrated in the supervised learning setting. While most existing work
studies the problem in the context of computer vision or console games, this
paper focuses on reinforcement learning in autonomous cyber defence under
partial observability. We demonstrate that under the black-box setting, where
the attacker has no direct access to the target RL model, causative
attacks---attacks that target the training process---can poison RL agents even
if the attacker only has partial observability of the environment. In addition,
we propose an inversion defence method that aims to apply the opposite
perturbation to that which an attacker might use to generate their adversarial
samples. Our experimental results illustrate that the countermeasure can
effectively reduce the impact of the causative attack, while not significantly
affecting the training process in non-attack scenarios.
| stat |
Semi-Conditional Normalizing Flows for Semi-Supervised Learning | This paper proposes a semi-conditional normalizing flow model for
semi-supervised learning. The model uses both labelled and unlabeled data to
learn an explicit model of joint distribution over objects and labels.
Semi-conditional architecture of the model allows us to efficiently compute a
value and gradients of the marginal likelihood for unlabeled objects. The
conditional part of the model is based on a proposed conditional coupling
layer. We demonstrate performance of the model for semi-supervised
classification problem on different datasets. The model outperforms the
baseline approach based on variational auto-encoders on MNIST dataset.
| stat |
Error Analysis for the Particle Filter: Methods and Theoretical Support | The particle filter is a popular Bayesian filtering algorithm for use in
cases where the state-space model is nonlinear and/or the random terms (initial
state or noises) are non-Gaussian distributed. We study the behavior of the
error in the particle filter algorithm as the number of particles gets large.
After a decomposition of the error into two terms, we show that the difference
between the estimator and the conditional mean is asymptotically normal when
the resampling is done at every step in the filtering process. Two
nonlinear/non-Gaussian examples are tested to verify this conclusion.
| stat |
Learning Deep-Latent Hierarchies by Stacking Wasserstein Autoencoders | Probabilistic models with hierarchical-latent-variable structures provide
state-of-the-art results amongst non-autoregressive, unsupervised density-based
models. However, the most common approach to training such models based on
Variational Autoencoders (VAEs) often fails to leverage deep-latent
hierarchies; successful approaches require complex inference and optimisation
schemes. Optimal Transport is an alternative, non-likelihood-based framework
for training generative models with appealing theoretical properties, in
principle allowing easier training convergence between distributions. In this
work we propose a novel approach to training models with deep-latent
hierarchies based on Optimal Transport, without the need for highly bespoke
models and inference networks. We show that our method enables the generative
model to fully leverage its deep-latent hierarchy, avoiding the well known
"latent variable collapse" issue of VAEs; therefore, providing qualitatively
better sample generations as well as more interpretable latent representation
than the original Wasserstein Autoencoder with Maximum Mean Discrepancy
divergence.
| stat |
Casting Multiple Shadows: High-Dimensional Interactive Data
Visualisation with Tours and Embeddings | Non-linear dimensionality reduction (NLDR) methods such as t-distributed
stochastic neighbour embedding (t-SNE) are ubiquitous in the natural sciences,
however, the appropriate use of these methods is difficult because of their
complex parameterisations; analysts must make trade-offs in order to identify
structure in the visualisation of an NLDR technique. We present visual
diagnostics for the pragmatic usage of NLDR methods by combining them with a
technique called the tour. A tour is a sequence of interpolated linear
projections of multivariate data onto a lower dimensional space. The sequence
is displayed as a dynamic visualisation, allowing a user to see the shadows the
high-dimensional data casts in a lower dimensional view. By linking the tour to
an NLDR view, we can preserve global structure and through user interactions
like linked brushing observe where the NLDR view may be misleading. We display
several case studies from both simulations and single cell transcriptomics,
that shows our approach is useful for cluster orientation tasks.
| stat |
MASSIVE: Tractable and Robust Bayesian Learning of Many-Dimensional
Instrumental Variable Models | The recent availability of huge, many-dimensional data sets, like those
arising from genome-wide association studies (GWAS), provides many
opportunities for strengthening causal inference. One popular approach is to
utilize these many-dimensional measurements as instrumental variables
(instruments) for improving the causal effect estimate between other pairs of
variables. Unfortunately, searching for proper instruments in a
many-dimensional set of candidates is a daunting task due to the intractable
model space and the fact that we cannot directly test which of these candidates
are valid, so most existing search methods either rely on overly stringent
modeling assumptions or fail to capture the inherent model uncertainty in the
selection process. We show that, as long as at least some of the candidates are
(close to) valid, without knowing a priori which ones, they collectively still
pose enough restrictions on the target interaction to obtain a reliable causal
effect estimate. We propose a general and efficient causal inference algorithm
that accounts for model uncertainty by performing Bayesian model averaging over
the most promising many-dimensional instrumental variable models, while at the
same time employing weaker assumptions regarding the data generating process.
We showcase the efficiency, robustness and predictive performance of our
algorithm through experimental results on both simulated and real-world data.
| stat |
Manifold Fitting in Ambient Space | Modern data sets in many applications no longer comprise samples of real
vectors in a real vector space but samples of much more complex structures
which may be represented as points in a space with certain underlying geometric
structure, namely a manifold. Manifold learning is an emerging field for
learning the underlying structure. The study of manifold learning can be split
into two main branches, namely dimension reduction and manifold fitting. With
the aim of interacting statistics and geometry, we tackle the problem of
manifold fitting in the ambient space. Inspired by the relation between the
eigenvalues of the Laplace-Beltrami operator and the geometry of a manifold, we
aim to find a small set of points that preserve the geometry of the underlying
manifold. Based on this relationship, we extend the idea of subsampling to
noisy datasets in high dimensional space and utilize the Moving Least Squares
(MLS) approach to approximate the underlying manifold. We analyze the two core
steps in our proposed method theoretically and also provide the bounds for the
MLS approach. Our simulation results and real data analysis demonstrate the
superiority of our method in estimating the underlying manifold from noisy
data.
| stat |
Using sex and gender in survey adjustment | Accounting for sex and gender characteristics is a complex, structural
challenge in social science research. While other methodology papers consider
issues surrounding appropriate measurement, we consider how gender and sex
impact adjustments for non-response patterns in sampling and survey estimates.
We consider the problem of survey adjustment arising from the recent push
toward measuring sex or gender as a non-binary construct. This is challenging
not only in that response categories differ between sex and gender measurement,
but also in that both of these attributes are potentially multidimensional. In
this manuscript we reflect on similarities to measuring race/ethnicity before
considering the ethical and statistical implications of the options available
to us. We do not conclude with a single best recommendation but rather an
awareness of the complexity of the issues surrounding this challenge and the
benefits and weaknesses of different approaches.
| stat |
A set of efficient methods to generate high-dimensional binary data with
specified correlation structures | High dimensional correlated binary data arise in many areas, such as observed
genetic variations in biomedical research. Data simulation can help researchers
evaluate efficiency and explore properties of different computational and
statistical methods. Also, some statistical methods, such as Monte-Carlo
methods, rely on data simulation. Lunn and Davies (1998) proposed linear time
complexity methods to generate correlated binary variables with three common
correlation structures. However, it is infeasible to specify unequal
probabilities in their methods. In this manuscript, we introduce several
computationally efficient algorithms that generate high-dimensional binary data
with specified correlation structures and unequal probabilities. Our algorithms
have linear time complexity with respect to the dimension for three commonly
studied correlation structures, namely exchangeable, decaying-product and
K-dependent correlation structures. In addition, we extend our algorithms to
generate binary data of general non-negative correlation matrices with
quadratic time complexity. We provide an R package, CorBin, to implement our
simulation methods. Compared to the existing packages for binary data
generation, the time cost to generate a 100-dimensional binary vector with the
common correlation structures and general correlation matrices can be reduced
up to $10^5$ folds and $10^3$ folds, respectively, and the efficiency can be
further improved with the increase of dimensions. The R package CorBin is
available on CRAN at https://cran.r-project.org/.
| stat |
Robust Hypothesis Testing and Model Selection for Parametric
Proportional Hazard Regression Models | The semi-parametric Cox proportional hazards regression model has been widely
used for many years in several applied sciences. However, a fully parametric
proportional hazards model, if appropriately assumed, can often lead to more
efficient inference. To tackle the extreme non-robustness of the traditional
maximum likelihood estimator in the presence of outliers in the data under such
fully parametric proportional hazard models, a robust estimation procedure has
recently been proposed extending the concept of the minimum density power
divergence estimator (MDPDE) under this set-up. In this paper, we consider the
problem of statistical inference under the parametric proportional hazards
model and develop robust Wald-type hypothesis testing and model selection
procedures using the MDPDEs. We have also derived the necessary asymptotic
results which are used to construct the testing procedure for general composite
hypothesis and study its asymptotic powers. The claimed robustness properties
are studied theoretically via appropriate influence function analyses. We have
studied the finite sample level and power of the proposed MDPDE based Wald type
test through extensive simulations where comparisons are also made with the
existing semi-parametric methods. The important issue of the selection of
appropriate robustness tuning parameter is also discussed. The practical
usefulness of the proposed robust testing and model selection procedures is
finally illustrated through three interesting real data examples.
| stat |
APTER: Aggregated Prognosis Through Exponential Reweighting | This paper considers the task of learning how to make a prognosis of a
patient based on his/her micro-array expression levels. The method is an
application of the aggregation method as recently proposed in the literature on
theoretical machine learning, and excels in its computational convenience and
capability to deal with high-dimensional data. A formal analysis of the method
is given, yielding rates of convergence similar to what traditional techniques
obtain, while it is shown to cope well with an exponentially large set of
features. Those results are supported by numerical simulations on a range of
publicly available survival-micro-array datasets. It is empirically found that
the proposed technique combined with a recently proposed preprocessing
technique gives excellent performances.
| stat |
Outlier detection and a tail-adjusted boxplot based on extreme value
theory | Whether an extreme observation is an outlier or not, depends strongly on the
corresponding tail behaviour of the underlying distribution. We develop an
automatic, data-driven method to identify extreme tail behaviour that deviates
from the intermediate and central characteristics. This allows for detecting
extreme outliers or sets of extreme data that show less spread than the bulk of
the data. To this end we extend a testing method proposed in Bhattacharya et al
2019 for the specific case of heavy tailed models, to all max-domains of
attraction. Consequently we propose a tail-adjusted boxplot which yields a more
accurate representation of possible outliers. Several examples and simulation
results illustrate the finite sample behaviour of this approach.
| stat |
Allocations of Cold Standbys to Series and Parallel Systems with
Dependent Components | In the context of industrial engineering, cold-standby redundancies
allocation strategy is usually adopted to improve the reliability of coherent
systems. This paper investigates optimal allocation strategies of cold standbys
for series and parallel systems comprised of dependent components with
left/right tail weakly stochastic arrangement increasing lifetimes. For the
case of heterogeneous and independent matched cold standbys, it is proved that
better redundancies should be put in the nodes having weaker [better]
components for series [parallel] systems. For the case of homogeneous and
independent cold standbys, it is shown that more redundancies should be put in
standby with weaker [better] components to enhance the reliability of series
[parallel] systems. The results developed here generalize and extend those
corresponding ones in the literature to the case of series and parallel systems
with dependent components. Numerical examples are also presented to provide
guidance for the practical use of our theoretical findings.
| stat |
VARENN: Graphical representation of spatiotemporal data and application
to climate studies | Analyzing and utilizing spatiotemporal big data are essential for studies
concerning climate change. However, such data are not fully integrated into
climate models owing to limitations in statistical frameworks. Herein, we
employ VARENN (visually augmented representation of environment for neural
networks) to efficiently summarize monthly observations of climate data for
1901-2016 into 2-dimensional graphical images. Using red, green, and blue
channels of color images, three different variables are simultaneously
represented in a single image. For global datasets, models were trained via
convolutional neural networks. These models successfully classified rises and
falls in temperature and precipitation. Moreover, similarities between the
input and target variables were observed to have a significant effect on model
accuracy. The input variables had both seasonal and interannual variations,
whose importance was quantified for model efficacy. VARENN is thus an effective
method to summarize spatiotemporal data objectively and accurately.
| stat |
Generalizing a causal effect: sensitivity analysis and missing
covariates | While a randomized controlled trial (RCT) readily measures the average
treatment effect (ATE), this measure may need to be shifted to generalize to a
different population. Standard estimators of the target population treatment
effect are based on the distributional shift in covariates, using inverse
propensity sampling weighting (IPSW) or modeling response with the g-formula.
However, these need covariates that are available both in the RCT and in an
observational sample, which often qualifies very few of them. Here we analyze
how the classic estimators behave when covariates are missing in at least one
of the two datasets - RCT or observational. In line with general
identifiability conditions, these estimators are consistent when including only
treatment effect modifiers that are shifted in the target population. We
compute the expected bias induced by a missing covariate, assuming Gaussian
covariates and a linear model for the conditional ATE function. This enables
sensitivity analysis for each missing covariate pattern. In addition, this
method is particularly useful as it gives the sign of the expected bias. We
also show that there is no gain imputing a partially-unobserved covariate.
Finally we study the replacement of a missing covariate by a proxy, and the
impact of imputation. We illustrate all these results on simulations, as well
as semi-synthetic benchmarks using data from the Tennessee Student/Teacher
Achievement Ratio (STAR), and with a real-world example from the critical care
medical domain.
| stat |
Review on the astronomical estimators of number of flare stars | Review is devoted to estimators (introduced in astronomy by famous astronomer
Ambartsumian and his followers) that are used to estimating the unknown number
of flare stars and other randomly flashing objects (stars with X-ray flares,
solar type stars with superflares). Some important astronomical applications of
them are presented. The development of these methods in astronomy have
proceeded regardless of development of analogous methods of mathematical
statistics. Brief history of this development and parallels with similar
statistical methods is presented.
| stat |
Epsilon Consistent Mixup: An Adaptive Consistency-Interpolation Tradeoff | In this paper we propose $\epsilon$-Consistent Mixup ($\epsilon$mu).
$\epsilon$mu is a data-based structural regularization technique that combines
Mixup's linear interpolation with consistency regularization in the Mixup
direction, by compelling a simple adaptive tradeoff between the two. This
learnable combination of consistency and interpolation induces a more flexible
structure on the evolution of the response across the feature space and is
shown to improve semi-supervised classification accuracy on the SVHN and
CIFAR10 benchmark datasets, yielding the largest gains in the most challenging
low label-availability scenarios. Empirical studies comparing $\epsilon$mu and
Mixup are presented and provide insight into the mechanisms behind
$\epsilon$mu's effectiveness. In particular, $\epsilon$mu is found to produce
more accurate synthetic labels and more confident predictions than Mixup.
| stat |
Estimating adaptive cruise control model parameters from on-board radar
units | Two new methods are presented for estimating car-following model parameters
using data collected from the Adaptive Cruise Control (ACC) enabled vehicles.
The vehicle is assumed to follow a constant time headway relative velocity
model in which the parameters are unknown and to be determined. The first
technique is a batch method that uses a least-squares approach to estimate the
parameters from time series data of the vehicle speed, space gap, and relative
velocity of a lead vehicle. The second method is an online approach that uses a
particle filter to simultaneously estimate both the state of the system and the
model parameters. Numerical experiments demonstrate the accuracy and
computational performance of the methods relative to a commonly used
simulation-based optimization approach. The methods are also assessed on
empirical data collected from a 2019 model year ACC vehicle driven in a highway
environment. Speed, space gap, and relative velocity data are recorded directly
from the factory-installed radar unit via the vehicle's CAN bus. All three
methods return similar mean absolute error values in speed and spacing compared
to the recorded data. The least-squares method has the fastest run-time
performance, and is up to 3 orders of magnitude faster than other methods. The
particle filter is faster than real-time, and therefore is suitable in
streaming applications in which the datasets can grow arbitrarily large.
| stat |
Finite Time Analysis of Linear Two-timescale Stochastic Approximation
with Markovian Noise | Linear two-timescale stochastic approximation (SA) scheme is an important
class of algorithms which has become popular in reinforcement learning (RL),
particularly for the policy evaluation problem. Recently, a number of works
have been devoted to establishing the finite time analysis of the scheme,
especially under the Markovian (non-i.i.d.) noise settings that are ubiquitous
in practice. In this paper, we provide a finite-time analysis for linear two
timescale SA. Our bounds show that there is no discrepancy in the convergence
rate between Markovian and martingale noise, only the constants are affected by
the mixing time of the Markov chain. With an appropriate step size schedule,
the transient term in the expected error bound is $o(1/k^c)$ and the
steady-state term is ${\cal O}(1/k)$, where $c>1$ and $k$ is the iteration
number. Furthermore, we present an asymptotic expansion of the expected error
with a matching lower bound of $\Omega(1/k)$. A simple numerical experiment is
presented to support our theory.
| stat |
Feature Selection Methods for Cost-Constrained Classification in Random
Forests | Cost-sensitive feature selection describes a feature selection problem, where
features raise individual costs for inclusion in a model. These costs allow to
incorporate disfavored aspects of features, e.g. failure rates of as measuring
device, or patient harm, in the model selection process. Random Forests define
a particularly challenging problem for feature selection, as features are
generally entangled in an ensemble of multiple trees, which makes a post hoc
removal of features infeasible. Feature selection methods therefore often
either focus on simple pre-filtering methods, or require many Random Forest
evaluations along their optimization path, which drastically increases the
computational complexity. To solve both issues, we propose Shallow Tree
Selection, a novel fast and multivariate feature selection method that selects
features from small tree structures. Additionally, we also adapt three standard
feature selection algorithms for cost-sensitive learning by introducing a
hyperparameter-controlled benefit-cost ratio criterion (BCR) for each method.
In an extensive simulation study, we assess this criterion, and compare the
proposed methods to multiple performance-based baseline alternatives on four
artificial data settings and seven real-world data settings. We show that all
methods using a hyperparameterized BCR criterion outperform the baseline
alternatives. In a direct comparison between the proposed methods, each method
indicates strengths in certain settings, but no one-fits-all solution exists.
On a global average, we could identify preferable choices among our BCR based
methods. Nevertheless, we conclude that a practical analysis should never rely
on a single method only, but always compare different approaches to obtain the
best results.
| stat |
Bayesian Changepoint Analysis | In my PhD thesis, we elaborate upon Bayesian changepoint analysis, whereby
our focus is on three big topics: approximate sampling via MCMC, exact
inference and uncertainty quantification. Besides, modeling matters are
discussed in an ongoing fashion. Our findings are underpinned through several
changepoint examples with a focus on a well-log drilling data.
| stat |
Gradient Based Memory Editing for Task-Free Continual Learning | We explore task-free continual learning (CL), in which a model is trained to
avoid catastrophic forgetting, but without being provided any explicit task
boundaries or identities. However, since CL models are continually updated, the
utility of stored seen examples may diminish over time. Here, we propose
Gradient based Memory EDiting (GMED), a framework for editing stored examples
in continuous input space via gradient updates, in order to create a wide range
of more ``challenging" examples for replay. GMED-edited examples remain similar
to their unedited forms, but can yield increased loss in the upcoming model
updates, thereby making the future replays more effective in overcoming
catastrophic forgetting. By construction, GMED can be seamlessly applied in
conjunction with other memory-based CL algorithms to bring further improvement.
Experiments on six datasets validate that GMED is effective, and our single
best method significantly outperforms existing approaches on three datasets.
Code and data can be found at https://github.com/INK-USC/GMED.
| stat |
Optimal Semi-supervised Estimation and Inference for High-dimensional
Linear Regression | There are many scenarios such as the electronic health records where the
outcome is much more difficult to collect than the covariates. In this paper,
we consider the linear regression problem with such a data structure under the
high dimensionality. Our goal is to investigate when and how the unlabeled data
can be exploited to improve the estimation and inference of the regression
parameters in linear models, especially in light of the fact that such linear
models may be misspecified in data analysis. In particular, we address the
following two important questions. (1) Can we use the labeled data as well as
the unlabeled data to construct a semi-supervised estimator such that its
convergence rate is faster than the supervised estimators? (2) Can we construct
confidence intervals or hypothesis tests that are guaranteed to be more
efficient or powerful than the supervised estimators? To address the first
question, we establish the minimax lower bound for parameter estimation in the
semi-supervised setting. We show that the upper bound from the supervised
estimators that only use the labeled data cannot attain this lower bound. We
close this gap by proposing a new semi-supervised estimator which attains the
lower bound. To address the second question, based on our proposed
semi-supervised estimator, we propose two additional estimators for
semi-supervised inference, the efficient estimator and the safe estimator. The
former is fully efficient if the unknown conditional mean function is estimated
consistently, but may not be more efficient than the supervised approach
otherwise. The latter usually does not aim to provide fully efficient
inference, but is guaranteed to be no worse than the supervised approach, no
matter whether the linear model is correctly specified or the conditional mean
function is consistently estimated.
| stat |
Health and Demographic Surveillance Systems and the 2030 Agenda:
Sustainable Development Goals | The health and demographic surveillance system (HDSS) is an old method for
intensively monitoring a population to assess the effects of healthcare or
other population-level interventions - often clinical trials. The strengths of
HDSS include very detailed descriptions of whole populations with frequent
updates. This often provides long time series of accurate population and health
indicators for the HDSS study population. The primary weakness of HDSS is that
the data describe only the HDSS study population and cannot be generalized
beyond that.
The 2030 agenda is the ecosystem of activities - many including
population-level monitoring - that relate to the United Nations (UN)
Sustainable Development Goals (SDG). With respect to the 2030 agenda, HDSS can
contribute by: continuing to conduct cause-and-effect studies; contributing to
data triangulation or amalgamation initiatives; characterizing the bias in and
calibrating 'big data'; and contributing more to the rapid training of
data-oriented professionals, especially in the population and health fields.
| stat |
Estimating Average Treatment Effects with a Double-Index Propensity
Score | We consider estimating average treatment effects (ATE) of a binary treatment
in observational data when data-driven variable selection is needed to select
relevant covariates from a moderately large number of available covariates
$\mathbf{X}$. To leverage covariates among $\mathbf{X}$ predictive of the
outcome for efficiency gain while using regularization to fit a parametric
propensity score (PS) model, we consider a dimension reduction of $\mathbf{X}$
based on fitting both working PS and outcome models using adaptive LASSO. A
novel PS estimator, the Double-index Propensity Score (DiPS), is proposed, in
which the treatment status is smoothed over the linear predictors for
$\mathbf{X}$ from both the initial working models. The ATE is estimated by
using the DiPS in a normalized inverse probability weighting (IPW) estimator,
which is found to maintain double-robustness and also local semiparametric
efficiency with a fixed number of covariates $p$. Under misspecification of
working models, the smoothing step leads to gains in efficiency and robustness
over traditional doubly-robust estimators. These results are extended to the
case where $p$ diverges with sample size and working models are sparse.
Simulations show the benefits of the approach in finite samples. We illustrate
the method by estimating the ATE of statins on colorectal cancer risk in an
electronic medical record (EMR) study and the effect of smoking on C-reactive
protein (CRP) in the Framingham Offspring Study.
| stat |
The Wang-Landau Algorithm as Stochastic Optimization and Its
Acceleration | We show that the Wang-Landau algorithm can be formulated as a stochastic
gradient descent algorithm minimizing a smooth and convex objective function,
of which the gradient is estimated using Markov chain Monte Carlo iterations.
The optimization formulation provides us a new way to establish the convergence
rate of the Wang-Landau algorithm, by exploiting the fact that almost surely,
the density estimates (on the logarithmic scale) remain in a compact set, upon
which the objective function is strongly convex. The optimization viewpoint
motivates us to improve the efficiency of the Wang-Landau algorithm using
popular tools including the momentum method and the adaptive learning rate
method. We demonstrate the accelerated Wang-Landau algorithm on a
two-dimensional Ising model and a two-dimensional ten-state Potts model.
| stat |
Challenging nostalgia and performance metrics in baseball | We show that the great baseball players that started their careers before
1950 are overrepresented among rankings of baseball's all time greatest
players. The year 1950 coincides with the decennial US Census that is closest
to when Major League Baseball (MLB) was integrated in 1947. We also show that
performance metrics used to compare players have substantial era biases that
favor players who started their careers before 1950. In showing that the these
players are overrepresented, no individual statistics or era adjusted metrics
are used. Instead, we argue that the eras in which players played are
fundamentally different and are not comparable. In particular, there were
significantly fewer eligible MLB players available at and before 1950. As a
consequence of this and other differences across eras, we argue that popular
opinion, performance metrics, and expert opinion over include players that
started their careers before 1950 in their rankings of baseball's all time
greatest players.
| stat |
Mixture models applied to heterogeneous populations | Mixture models provide a flexible representation of heterogeneity in a finite
number of latent classes. From the Bayesian point of view, Markov Chain Monte
Carlo methods provide a way to draw inferences from these models. In
particular, when the number of subpopulations is considered unknown, more
sophisticated methods are required to perform Bayesian analysis. The Reversible
Jump Markov Chain Monte Carlo is an alternative method for computing the
posterior distribution by simulation in this case. Some problems associated
with the Bayesian analysis of these class of models are frequent, such as the
so-called "label-switching" problem. However, as the level of heterogeneity in
the population increases, these problems are expected to become less frequent
and the model's performance to improve. Thus, the aim of this work is to
evaluate the normal mixture model fit using simulated data under different
settings of heterogeneity and prior information about the mixture proportions.
A simulation study is also presented to evaluate the model's performance
considering the number of components known and estimating it. Finally, the
model is applied to a censored real dataset containing antibody levels of
Cytomegalovirus in individuals.
| stat |
No Free Lunch for Approximate MCMC | It is widely known that the performance of Markov chain Monte Carlo (MCMC)
can degrade quickly when targeting computationally expensive posterior
distributions, such as when the sample size is large. This has motivated the
search for MCMC variants that scale well to large datasets. One general
approach has been to look at only a subsample of the data at every step. In
this note, we point out that well-known MCMC convergence results often imply
that these "subsampling" MCMC algorithms cannot greatly improve performance. We
apply these generic results to realistic statistical problems and proposed
algorithms, and also discuss some design principles suggested by the results.
| stat |
Using social contact data to improve the overall effect estimate of a
cluster-randomized influenza vaccination program in Senegal | This study estimates the overall effect of two influenza vaccination programs
consecutively administered in a cluster-randomized trial in western Senegal
over the course of two influenza seasons from 2009-2011. We apply cutting-edge
methodology combining social contact data with infection data to reduce bias in
estimation arising from contamination between clusters. Our time-varying
estimates reveal a reduction in seasonal influenza from the intervention and a
nonsignificant increase in H1N1 pandemic influenza. We estimate an additive
change in overall cumulative incidence (which was 6.13% in the control arm) of
-0.68 percentage points during Year 1 of the study (95% CI: -2.53, 1.18). When
H1N1 pandemic infections were excluded from analysis, the estimated change was
-1.45 percentage points and was significant (95% CI, -2.81, -0.08). Because
cross-cluster contamination was low (0-3% of contacts for most villages), an
estimator assuming no contamination was only slightly attenuated (-0.65
percentage points). These findings are encouraging for studies carefully
designed to minimize spillover. Further work is needed to estimate
contamination, and its effect on estimation, in a variety of settings.
| stat |
Why is the Mahalanobis Distance Effective for Anomaly Detection? | The Mahalanobis distance-based confidence score, a recently proposed anomaly
detection method for pre-trained neural classifiers, achieves state-of-the-art
performance on both out-of-distribution (OoD) and adversarial examples
detection. This work analyzes why this method exhibits such strong performance
in practical settings while imposing an implausible assumption; namely, that
class conditional distributions of pre-trained features have tied covariance.
Although the Mahalanobis distance-based method is claimed to be motivated by
classification prediction confidence, we find that its superior performance
stems from information not useful for classification. This suggests that the
reason the Mahalanobis confidence score works so well is mistaken, and makes
use of different information from ODIN, another popular OoD detection method
based on prediction confidence. This perspective motivates us to combine these
two methods, and the combined detector exhibits improved performance and
robustness. These findings provide insight into the behavior of neural
classifiers in response to anomalous inputs.
| stat |
Large-Scale Spatiotemporal Density Smoothing with the Graph-fused
Elastic Net: Application to Ride-sourcing Driver Productivity | Ride-sourcing or transportation network companies (TNCs) provide on-demand
transportation service for compensation, connecting drivers of personal
vehicles with passengers through the use of smartphone applications. This
article considers the problem of estimating a spatiotemporally varying
probability distribution for the productivity of a TNC driver, using data on
more than 1 million ride-sourcing trips in Austin, Texas. We extend existing
methods for spatial density smoothing on very large general graphs to the
spatiotemporal setting. Our proposed model allows for distinct spatial and
temporal dynamics, including different degrees of smoothness, and it
appropriately handles vertices with little or no data. Core to our method is an
extension of the Graph-fused Lasso that we refer to as the Graph-fused Elastic
Net (GFEN). Two key empirical findings made possible by our method include 1)
the probability that a TNC driver can expect to earn a living wage in Austin
exhibits high variability in space and time, from as low as 40% to as high as
90%; and 2) for some times and locations, drivers suffer considerable "tail
risk", with the bottom 10% of the earnings distribution falling below $10 per
hour -- grossly below a living wage in Austin for a single adult. All code and
data for the paper are publicly available, as a Shiny app for visualizing the
results and a software package in Julia for implementing the GFEN.
| stat |
DoubleML -- An Object-Oriented Implementation of Double Machine Learning
in Python | DoubleML is an open-source Python library implementing the double machine
learning framework of Chernozhukov et al. (2018) for a variety of causal
models. It contains functionalities for valid statistical inference on causal
parameters when the estimation of nuisance parameters is based on machine
learning methods. The object-oriented implementation of DoubleML provides a
high flexibility in terms of model specifications and makes it easily
extendable. The package is distributed under the MIT license and relies on core
libraries from the scientific Python ecosystem: scikit-learn, numpy, pandas,
scipy, statsmodels and joblib. Source code, documentation and an extensive user
guide can be found at https://github.com/DoubleML/doubleml-for-py and
https://docs.doubleml.org.
| stat |
The Sylvester Graphical Lasso (SyGlasso) | This paper introduces the Sylvester graphical lasso (SyGlasso) that captures
multiway dependencies present in tensor-valued data. The model is based on the
Sylvester equation that defines a generative model. The proposed model
complements the tensor graphical lasso (Greenewald et al., 2019) that imposes a
Kronecker sum model for the inverse covariance matrix by providing an
alternative Kronecker sum model that is generative and interpretable. A
nodewise regression approach is adopted for estimating the conditional
independence relationships among variables. The statistical convergence of the
method is established, and empirical studies are provided to demonstrate the
recovery of meaningful conditional dependency graphs. We apply the SyGlasso to
an electroencephalography (EEG) study to compare the brain connectivity of
alcoholic and nonalcoholic subjects. We demonstrate that our model can
simultaneously estimate both the brain connectivity and its temporal
dependencies.
| stat |
Score Matched Conditional Exponential Families for Likelihood-Free
Inference | To perform Bayesian inference for stochastic simulator models for which the
likelihood is not accessible, Likelihood-Free Inference (LFI) relies on
simulations from the model. Standard LFI methods can be split according to how
these simulations are used: to build an explicit Surrogate Likelihood, or to
accept/reject parameter values according to a measure of distance from the
observations (Approximate Bayesian Computation (ABC)). In both cases,
simulations are adaptively tailored to the value of the observation. Here, we
generate parameter-simulation pairs from the model independently on the
observation, and use them to learn a conditional exponential family likelihood
approximation; to parametrize it, we use Neural Networks whose weights are
tuned with Score Matching. With our likelihood approximation, we can employ
MCMC for doubly intractable distributions to draw samples from the posterior
for any number of observations without additional model simulations, with
performance competitive to comparable approaches. Further, the sufficient
statistics of the exponential family can be used as summaries in ABC,
outperforming the state-of-the-art method in five different models with known
likelihood. Finally, we apply our method to a challenging model from
meteorology.
| stat |
On weakly informative prior distributions for the heterogeneity
parameter in Bayesian random-effects meta-analysis | The normal-normal hierarchical model (NNHM) constitutes a simple and widely
used framework for meta-analysis. In the common case of only few studies
contributing to the meta-analysis, standard approaches to inference tend to
perform poorly, and Bayesian meta-analysis has been suggested as a potential
solution. The Bayesian approach, however, requires the sensible specification
of prior distributions. While non-informative priors are commonly used for the
overall mean effect, the use of weakly informative priors has been suggested
for the heterogeneity parameter, in particular in the setting of (very) few
studies. To date, however, a consensus on how to generally specify a weakly
informative heterogeneity prior is lacking. Here we investigate the problem
more closely and provide some guidance on prior specification.
| stat |
A strategy for the matching of mobile phone signals with census data | Administrative data allows us to count for the number of residents. The
geo-localization of people by mobile phone, by quantifying the number of people
at a given moment in time, enriches the amount of useful information for
"smart" (cities) evaluations. However, using Telecom Italia Mobile (TIM) data,
we are able to characterize the spatio-temporal dynamic of the presences in the
city of just TIM users. A strategy to estimate total presences is needed. In
this paper we propose a strategy to extrapolate the number of total people by
using TIM data only. To do so, we apply a spatial record linkage of mobile
phone data with administrative archives using the number of residents at the
level of sezione di censimento.
| stat |
BEST : A decision tree algorithm that handles missing values | The main contribution of this paper is the development of a new decision tree
algorithm. The proposed approach allows users to guide the algorithm through
the data partitioning process. We believe this feature has many applications
but in this paper we demonstrate how to utilize this algorithm to analyse data
sets containing missing values. We tested our algorithm against simulated data
sets with various missing data structures and a real data set. The results
demonstrate that this new classification procedure efficiently handles missing
values and produces results that are slightly more accurate and more
interpretable than most common procedures without any imputations or
pre-processing.
| stat |
An Adaptive Empirical Bayesian Method for Sparse Deep Learning | We propose a novel adaptive empirical Bayesian method for sparse deep
learning, where the sparsity is ensured via a class of self-adaptive
spike-and-slab priors. The proposed method works by alternatively sampling from
an adaptive hierarchical posterior distribution using stochastic gradient
Markov Chain Monte Carlo (MCMC) and smoothly optimizing the hyperparameters
using stochastic approximation (SA). We further prove the convergence of the
proposed method to the asymptotically correct distribution under mild
conditions. Empirical applications of the proposed method lead to the
state-of-the-art performance on MNIST and Fashion MNIST with shallow
convolutional neural networks and the state-of-the-art compression performance
on CIFAR10 with Residual Networks. The proposed method also improves resistance
to adversarial attacks.
| stat |
Environmental contours as Voronoi cells | Environmental contours are widely used as basis for design of structures
exposed to environmental loads. The basic idea of the method is to decouple the
environmental description from the structural response. This is done by
establishing an envelope of environmental conditions, such that any structure
tolerating loads on this envelope will have a failure probability smaller than
a prescribed value. Specifically, given an $n$-dimensional random variable
$\mathbf{X}$ and a target probability of failure $p_{e}$, an environmental
contour is the boundary of a set $\mathcal{B} \subset \mathbb{R}^{n}$ with the
following property: For any failure set $\mathcal{F} \subset \mathbb{R}^{n}$,
if $\mathcal{F}$ does not intersect the interior of $\mathcal{B}$, then the
probability of failure, $P(\mathbf{X} \in \mathcal{F})$, is bounded above by
$p_{e}$. As is common for many real-world applications, we work under the
assumption that failure sets are convex.
In this paper, we show that such environmental contours may be regarded as
boundaries of Voronoi cells. This geometric interpretation leads to new
theoretical insights and suggests a simple novel construction algorithm that
guarantees the desired probabilistic properties. The method is illustrated with
examples in two and three dimensions, but the results extend to environmental
contours in arbitrary dimensions. Inspired by the Voronoi-Delaunay duality in
the numerical discrete scenario, we are also able to derive an analytical
representation where the environmental contour is considered as a
differentiable manifold, and a criterion for its existence is established.
| stat |
Integrating multi-source block-wise missing data in model selection | For multi-source data, blocks of variable information from certain sources
are likely missing. Existing methods for handling missing data do not take
structures of block-wise missing data into consideration. In this paper, we
propose a Multiple Block-wise Imputation (MBI) approach, which incorporates
imputations based on both complete and incomplete observations. Specifically,
for a given missing pattern group, the imputations in MBI incorporate more
samples from groups with fewer observed variables in addition to the group with
complete observations. We propose to construct estimating equations based on
all available information, and optimally integrate informative estimating
functions to achieve efficient estimators. We show that the proposed method has
estimation and model selection consistency under both fixed-dimensional and
high-dimensional settings. Moreover, the proposed estimator is asymptotically
more efficient than the estimator based on a single imputation from complete
observations only. In addition, the proposed method is not restricted to
missing completely at random. Numerical studies and ADNI data application
confirm that the proposed method outperforms existing variable selection
methods under various missing mechanisms.
| stat |
ensr: R Package for Simultaneous Selection of Elastic Net Tuning
Parameters | Motivation: Elastic net regression is a form of penalized regression that
lies between ridge and least absolute shrinkage and selection operator (LASSO)
regression. The elastic net penalty is a powerful tool controlling the impact
of correlated predictors and the overall complexity of generalized linear
regression models. The elastic net penalty has two tuning parameters:
${\lambda}$ for the complexity and ${\alpha}$ for the compromise between LASSO
and ridge. The R package glmnet provides efficient tools for fitting elastic
net models and selecting ${\lambda}$ for a given ${\alpha}.$ However, glmnet
does not simultaneously search the ${\lambda} - {\alpha}$ space for the
optional elastic net model.
Results: We built the R package ensr, elastic net searcher. enser extends the
functionality of glment to search the ${\lambda} - {\alpha}$ space and identify
an optimal ${\lambda} - {\alpha}$ pair.
Availability: ensr is available from the Comprehensive R Archive Network at
https://cran.r-project.org/package=ensr
| stat |
Variance Reduced Stochastic Proximal Algorithm for AUC Maximization | Stochastic Gradient Descent has been widely studied with classification
accuracy as a performance measure. However, these stochastic algorithms cannot
be directly used when non-decomposable pairwise performance measures are used
such as Area under the ROC curve (AUC) which is a common performance metric
when the classes are imbalanced. There have been several algorithms proposed
for optimizing AUC as a performance metric, and one of the recent being a
stochastic proximal gradient algorithm (SPAM). But the downside of the
stochastic methods is that they suffer from high variance leading to slower
convergence. To combat this issue, several variance reduced methods have been
proposed with faster convergence guarantees than vanilla stochastic gradient
descent. Again, these variance reduced methods are not directly applicable when
non-decomposable performance measures are used. In this paper, we develop a
Variance Reduced Stochastic Proximal algorithm for AUC Maximization
(\textsc{VRSPAM}) and perform a theoretical analysis as well as empirical
analysis to show that our algorithm converges faster than SPAM which is the
previous state-of-the-art for the AUC maximization problem.
| stat |
Individual-Level Modelling of Infectious Disease Data: EpiILM | In this article, we introduce the R package EpiILM, which provides tools for
simulation from, and inference for, discrete-time individual-level models of
infectious disease transmission proposed by Deardon et al. (2010). The
inference is set in a Bayesian framework and is carried out via
Metropolis-Hastings Markov chain Monte Carlo (MCMC). For its fast
implementation, key functions are coded in Fortran. Both spatial and contact
network models are implemented in the package and can be set in either
susceptible-infected (SI) or susceptible-infected-removed (SIR) compartmental
frameworks. The use of the package is demonstrated through examples involving
both simulated and real data.
| stat |
Integral Privacy for Sampling | Differential privacy is a leading protection setting, focused by design on
individual privacy. Many applications, in medical / pharmaceutical domains or
social networks, rather posit privacy at a group level, a setting we call
integral privacy. We aim for the strongest form of privacy: the group size is
in particular not known in advance. We study a problem with related
applications in domains cited above that have recently met with substantial
recent press: sampling.
Keeping correct utility levels in such a strong model of statistical
indistinguishability looks difficult to be achieved with the usual differential
privacy toolbox because it would typically scale in the worst case the
sensitivity by the sample size and so the noise variance by up to its square.
We introduce a trick specific to sampling that bypasses the sensitivity
analysis. Privacy enforces an information theoretic barrier on approximation,
and we show how to reach this barrier with guarantees on the approximation of
the target non private density. We do so using a recent approach to non private
density estimation relying on the original boosting theory, learning the
sufficient statistics of an exponential family with classifiers. Approximation
guarantees cover the mode capture problem. In the context of learning, the
sampling problem is particularly important: because integral privacy enjoys the
same closure under post-processing as differential privacy does, any algorithm
using integrally privacy sampled data would result in an output equally
integrally private. We also show that this brings fairness guarantees on
post-processing that would eventually elude classical differential privacy: any
decision process has bounded data-dependent bias when the data is integrally
privately sampled. Experimental results against private kernel density
estimation and private GANs displays the quality of our results.
| stat |
Bayesian Bootstraps for Massive Data | In this article, we present data-subsetting algorithms that allow for the
approximate and scalable implementation of the Bayesian bootstrap. They are
analogous to two existing algorithms in the frequentist literature: the bag of
little bootstraps (Kleiner et al., 2014) and the subsampled double bootstrap
(SDB; Sengupta et al., 2016). Our algorithms have appealing theoretical and
computational properties that are comparable to those of their frequentist
counterparts. Additionally, we provide a strategy for performing lossless
inference for a class of functionals of the Bayesian bootstrap, and briefly
introduce extensions to the Dirichlet Process.
| stat |
Discussion Paper: Should statistics rescue mathematical modelling? | Statistics experiences a storm around the perceived misuse and possible abuse
of its methods in the context of the so-called reproducibility crisis. The
methods and styles of quantification practiced in mathematical modelling rarely
make it to the headlines, though modelling practitioners writing in
disciplinary journals flag a host of problems in the field. Technical, cultural
and ethical dimensions are simultaneously at play in the current predicaments
of both statistics and mathematical modelling. Since mathematical modelling is
not a discipline like statistics, its shortcomings risk remaining untreated
longer. We suggest that the tools of statistics and its disciplinary
organisation might offer a remedial contribution to mathematical modelling,
standardising methodologies and disseminating good practices. Statistics could
provide scientists and engineers from all disciplines with a point of anchorage
for sound modelling work. This is a vast and long-term undertaking. A step in
the proposed direction is offered here by focusing on the use of statistical
tools for quality assurance of mathematical models. By way of illustration,
techniques for uncertainty quantification, sensitivity analysis and sensitivity
auditing are suggested for incorporation in statistical syllabuses and
practices.
| stat |
Paired Test of Matrix Graphs and Brain Connectivity Analysis | Inferring brain connectivity network and quantifying the significance of
interactions between brain regions are of paramount importance in neuroscience.
Although there have recently emerged some tests for graph inference based on
independent samples, there is no readily available solution to test the change
of brain network for paired and correlated samples. In this article, we develop
a paired test of matrix graphs to infer brain connectivity network when the
groups of samples are correlated. The proposed test statistic is both bias
corrected and variance corrected, and achieves a small estimation error rate.
The subsequent multiple testing procedure built on this test statistic is
guaranteed to asymptotically control the false discovery rate at the
pre-specified level. Both the methodology and theory of the new test are
considerably different from the two independent samples framework, owing to the
strong correlations of measurements on the same subjects before and after the
stimulus activity. We illustrate the efficacy of our proposal through
simulations and an analysis of an Alzheimer's Disease Neuroimaing Initiative
dataset.
| stat |
Geostatistical Modeling of Positive Definite Matrices: An Application to
Diffusion Tensor Imaging | Geostatistical modeling for continuous point-referenced data has been
extensively applied to neuroimaging because it produces efficient and valid
statistical inference. However, diffusion tensor imaging (DTI), a neuroimaging
characterizing the brain structure produces a positive definite (p.d.) matrix
for each voxel. Current geostatistical modeling has not been extended to p.d.
matrices because introducing spatial dependence among positive definite
matrices properly is challenging. In this paper, we use the spatial Wishart
process, a spatial stochastic process (random field) where each p.d.
matrix-variate marginally follows a Wishart distribution, and spatial
dependence between random matrices is induced by latent Gaussian processes.
This process is valid on an uncountable collection of spatial locations and is
almost surely continuous, leading to a reasonable means of modeling spatial
dependence. Motivated by a DTI dataset of cocaine users, we propose a spatial
matrix-variate regression model based on the spatial Wishart process. A
problematic issue is that the spatial Wishart process has no closed-form
density function. Hence, we propose approximation methods to obtain a feasible
working model. A local likelihood approximation method is also applied to
achieve fast computation. The simulation studies and real data analysis
demonstrate that the working model produces reliable inference and improved
performance compared to other methods.
| stat |
An Email Experiment to Identify the Effect of Racial Discrimination on
Access to Lawyers: A Statistical Approach | We consider the problem of conducting an experiment to study the prevalence
of racial bias against individuals seeking legal assistance, in particular
whether lawyers use clues about a potential client's race in deciding whether
to reply to e-mail requests for representations. The problem of discriminating
between potential linear and non-linear effects of a racial signal is
formulated as a statistical inference problem, whose objective is to infer a
parameter determining the shape of a specific function. Various complexities
associated with the design and analysis of this experiment are handled by
applying a novel combination of rigorous, semi-rigorous and rudimentary
statistical techniques. The actual experiment was attempted with a population
of lawyers in Florida, but could not be performed with the desired sample size
due to resource limitations. Nonetheless, it provides a nice demonstration of
the proposed steps involved in conducting such a study.
| stat |
Monte Carlo Co-Ordinate Ascent Variational Inference | In Variational Inference (VI), coordinate-ascent and gradient-based
approaches are two major types of algorithms for approximating
difficult-to-compute probability densities. In real-world implementations of
complex models, Monte Carlo methods are widely used to estimate expectations in
coordinate-ascent approaches and gradients in derivative-driven ones. We
discuss a Monte Carlo Co-ordinate Ascent VI (MC-CAVI) algorithm that makes use
of Markov chain Monte Carlo (MCMC) methods in the calculation of expectations
required within Co-ordinate Ascent VI (CAVI). We show that, under regularity
conditions, an MC-CAVI recursion will get arbitrarily close to a maximiser of
the evidence lower bound (ELBO) with any given high probability. In numerical
examples, the performance of MC-CAVI algorithm is compared with that of MCMC
and -- as a representative of derivative-based VI methods -- of Black Box VI
(BBVI). We discuss and demonstrate MC-CAVI's suitability for models with hard
constraints in simulated and real examples. We compare MC-CAVI's performance
with that of MCMC in an important complex model used in Nuclear Magnetic
Resonance (NMR) spectroscopy data analysis -- BBVI is nearly impossible to be
employed in this setting due to the hard constraints involved in the model.
| stat |
Dimension Reduction Forests: Local Variable Importance using Structured
Random Forests | Random forests are one of the most popular machine learning methods due to
their accuracy and variable importance assessment. However, random forests only
provide variable importance in a global sense. There is an increasing need for
such assessments at a local level, motivated by applications in personalized
medicine, policy-making, and bioinformatics. We propose a new nonparametric
estimator that pairs the flexible random forest kernel with local sufficient
dimension reduction to adapt to a regression function's local structure. This
allows us to estimate a meaningful directional local variable importance
measure at each prediction point. We develop a computationally efficient
fitting procedure and provide sufficient conditions for the recovery of the
splitting directions. We demonstrate significant accuracy gains of our proposed
estimator over competing methods on simulated and real regression problems.
Finally, we apply the proposed method to seasonal particulate matter
concentration data collected in Beijing, China, which yields meaningful local
importance measures. The methods presented here are available in the drforest
Python package.
| stat |
A Bridge between Cross-validation Bayes Factors and Geometric Intrinsic
Bayes Factors | Model Selections in Bayesian Statistics are primarily made with statistics
known as Bayes Factors, which are directly related to Posterior Probabilities
of models. Bayes Factors require a careful assessment of prior distributions as
in the Intrinsic Priors of Berger and Pericchi (1996a) and integration over the
parameter space, which may be highly dimensional. Recently researchers have
been proposing alternatives to Bayes Factors that require neither integration
nor specification of priors. These developments are still in a very early stage
and are known as Prior-free Bayes Factors, Cross-Validation Bayes Factors
(CVBF), and Bayesian "Stacking." This kind of method and Intrinsic Bayes Factor
(IBF) both avoid the specification of prior. However, this Prior-free Bayes
factor might need a careful choice of a training sample size. In this article,
a way of choosing training sample sizes for the Prior-free Bayes factor based
on Geometric Intrinsic Bayes Factors (GIBFs) is proposed and studied. We
present essential examples with a different number of parameters and study the
statistical behavior both numerically and theoretically to explain the ideas
for choosing a feasible training sample size for Prior-free Bayes Factors. We
put forward the "Bridge Rule" as an assignment of a training sample size for
CVBF's that makes them close to Geometric IBFs. We conclude that even though
tractable Geometric IBFs are preferable, CVBF's, using the Bridge Rule, are
useful and economical approximations to Bayes Factors.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.