title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Robust Clustering Using Tau-Scales | K means is a popular non-parametric clustering procedure introduced by
Steinhaus (1956) and further developed by MacQueen (1967). It is known,
however, that K means does not perform well in the presence of outliers.
Cuesta-Albertos et al (1997) introduced a robust alternative, trimmed K means,
which can be tuned to be robust or efficient, but cannot achieve these two
properties simultaneously in an adaptive way. To overcome this limitation we
propose a new robust clustering procedure called K Tau Centers, which is based
on the concept of Tau scale introduced by Yohai and Zamar (1988). We show that
K Tau Centers performs well in extensive simulation studies and real data
examples. We also show that the centers found by the proposed method are
consistent estimators of the "true" centers defined as the minimizers of the
the objective function at the population level.
| stat |
Efficient kernel-based variable selection with sparsistency | Variable selection is central to high-dimensional data analysis, and various
algorithms have been developed. Ideally, a variable selection algorithm shall
be flexible, scalable, and with theoretical guarantee, yet most existing
algorithms cannot attain these properties at the same time. In this article, a
three-step variable selection algorithm is developed, involving kernel-based
estimation of the regression function and its gradient functions as well as a
hard thresholding. Its key advantage is that it assumes no explicit model
assumption, admits general predictor effects, allows for scalable computation,
and attains desirable asymptotic sparsistency. The proposed algorithm can be
adapted to any reproducing kernel Hilbert space (RKHS) with different kernel
functions, and can be extended to interaction selection with slight
modification. Its computational cost is only linear in the data dimension, and
can be further improved through parallel computing. The sparsistency of the
proposed algorithm is established for general RKHS under mild conditions,
including linear and Gaussian kernels as special cases. Its effectiveness is
also supported by a variety of simulated and real examples.
| stat |
Sparse-Input Neural Networks for High-dimensional Nonparametric
Regression and Classification | Neural networks are usually not the tool of choice for nonparametric
high-dimensional problems where the number of input features is much larger
than the number of observations. Though neural networks can approximate complex
multivariate functions, they generally require a large number of training
observations to obtain reasonable fits, unless one can learn the appropriate
network structure. In this manuscript, we show that neural networks can be
applied successfully to high-dimensional settings if the true function falls in
a low dimensional subspace, and proper regularization is used. We propose
fitting a neural network with a sparse group lasso penalty on the first-layer
input weights. This results in a neural net that only uses a small subset of
the original features. In addition, we characterize the statistical convergence
of the penalized empirical risk minimizer to the optimal neural network: we
show that the excess risk of this penalized estimator only grows with the
logarithm of the number of input features; and we show that the weights of
irrelevant features converge to zero. Via simulation studies and data analyses,
we show that these sparse-input neural networks outperform existing
nonparametric high-dimensional estimation methods when the data has complex
higher-order interactions.
| stat |
Pathwise Conditioning of Gaussian Processes | As Gaussian processes are integrated into increasingly complex problem
settings, analytic solutions to quantities of interest become scarcer and
scarcer. Monte Carlo methods act as a convenient bridge for connecting
intractable mathematical expressions with actionable estimates via sampling.
Conventional approaches for simulating Gaussian process posteriors view samples
as vectors drawn from marginal distributions over process values at a finite
number of input location. This distribution-based characterization leads to
generative strategies that scale cubically in the size of the desired random
vector. These methods are, therefore, prohibitively expensive in cases where
high-dimensional vectors - let alone continuous functions - are required. In
this work, we investigate a different line of reasoning. Rather than focusing
on distributions, we articulate Gaussian conditionals at the level of random
variables. We show how this pathwise interpretation of conditioning gives rise
to a general family of approximations that lend themselves to fast sampling
from Gaussian process posteriors. We analyze these methods, along with the
approximation errors they introduce, from first principles. We then complement
this theory, by exploring the practical ramifications of pathwise conditioning
in a various applied settings.
| stat |
Outlier-Detection Based Robust Information Fusion for Networked Systems | We consider state estimation for networked systems where measurements from
sensor nodes are contaminated by outliers. A new hierarchical measurement model
is formulated for outlier detection by integrating the outlier-free measurement
model with a binary indicator variable. The binary indicator variable, which is
assigned a beta-Bernoulli prior, is utilized to characterize if the sensor's
measurement is nominal or an outlier. Based on the proposed outlier-detection
measurement model, both centralized and decentralized information fusion
filters are developed. Specifically, in the centralized approach, all
measurements are sent to a fusion center where the state and outlier indicators
are jointly estimated by employing the mean-field variational Bayesian
inference in an iterative manner. In the decentralized approach, however, every
node shares its information, including the prior and likelihood, only with its
neighbors based on a hybrid consensus strategy. Then each node independently
performs the estimation task based on its own and shared information. In
addition, an approximation distributed solution is proposed to reduce the local
computational complexity and communication overhead. Simulation results reveal
that the proposed algorithms are effective in dealing with outliers compared
with several recent robust solutions.
| stat |
Structured Semantic Model supported Deep Neural Network for
Click-Through Rate Prediction | With the rapid development of online advertising and recommendation systems,
click-through rate prediction is expected to play an increasingly important
role.Recently many DNN-based models which follow a similar Embedding&MLP
paradigm have been proposed, and have achieved good result in image/voice and
nlp fields. In these methods the Wide&Deep model announced by Google plays a
key role.Most models first map large scale sparse input features into
low-dimensional vectors which are transformed to fixed-length vectors, then
concatenated together before being fed into a multilayer perceptron (MLP) to
learn non-linear relations among input features. The number of trainable
variables normally grow dramatically the number of feature fields and the
embedding dimension grow. It is a big challenge to get state-of-the-art result
through training deep neural network and embedding together, which falls into
local optimal or overfitting easily. In this paper, we propose an Structured
Semantic Model (SSM) to tackles this challenge by designing a orthogonal base
convolution and pooling model which adaptively learn the multi-scale base
semantic representation between features supervised by the click label.The
output of SSM are then used in the Wide&Deep for CTR prediction.Experiments on
two public datasets as well as real Weibo production dataset with over 1
billion samples have demonstrated the effectiveness of our proposed approach
with superior performance comparing to state-of-the-art methods.
| stat |
Robust Discrimination between Long-Range Dependence and a Change in Mean | In this paper we introduce a robust to outliers Wilcoxon change-point testing
procedure, for distinguishing between short-range dependent time series with a
change in mean at unknown time and stationary long-range dependent time series.
We establish the asymptotic distribution of the test statistic under the null
hypothesis for $L_1$ near epoch dependent processes and show its consistency
under the alternative. The Wilcoxon-type testing procedure similarly as the
CUSUM-type testing procedure of Berkes, Horv\'ath, Kokoszka and Shao (2006),
requires estimation of the location of a possible change-point, and then using
pre- and post-break subsamples to discriminate between short and long-range
dependence. A simulation study examines the empirical size and power of the
Wilcoxon-type testing procedure in standard cases and with disturbances by
outliers. It shows that in standard cases the Wilcoxon-type testing procedure
behaves equally well as the CUSUM-type testing procedure but outperforms it in
presence of outliers. We also apply both testing procedure to hydrologic data.
| stat |
Deep recurrent Gaussian process with variational Sparse Spectrum
approximation | Modeling sequential data has become more and more important in practice. Some
applications are autonomous driving, virtual sensors and weather forecasting.
To model such systems, so called recurrent models are frequently used. In this
paper we introduce several new Deep recurrent Gaussian process (DRGP) models
based on the Sparse Spectrum Gaussian process (SSGP) and the improved version,
called variational Sparse Spectrum Gaussian process (VSSGP). We follow the
recurrent structure given by an existing DRGP based on a specific variational
sparse Nystr\"om approximation, the recurrent Gaussian process (RGP). Similar
to previous work, we also variationally integrate out the input-space and hence
can propagate uncertainty through the Gaussian process (GP) layers. Our
approach can deal with a larger class of covariance functions than the RGP,
because its spectral nature allows variational integration in all stationary
cases. Furthermore, we combine the (variational) Sparse Spectrum ((V)SS)
approximations with a well known inducing-input regularization framework. We
improve over current state of the art methods in prediction accuracy for
experimental data-sets used for their evaluation and introduce a new data-set
for engine control, named Emission.
| stat |
Multi-task Causal Learning with Gaussian Processes | This paper studies the problem of learning the correlation structure of a set
of intervention functions defined on the directed acyclic graph (DAG) of a
causal model. This is useful when we are interested in jointly learning the
causal effects of interventions on different subsets of variables in a DAG,
which is common in field such as healthcare or operations research. We propose
the first multi-task causal Gaussian process (GP) model, which we call DAG-GP,
that allows for information sharing across continuous interventions and across
experiments on different variables. DAG-GP accommodates different assumptions
in terms of data availability and captures the correlation between functions
lying in input spaces of different dimensionality via a well-defined integral
operator. We give theoretical results detailing when and how the DAG-GP model
can be formulated depending on the DAG. We test both the quality of its
predictions and its calibrated uncertainties. Compared to single-task models,
DAG-GP achieves the best fitting performance in a variety of real and synthetic
settings. In addition, it helps to select optimal interventions faster than
competing approaches when used within sequential decision making frameworks,
like active learning or Bayesian optimization.
| stat |
Differentiable Likelihoods for Fast Inversion of 'Likelihood-Free'
Dynamical Systems | Likelihood-free (a.k.a. simulation-based) inference problems are inverse
problems with expensive, or intractable, forward models. ODE inverse problems
are commonly treated as likelihood-free, as their forward map has to be
numerically approximated by an ODE solver. This, however, is not a fundamental
constraint but just a lack of functionality in classic ODE solvers, which do
not return a likelihood but a point estimate. To address this shortcoming, we
employ Gaussian ODE filtering (a probabilistic numerical method for ODEs) to
construct a local Gaussian approximation to the likelihood. This approximation
yields tractable estimators for the gradient and Hessian of the
(log-)likelihood. Insertion of these estimators into existing gradient-based
optimization and sampling methods engenders new solvers for ODE inverse
problems. We demonstrate that these methods outperform standard likelihood-free
approaches on three benchmark-systems.
| stat |
Assaying Large-scale Testing Models to Interpret COVID-19 Case Numbers | Large-scale testing is considered key to assess the state of the current
COVID-19 pandemic. Yet, the link between the reported case numbers and the true
state of the pandemic remains elusive. We develop mathematical models based on
competing hypotheses regarding this link, thereby providing different
prevalence estimates based on case numbers, and validate them by predicting
SARS-CoV-2-attributed death rate trajectories. Assuming that individuals were
tested based solely on a predefined risk of being infectious implies the
absolute case numbers reflect the prevalence, but turned out to be a poor
predictor, consistently overestimating growth rates at the beginning of two
COVID-19 epidemic waves. In contrast, assuming that testing capacity is fully
exploited performs better. This leads to using the percent-positive rate as a
more robust indicator of epidemic dynamics, however we find it is subject to a
saturation phenomenon that needs to be accounted for as the number of tests
becomes larger.
| stat |
Continuous Time Individual-Level Models of Infectious Disease: a Package
EpiILMCT | This paper describes the R package EpiILMCT, which allows users to study the
spread of infectious disease using continuous time individual level models
(ILMs). The package provides tools for simulation from continuous time ILMs
that are based on either spatial demographic, contact network, or a combination
of both of them, and for the graphical summarization of epidemics. Model
fitting is carried out within a Bayesian Markov Chain Monte Carlo (MCMC)
framework. The continuous time ILMs can be implemented within either
susceptible-infected-removed (SIR) or susceptible-infected-notified-removed
(SINR) compartmental frameworks. As infectious disease data is often partially
observed, data uncertainties in the form of missing infection times - and in
some situations missing removal times - are accounted for using data
augmentation techniques. The package is illustrated using both simulated and an
experimental data set on the spread of the tomato spotted wilt virus (TSWV)
disease.
| stat |
The Pearson Bayes factor: An analytic formula for computing evidential
value from minimal summary statistics | In Bayesian hypothesis testing, evidence for a statistical model is
quantified by the Bayes factor, which represents the relative likelihood of
observed data under that model compared to another competing model. In general,
computing Bayes factors is difficult, as computing the marginal likelihood of
data under a given model requires integrating over a prior distribution of
model parameters. In this paper, I capitalize on a particular choice of prior
distribution that allows the Bayes factor to be expressed without integral
representation and I develop a simple formula -- the Pearson Bayes factor --
that requires only minimal summary statistics commonly reported in scientific
papers, such as the $t$ or $F$ score and the degrees of freedom. In addition to
presenting this new result, I provide several examples of its use and report a
simulation study validating its performance. Importantly, the Pearson Bayes
factor gives applied researchers the ability to compute exact Bayes factors
from minimal summary data, and thus easily assess the evidential value of any
data for which these summary statistics are provided, even when the original
data is not available.
| stat |
Negative binomial-reciprocal inverse Gaussian distribution: Statistical
properties with applications | In this article, we propose a new three parameter distribution by compounding
negative binomial with reciprocal inverse Gaussian model called negative
binomial-reciprocal inverse Gaussian distribution. This model is tractable with
some important properties not only in actuarial science but in other fields as
well where overdispersion pattern is seen. Some basic properties including
recurrence relation of probabilities for computation of successive
probabilities have been discussed. In its compound version, when the claims are
absolutely continuous, an integral equation for the probability density
function is discussed. Brief discussion about extension of univariate version
have also been done to its respective multivariate version. parameters involved
in the model have been estimated by Maximum Likelihood Estimation technique.
Applications of the proposed distribution are carried out by taking two real
count data sets. The result shown that the negative binomial-reciprocal inverse
Gaussian distribution gives better fit when compared to the Poisson and
negative binomial distributions.
| stat |
BAREB: A Bayesian repulsive biclustering model for periodontal data | Preventing periodontal diseases (PD) and maintaining the structure and
function of teeth are important goals for personal oral care. To understand the
heterogeneity in patients with diverse PD patterns, we develop BAREB, a
Bayesian repulsive biclustering method that can simultaneously cluster the PD
patients and their tooth sites after taking the patient- and site- level
covariates into consideration. BAREB uses the determinantal point process (DPP)
prior to induce diversity among different biclusters to facilitate parsimony
and interpretability. Since PD progression is hypothesized to be
spatially-referenced, BAREB factors in the spatial dependence among tooth
sites. In addition, since PD is the leading cause for tooth loss, the missing
data mechanism is non-ignorable. Such nonrandom missingness is incorporated
into BAREB. For the posterior inference, we design an efficient reversible jump
Markov chain Monte Carlo sampler. Simulation studies show that BAREB is able to
accurately estimate the biclusters, and compares favorably to alternatives. For
real world application, we apply BAREB to a dataset from a clinical PD study,
and obtain desirable and interpretable results. A major contribution of this
paper is the Rcpp implementation of BAREB, available at
https://github.com/YanxunXu/ BAREB.
| stat |
Ordered SGD: A New Stochastic Optimization Framework for Empirical Risk
Minimization | We propose a new stochastic optimization framework for empirical risk
minimization problems such as those that arise in machine learning. The
traditional approaches, such as (mini-batch) stochastic gradient descent (SGD),
utilize an unbiased gradient estimator of the empirical average loss. In
contrast, we develop a computationally efficient method to construct a gradient
estimator that is purposely biased toward those observations with higher
current losses. On the theory side, we show that the proposed method minimizes
a new ordered modification of the empirical average loss, and is guaranteed to
converge at a sublinear rate to a global optimum for convex loss and to a
critical point for weakly convex (non-convex) loss. Furthermore, we prove a new
generalization bound for the proposed algorithm. On the empirical side, the
numerical experiments show that our proposed method consistently improves the
test errors compared with the standard mini-batch SGD in various models
including SVM, logistic regression, and deep learning problems.
| stat |
On the convergence of Hamiltonian Monte Carlo | This paper discusses the irreducibility and geometric ergodicity of the
Hamiltonian Monte Carlo (HMC) algorithm. We consider cases where the number of
steps of the symplectic integrator is either fixed or random. Under mild
conditions on the potential $\F$ associated with target distribution $\pi$, we
first show that the Markov kernel associated to the HMC algorithm is
irreducible and recurrent. Under more stringent conditions, we then establish
that the Markov kernel is Harris recurrent. Finally, we provide verifiable
conditions on $\F$ under which the HMC sampler is geometrically ergodic.
| stat |
Usage of multiple RTL features for Earthquake prediction | We construct a classification model that predicts if an earthquake with the
magnitude above a threshold will take place at a given location in a time range
30-180 days from a given moment of time. A common approach is to use expert
forecasts based on features like Region-Time-Length (RTL) characteristics. The
proposed approach uses machine learning on top of multiple RTL features to take
into account effects at various scales and to improve prediction accuracy. For
historical data about Japan earthquakes 1992-2005 and predictions at locations
given in this database the best model has precision up to ~ 0.95 and recall up
to ~ 0.98.
| stat |
A Sequential Significance Test for Treatment by Covariate Interactions | Due to patient heterogeneity in response to various aspects of any treatment
program, biomedical and clinical research is gradually shifting from the
traditional "one-size-fits-all" approach to the new paradigm of personalized
medicine. An important step in this direction is to identify the treatment by
covariate interactions. We consider the setting in which there are potentially
a large number of covariates of interest. Although a number of novel machine
learning methodologies have been developed in recent years to aid in treatment
selection in this setting, few, if any, have adopted formal hypothesis testing
procedures. In this article, we present a novel testing procedure based on
m-out-of-n bootstrap that can be used to sequentially identify variables that
interact with treatment. We study the theoretical properties of the method and
show that it is more effective in controlling the type I error rate and
achieving a satisfactory power as compared to competing methods, via extensive
simulations. Furthermore, the usefulness of the proposed method is illustrated
using real data examples, both from a randomized trial and from an
observational study.
| stat |
Extending inferences from a randomized trial to a new target population | When treatment effect modifiers influence the decision to participate in a
randomized trial, the average treatment effect in the population represented by
the randomized individuals will differ from the effect in other populations. In
this tutorial, we consider methods for extending causal inferences about
time-fixed treatments from a trial to a new target population of
non-participants, using data from a completed randomized trial and baseline
covariate data from a sample from the target population. We examine methods
based on modeling the expectation of the outcome, the probability of
participation, or both (doubly robust). We compare the methods in a simulation
study and show how they can be implemented in software. We apply the methods to
a randomized trial nested within a cohort of trial-eligible patients to compare
coronary artery surgery plus medical therapy versus medical therapy alone for
patients with chronic coronary artery disease. We conclude by discussing issues
that arise when using the methods in applied analyses.
| stat |
Is Group Testing Ready for Prime-time in Disease Identification? | Large scale disease screening is a complicated process in which high costs
must be balanced against pressing public health needs. When the goal is
screening for infectious disease, one approach is group testing in which
samples are initially tested in pools and individual samples are retested only
if the initial pooled test was positive. Intuitively, if the prevalence of
infection is small, this could result in a large reduction of the total number
of tests required. Despite this, the use of group testing in medical studies
has been limited, largely due to skepticism about the impact of pooling on the
accuracy of a given assay. While there is a large body of research addressing
the issue of testing errors in group testing studies, it is customary to assume
that the misclassification parameters are known from an external population
and/or that the values do not change with the group size. Both of these
assumptions are highly questionable for many medical practitioners considering
group testing in their study design. In this article, we explore how the
failure of these assumptions might impact the efficacy of a group testing
design and, consequently, whether group testing is currently feasible for
medical screening. Specifically, we look at how incorrect assumptions about the
sensitivity function at the design stage can lead to poor estimation of a
procedure's overall sensitivity and expected number of tests. Furthermore, if a
validation study is used to estimate the pooled misclassification parameters of
a given assay, we show that the sample sizes required are so large as to be
prohibitive in all but the largest screening programs
| stat |
Bayesian estimation of the functional spatial lag model | The spatial lag model (SLM) has been widely studied in the literature for
spatialised data modeling in various disciplines such as geography, economics,
demography, regional sciences, etc. This is an extension of the classical
linear model that takes into account the proximity of spatial units in
modeling. The extension of the SLM model in the functional framework (the FSLM
model) as well as its estimation by the truncated maximum likelihood technique
have been proposed by \cite{Ahmed}. In this paper, we propose a Bayesian
estimation of the FSLM model. The Bayesian MCMC technique is used as estimation
methods of the parameters of the model. A simulation study is conducted in
order to compare the results of the Bayesian estimation method with the
truncated maximum likelihood method. As an illustration, the proposed Bayesian
method is used to establish a relationship between the unemployment rate and
the curves of illiteracy rate observed in the 45 departments of Senegal.
| stat |
Optimizing tail risks using an importance sampling based extrapolation
for heavy-tailed objectives | Motivated by the prominence of Conditional Value-at-Risk (CVaR) as a measure
for tail risk in settings affected by uncertainty, we develop a new formula for
approximating CVaR based optimization objectives and their gradients from
limited samples. A key difficulty that limits the widespread practical use of
these optimization formulations is the large amount of data required by the
state-of-the-art sample average approximation schemes to approximate the CVaR
objective with high fidelity. Unlike the state-of-the-art sample average
approximations which require impractically large amounts of data in tail
probability regions, the proposed approximation scheme exploits the
self-similarity of heavy-tailed distributions to extrapolate data from suitable
lower quantiles. The resulting approximations are shown to be statistically
consistent and are amenable for optimization by means of conventional gradient
descent. The approximation is guided by means of a systematic
importance-sampling scheme whose asymptotic variance reduction properties are
rigorously examined. Numerical experiments demonstrate the superiority of the
proposed approximations and the ease of implementation points to the
versatility of settings to which the approximation scheme can be applied.
| stat |
Approximate Inference for Constructing Astronomical Catalogs from Images | We present a new, fully generative model for constructing astronomical
catalogs from optical telescope image sets. Each pixel intensity is treated as
a random variable with parameters that depend on the latent properties of stars
and galaxies. These latent properties are themselves modeled as random. We
compare two procedures for posterior inference. One procedure is based on
Markov chain Monte Carlo (MCMC) while the other is based on variational
inference (VI). The MCMC procedure excels at quantifying uncertainty, while the
VI procedure is 1000 times faster. On a supercomputer, the VI procedure
efficiently uses 665,000 CPU cores to construct an astronomical catalog from 50
terabytes of images in 14.6 minutes, demonstrating the scaling characteristics
necessary to construct catalogs for upcoming astronomical surveys.
| stat |
Neuron Shapley: Discovering the Responsible Neurons | We develop Neuron Shapley as a new framework to quantify the contribution of
individual neurons to the prediction and performance of a deep network. By
accounting for interactions across neurons, Neuron Shapley is more effective in
identifying important filters compared to common approaches based on activation
patterns. Interestingly, removing just 30 filters with the highest Shapley
scores effectively destroys the prediction accuracy of Inception-v3 on
ImageNet. Visualization of these few critical filters provides insights into
how the network functions. Neuron Shapley is a flexible framework and can be
applied to identify responsible neurons in many tasks. We illustrate additional
applications of identifying filters that are responsible for biased prediction
in facial recognition and filters that are vulnerable to adversarial attacks.
Removing these filters is a quick way to repair models. Enabling all these
applications is a new multi-arm bandit algorithm that we developed to
efficiently estimate Neuron Shapley values.
| stat |
Defining and Estimating Subgroup Mediation Effects with Semi-Competing
Risks Data | In many medical studies, an ultimate failure event such as death is likely to
be affected by the occurrence and timing of other intermediate clinical events.
Both event times are subject to censoring by loss-to-follow-up but the
nonterminal event may further be censored by the occurrence of the primary
outcome, but not vice versa. To study the effect of an intervention on both
events, the intermediate event may be viewed as a mediator, but conventional
definition of direct and indirect effects is not applicable due to
semi-competing risks data structure. We define three principal strata based on
whether the potential intermediate event occurs before the potential failure
event, which allow proper definition of direct and indirect effects in one
stratum whereas total effects are defined for all strata. We discuss the
identification conditions for stratum-specific effects, and proposed a
semiparametric estimator based on a multivariate logistic stratum membership
model and within-stratum proportional hazards models for the event times. By
treating the unobserved stratum membership as a latent variable, we propose an
EM algorithm for computation. We study the asymptotic properties of the
estimators by the modern empirical process theory and examine the performance
of the estimators in numerical studies.
| stat |
A framework for streamlined statistical prediction using topic models | In the Humanities and Social Sciences, there is increasing interest in
approaches to information extraction, prediction, intelligent linkage, and
dimension reduction applicable to large text corpora. With approaches in these
fields being grounded in traditional statistical techniques, the need arises
for frameworks whereby advanced NLP techniques such as topic modelling may be
incorporated within classical methodologies. This paper provides a classical,
supervised, statistical learning framework for prediction from text, using
topic models as a data reduction method and the topics themselves as
predictors, alongside typical statistical tools for predictive modelling. We
apply this framework in a Social Sciences context (applied animal behaviour) as
well as a Humanities context (narrative analysis) as examples of this
framework. The results show that topic regression models perform comparably to
their much less efficient equivalents that use individual words as predictors.
| stat |
Simultaneous Decorrelation of Matrix Time Series | We propose a contemporaneous bilinear transformation for matrix time series
to alleviate the difficulties in modelling and forecasting large number of time
series together. More precisely the transformed matrix splits into several
small matrices, and those small matrix series are uncorrelated across all
times. Hence an effective dimension-reduction is achieved by modelling each of
those small matrix series separately without the loss of information on the
overall linear dynamics. We adopt the bilinear transformation such that the
rows and the columns of the matrix do not mix together, as they typically
represent radically different features. As the targeted transformation is not
unique, we identify an ideal version through a new normalization, which
facilitates the no-cancellation accumulation of the information from different
time lags. The non-asymptotic error bounds of the estimated transformation are
derived, leading to the uniform convergence rates of the estimation. The
proposed method is illustrated numerically via both simulated and real data
examples.
| stat |
Boost-S: Gradient Boosted Trees for Spatial Data and Its Application to
FDG-PET Imaging Data | Boosting Trees are one of the most successful statistical learning approaches
that involve sequentially growing an ensemble of simple regression trees (i.e.,
"weak learners"). However, gradient boosted trees are not yet available for
spatially correlated data. This paper proposes a new gradient Boosted Trees
algorithm for Spatial Data (Boost-S) with covariate information. Boost-S
integrates the spatial correlation structure into the classical framework of
gradient boosted trees. Each tree is grown by solving a regularized
optimization problem, where the objective function involves two penalty terms
on tree complexity and takes into account the underlying spatial correlation. A
computationally-efficient algorithm is proposed to obtain the ensemble trees.
The proposed Boost-S is applied to the spatially-correlated FDG-PET
(fluorodeoxyglucose-positron emission tomography) imaging data collected during
cancer chemoradiotherapy. Our numerical investigations successfully demonstrate
the advantages of the proposed Boost-S over existing approaches for this
particular application.
| stat |
A fast subsampling method for estimating the distribution of
signal-to-noise ratio statistics in nonparametric time series regression
models | Signal-to-noise ratio (SNR) statistics play a central role in many
applications. A common situation where SNR is studied is when a continuous time
signal is sampled at a fixed frequency with some noise in the background. While
estimation methods exist, little is known about its distribution when the noise
is not weakly stationary. In this paper we develop a nonparametric method to
estimate the distribution of an SNR statistic when the noise belongs to a
fairly general class of stochastic processes that encompasses both short and
long-range dependence, as well as nonlinearities. The method is based on a
combination of smoothing and subsampling techniques. Computations are only
operated at the subsample level, and this allows to manage the typical enormous
sample size produced by modern data acquisition technologies. We derive
asymptotic guarantees for the proposed method, and we show the finite sample
performance based on numerical experiments. Finally, we propose an application
to electroencephalography (EEG) data.
| stat |
Contextualizing E-values for Interpretable Sensitivity to Unmeasured
Confounding Analyses | The strength of evidence provided by epidemiological and observational
studies is inherently limited by the potential for unmeasured confounding.
Researchers should present a quantified sensitivity to unmeasured confounding
analysis that is contextualized by the study's observed covariates. VanderWeele
and Ding's E-value provides an easily calculated metric for the magnitude of
the hypothetical unmeasured confounding required to render the study's result
inconclusive. We propose the Observed Covariate E-value to contextualize the
sensitivity analysis' hypothetical E-value within the actual impact of observed
covariates, individually or within groups. We introduce a sensitivity analysis
figure that presents the Observed Covariate E-values, on the E-value scale,
next to their corresponding observed bias effects, on the original scale of the
study results. This observed bias plot allows easy comparison of the
hypothetical E-values, Observed Covariate E-values, and observed bias effects.
We illustrate the methods with a specific example and provide a supplemental
appendix with modifiable code that teaches how to implement the method and
create a publication quality figure.
| stat |
An Apparent Paradox: A Classifier Trained from a Partially Classified
Sample May Have Smaller Expected Error Rate Than That If the Sample Were
Completely Classified | There has been increasing interest in using semi-supervised learning to form
a classifier. As is well known, the (Fisher) information in an unclassified
feature with unknown class label is less (considerably less for weakly
separated classes) than that of a classified feature which has known class
label. Hence assuming that the labels of the unclassified features are randomly
missing or their missing-label mechanism is simply ignored, the expected error
rate of a classifier formed from a partially classified sample is greater than
that if the sample were completely classified. We propose to treat the labels
of the unclassified features as missing data and to introduce a framework for
their missingness in situations where these labels are not randomly missing. An
examination of several partially classified data sets in the literature
suggests that the unclassified features are not occurring at random but rather
tend to be concentrated in regions of relatively high entropy in the feature
space. Here in the context of two normal classes with a common covariance
matrix we consider the situation where the missingness of the labels of the
unclassified features can be modelled by a logistic model in which the
probability of a missing label for a feature depends on its entropy. Rather
paradoxically, we show that the classifier so formed from the partially
classified sample may have smaller expected error rate that that if the sample
were completely classified.
| stat |
Covariate-adjusted Fisher randomization tests for the average treatment
effect | Fisher's randomization test (FRT) delivers exact $p$-values under the strong
null hypothesis of no treatment effect on any units whatsoever and allows for
flexible covariate adjustment to improve the power. Of interest is whether the
procedure could also be valid for testing the weak null hypothesis of zero
average treatment effect. Towards this end, we evaluate two general strategies
for FRT with covariate-adjusted test statistics: that based on the residuals
from an outcome model with only the covariates, and that based on the output
from an outcome model with both the treatment and the covariates. Based on
theory and simulation, we recommend using the ordinary least squares (OLS) fit
of the observed outcome on the treatment, centered covariates, and their
interactions for covariate adjustment, and conducting FRT with the robust
$t$-value of the treatment as the test statistic. The resulting FRT is
finite-sample exact for the strong null hypothesis, asymptotically valid for
the weak null hypothesis, and more powerful than the unadjusted analog under
alternatives, all irrespective of whether the linear model is correctly
specified or not. We develop the theory for complete randomization, cluster
randomization, stratified randomization, and rerandomization, respectively, and
give a recommendation for the test procedure and test statistic under each
design. We first focus on the finite-population perspective and then extend the
result to the super-population perspective, highlighting the difference in
standard errors. Motivated by the similarity in procedure, we also evaluate the
design-based properties of five existing permutation tests originally for
linear models and show the superiority of the proposed FRT for testing the
treatment effects.
| stat |
Bayesian Optimisation over Multiple Continuous and Categorical Inputs | Efficient optimisation of black-box problems that comprise both continuous
and categorical inputs is important, yet poses significant challenges. We
propose a new approach, Continuous and Categorical Bayesian Optimisation
(CoCaBO), which combines the strengths of multi-armed bandits and Bayesian
optimisation to select values for both categorical and continuous inputs. We
model this mixed-type space using a Gaussian Process kernel, designed to allow
sharing of information across multiple categorical variables, each with
multiple possible values; this allows CoCaBO to leverage all available data
efficiently. We extend our method to the batch setting and propose an efficient
selection procedure that dynamically balances exploration and exploitation
whilst encouraging batch diversity. We demonstrate empirically that our method
outperforms existing approaches on both synthetic and real-world optimisation
tasks with continuous and categorical inputs.
| stat |
Linear Thompson Sampling Revisited | We derive an alternative proof for the regret of Thompson sampling (\ts) in
the stochastic linear bandit setting. While we obtain a regret bound of order
$\widetilde{O}(d^{3/2}\sqrt{T})$ as in previous results, the proof sheds new
light on the functioning of the \ts. We leverage on the structure of the
problem to show how the regret is related to the sensitivity (i.e., the
gradient) of the objective function and how selecting optimal arms associated
to \textit{optimistic} parameters does control it. Thus we show that \ts can be
seen as a generic randomized algorithm where the sampling distribution is
designed to have a fixed probability of being optimistic, at the cost of an
additional $\sqrt{d}$ regret factor compared to a UCB-like approach.
Furthermore, we show that our proof can be readily applied to regularized
linear optimization and generalized linear model problems.
| stat |
Panoramic Robust PCA for Foreground-Background Separation on Noisy,
Free-Motion Camera Video | This work presents a new robust PCA method for foreground-background
separation on freely moving camera video with possible dense and sparse
corruptions. Our proposed method registers the frames of the corrupted video
and then encodes the varying perspective arising from camera motion as missing
data in a global model. This formulation allows our algorithm to produce a
panoramic background component that automatically stitches together corrupted
data from partially overlapping frames to reconstruct the full field of view.
We model the registered video as the sum of a low-rank component that captures
the background, a smooth component that captures the dynamic foreground of the
scene, and a sparse component that isolates possible outliers and other sparse
corruptions in the video. The low-rank portion of our model is based on a
recent low-rank matrix estimator (OptShrink) that has been shown to yield
superior low-rank subspace estimates in practice. To estimate the smooth
foreground component of our model, we use a weighted total variation framework
that enables our method to reliably decouple the true foreground of the video
from sparse corruptions. We perform extensive numerical experiments on both
static and moving camera video subject to a variety of dense and sparse
corruptions. Our experiments demonstrate the state-of-the-art performance of
our proposed method compared to existing methods both in terms of foreground
and background estimation accuracy.
| stat |
Estimation of Markovian-regime-switching models with independent regimes | Markovian-regime-switching (MRS) models are commonly used for modelling
economic time series, including electricity prices where independent regime
models are used, since they can more accurately and succinctly capture
electricity price dynamics than dependent regime MRS models can. We can think
of these independent regime MRS models for electricity prices as a collection
of independent AR(1) processes, of which only one process is observed at each
time; which is observed is determined by a (hidden) Markov chain. Here we
develop novel, computationally feasible methods for MRS models with independent
regimes including forward, backward and EM algorithms. The key idea is to
augment the hidden process with a counter which records the time since the
hidden Markov chain last visited each state that corresponding to an AR(1)
process.
| stat |
Symmetry constrained machine learning | Symmetry, a central concept in understanding the laws of nature, has been
used for centuries in physics, mathematics, and chemistry, to help make
mathematical models tractable. Yet, despite its power, symmetry has not been
used extensively in machine learning, until rather recently. In this article we
show a general way to incorporate symmetries into machine learning models. We
demonstrate this with a detailed analysis on a rather simple real world machine
learning system - a neural network for classifying handwritten digits, lacking
bias terms for every neuron. We demonstrate that ignoring symmetries can have
dire over-fitting consequences, and that incorporating symmetry into the model
reduces over-fitting, while at the same time reducing complexity, ultimately
requiring less training data, and taking less time and resources to train.
| stat |
Large Dimensional Analysis and Improvement of Multi Task Learning | Multi Task Learning (MTL) efficiently leverages useful information contained
in multiple related tasks to help improve the generalization performance of all
tasks. This article conducts a large dimensional analysis of a simple but, as
we shall see, extremely powerful when carefully tuned, Least Square Support
Vector Machine (LSSVM) version of MTL, in the regime where the dimension $p$ of
the data and their number $n$ grow large at the same rate.
Under mild assumptions on the input data, the theoretical analysis of the
MTL-LSSVM algorithm first reveals the "sufficient statistics" exploited by the
algorithm and their interaction at work. These results demonstrate, as a
striking consequence, that the standard approach to MTL-LSSVM is largely
suboptimal, can lead to severe effects of negative transfer but that these
impairments are easily corrected. These corrections are turned into an improved
MTL-LSSVM algorithm which can only benefit from additional data, and the
theoretical performance of which is also analyzed.
As evidenced and theoretically sustained in numerous recent works, these
large dimensional results are robust to broad ranges of data distributions,
which our present experiments corroborate. Specifically, the article reports a
systematically close behavior between theoretical and empirical performances on
popular datasets, which is strongly suggestive of the applicability of the
proposed carefully tuned MTL-LSSVM method to real data. This fine-tuning is
fully based on the theoretical analysis and does not in particular require any
cross validation procedure. Besides, the reported performances on real datasets
almost systematically outperform much more elaborate and less intuitive
state-of-the-art multi-task and transfer learning methods.
| stat |
Area-covering postprocessing of ensemble precipitation forecasts using
topographical and seasonal conditions | Probabilistic weather forecasts from ensemble systems require statistical
postprocessing to yield calibrated and sharp predictive distributions. This
paper presents an area-covering postprocessing method for ensemble
precipitation predictions. We rely on the ensemble model output statistics
(EMOS) approach, which generates probabilistic forecasts with a parametric
distribution whose parameters depend on (statistics of) the ensemble
prediction. A case study with daily precipitation predictions across
Switzerland highlights that postprocessing at observation locations indeed
improves high-resolution ensemble forecasts, with 4.5% CRPS reduction on
average in the case of a lead time of 1 day. Our main aim is to achieve such an
improvement without binding the model to stations, by leveraging topographical
covariates. Specifically, regression coefficients are estimated by weighting
the training data in relation to the topographical similarity between their
station of origin and the prediction location. In our case study, this approach
is found to reproduce the performance of the local model without using local
historical data for calibration. We further identify that one key difficulty is
that postprocessing often degrades the performance of the ensemble forecast
during summer and early autumn. To mitigate, we additionally estimate on the
training set whether postprocessing at a specific location is expected to
improve the prediction. If not, the direct model output is used. This extension
reduces the CRPS of the topographical model by up to another 1.7% on average at
the price of a slight degradation in calibration. In this case, the highest
improvement is achieved for a lead time of 4 days.
| stat |
Spatio-temporal small area surveillance of the Covid-19 pandemics | The emergence of Covid-19 requires new effective tools for epidemiological
surveillance. Spatio-temporal disease mapping models, which allow dealing with
highly disaggregated spatial and temporal units of analysis, are a priority in
this sense. Spatio-temporal models provide a geographically detailed and
temporally updated overview of the current state of the pandemics, making
public health interventions to be more effective. Moreover, the use of
spatio-temporal disease mapping models in the new Covid-19 epidemic context,
facilitates estimating newly demanded epidemiological indicators, such as the
instantaneous reproduction number (R_t), even for small areas. This, in turn,
allows to adapt traditional disease mapping models to these new circumstancies
and make their results more useful in this particular context.
In this paper we propose a new spatio-temporal disease mapping model,
particularly suited to Covid-19 surveillance. As an additional result, we
derive instantaneous reproduction number estimates for small areas, enabling
monitoring this parameter with a high spatial disaggregation. We illustrate the
use of our proposal with the separate study of the disease pandemics in two
Spanish regions. As a result, we illustrate how touristic flows could haved
shaped the spatial distribution of the disease. In these real studies, we also
propose new surveillance tools that can be used by regional public health
services to make a more efficient use of their resources.
| stat |
Spatial modeling of randomly acquired characteristics on outsoles with
application to forensic shoeprint analysis | Footwear comparison is used to link between a suspect's shoe and a footprint
found at a crime scene. Investigators compare the two items using randomly
acquired characteristics (RACs), such as scratches or holes. However, to date,
the distribution of RAC characteristics has not been investigated thoroughly,
and the evidential value of RACs is yet to be explored. An important question
concerns the distribution of the location of RACs on shoe soles, which can
serve as a benchmark for comparison. The location of RACs is modeled here as a
point process over the shoe sole and a data set of 386 independent shoes is
used to estimate its rate function. The analysis is somewhat complicated as the
shoes are differentiated by shape, level of wear and tear and contact surface.
This paper presents methods that take into account these challenges, either by
using natural cubic splines on high resolution data, or by using a
piecewise-constant model on larger regions defined by experts' knowledge. It is
shown that RACs are likely to appear at certain locations, corresponding to the
foot's morphology. The results can guide investigators in determining the
evidential value of footprint comparison.
| stat |
Nonlinear Functional Output Regression: a Dictionary Approach | To address functional-output regression, we introduce projection learning
(PL), a novel dictionary-based approach that learns to predict a function that
is expanded on a dictionary while minimizing an empirical risk based on a
functional loss. PL makes it possible to use non orthogonal dictionaries and
can then be combined with dictionary learning; it is thus much more flexible
than expansion-based approaches relying on vectorial losses. This general
method is instantiated with reproducing kernel Hilbert spaces of vector-valued
functions as kernel-based projection learning (KPL). For the functional square
loss, two closed-form estimators are proposed, one for fully observed output
functions and the other for partially observed ones. Both are backed
theoretically by an excess risk analysis. Then, in the more general setting of
integral losses based on differentiable ground losses, KPL is implemented using
first-order optimization for both fully and partially observed output
functions. Eventually, several robustness aspects of the proposed algorithms
are highlighted on a toy dataset; and a study on two real datasets shows that
they are competitive compared to other nonlinear approaches. Notably, using the
square loss and a learnt dictionary, KPL enjoys a particularily attractive
trade-off between computational cost and performances.
| stat |
Surfing: Iterative optimization over incrementally trained deep networks | We investigate a sequential optimization procedure to minimize the empirical
risk functional $f_{\hat\theta}(x) = \frac{1}{2}\|G_{\hat\theta}(x) - y\|^2$
for certain families of deep networks $G_{\theta}(x)$. The approach is to
optimize a sequence of objective functions that use network parameters obtained
during different stages of the training process. When initialized with random
parameters $\theta_0$, we show that the objective $f_{\theta_0}(x)$ is "nice''
and easy to optimize with gradient descent. As learning is carried out, we
obtain a sequence of generative networks $x \mapsto G_{\theta_t}(x)$ and
associated risk functions $f_{\theta_t}(x)$, where $t$ indicates a stage of
stochastic gradient descent during training. Since the parameters of the
network do not change by very much in each step, the surface evolves slowly and
can be incrementally optimized. The algorithm is formalized and analyzed for a
family of expansive networks. We call the procedure {\it surfing} since it
rides along the peak of the evolving (negative) empirical risk function,
starting from a smooth surface at the beginning of learning and ending with a
wavy nonconvex surface after learning is complete. Experiments show how surfing
can be used to find the global optimum and for compressed sensing even when
direct gradient descent on the final learned network fails.
| stat |
Multichannel adaptive signal detection: Basic theory and literature
review | Multichannel adaptive signal detection jointly uses the test and training
data to form an adaptive detector, and then make a decision on whether a target
exists or not. Remarkably, the resulting adaptive detectors usually possess the
constant false alarm rate (CFAR) properties, and hence no additional CFAR
processing is needed. Filtering is not needed as a processing procedure either,
since the function of filtering is embedded in the adaptive detector. Moreover,
adaptive detection usually exhibits better detection performance than the
filtering-then-CFAR detection technique. Multichannel adaptive signal detection
has been more than 30 years since the first multichannel adaptive detector was
proposed by Kelly in 1986. However, there are fewer overview articles on this
topic. In this paper we give a tutorial overview of multichannel adaptive
signal detection, with emphasis on Gaussian background. We present the main
deign criteria for adaptive detectors, investigate the relationship between
adaptive detection and filtering-then-CFAR detection, relationship between
adaptive detectors and adaptive filters, summarize typical adaptive detectors,
show numerical examples, give comprehensive literature review, and discuss some
possible further research tracks.
| stat |
Bayesian Nonparametric Density Autoregression with Lag Selection | We develop a Bayesian nonparametric autoregressive model applied to flexibly
estimate general transition densities exhibiting nonlinear lag dependence. Our
approach is related to Bayesian density regression using Dirichlet process
mixtures, with the Markovian likelihood defined through the conditional
distribution obtained from the mixture. This results in a Bayesian
nonparametric extension of a mixtures-of-experts model formulation. We address
computational challenges to posterior sampling that arise from the Markovian
structure in the likelihood. The base model is illustrated with synthetic data
from a classical model for population dynamics, as well as a series of waiting
times between eruptions of Old Faithful Geyser. We study inferences available
through the base model before extending the methodology to include automatic
relevance detection among a pre-specified set of lags. Inference for global and
local lag selection is explored with additional simulation studies, and the
methods are illustrated through analysis of an annual time series of pink
salmon abundance in a stream in Alaska. We further explore and compare
transition density estimation performance for alternative configurations of the
proposed model.
| stat |
Model-based clustering of partial records | Partially recorded data are frequently encountered in many applications. In
practice, such datasets are usually clustered by removing incomplete cases or
features with missing values, or by imputing missing values, followed by
application of a clustering algorithm to the resulting altered data set. Here,
we develop clustering methodology through a model-based approach using the
marginal density for the observed values, using a finite mixture model of
multivariate $t$ distributions. We compare our algorithm to the corresponding
full expectation-maximization (EM) approach that considers the missing values
in the incomplete data set and makes a missing at random (MAR) assumption, as
well as case deletion and imputation. Since only the observed values are
utilized, our approach is computationally more efficient than imputation or
full EM. Simulation studies demonstrate that our approach has favorable
recovery of the true cluster partition compared to case deletion and imputation
under various missingness mechanisms, and is more robust to extreme MAR
violations than the full EM approach since it does not use the observed values
to inform those that are missing. Our methodology is demonstrated on a problem
of clustering gamma-ray bursts and is implemented in the
https://github.com/emilygoren/MixtClust R package.
| stat |
Mixed Effect Modeling and Variable Selection for Quantile Regression | It is known that the estimating equations for quantile regression (QR) can be
solved using an EM algorithm in which the M-step is computed via weighted least
squares, with weights computed at the E-step as the expectation of independent
generalized inverse-Gaussian variables. This fact is exploited here to extend
QR to allow for random effects in the linear predictor. Convergence of the
algorithm in this setting is established by showing that it is a generalized
alternating minimization (GAM) procedure. Another modification of the EM
algorithm also allows us to adapt a recently proposed method for variable
selection in mean regression models to the QR setting. Simulations show the
resulting method significantly outperforms variable selection in QR models
using the lasso penalty. Applications to real data include a frailty QR
analysis of hospital stays, and variable selection for age at onset of lung
cancer and for riboflavin production rate using high-dimensional gene
expression arrays for prediction.
| stat |
Discovering contemporaneous and lagged causal relations in
autocorrelated nonlinear time series datasets | We consider causal discovery from time series using conditional independence
(CI) based network learning algorithms such as the PC algorithm. The PC
algorithm is divided into a skeleton phase where adjacencies are determined
based on efficiently selected CI tests and subsequent phases where links are
oriented utilizing the Markov and Faithfulness assumptions. Here we show that
autocorrelation makes the PC algorithm much less reliable with very low
adjacency and orientation detection rates and inflated false positives. We
propose a new algorithm, called PCMCI$^+$ that extends the PCMCI method from
[Runge et al., 2019b] to also include discovery of contemporaneous links. It
separates the skeleton phase for lagged and contemporaneous conditioning sets
and modifies the conditioning sets for the individual CI tests. We show that
this algorithm now benefits from increasing autocorrelation and yields much
more adjacency detection power and especially more orientation recall for
contemporaneous links while controlling false positives and having much shorter
runtimes. Numerical experiments indicate that the algorithm can be of
considerable use in many application scenarios for dozens of variables and
large time delays.
| stat |
Boosting Functional Response Models for Location, Scale and Shape with
an Application to Bacterial Competition | We extend Generalized Additive Models for Location, Scale, and Shape (GAMLSS)
to regression with functional response. This allows us to simultaneously model
point-wise mean curves, variances and other distributional parameters of the
response in dependence of various scalar and functional covariate effects. In
addition, the scope of distributions is extended beyond exponential families.
The model is fitted via gradient boosting, which offers inherent model
selection and is shown to be suitable for both complex model structures and
highly auto-correlated response curves. This enables us to analyze bacterial
growth in \textit{Escherichia coli} in a complex interaction scenario,
fruitfully extending usual growth models.
| stat |
Graph Kernels: A Survey | Graph kernels have attracted a lot of attention during the last decade, and
have evolved into a rapidly developing branch of learning on structured data.
During the past 20 years, the considerable research activity that occurred in
the field resulted in the development of dozens of graph kernels, each focusing
on specific structural properties of graphs. Graph kernels have proven
successful in a wide range of domains, ranging from social networks to
bioinformatics. The goal of this survey is to provide a unifying view of the
literature on graph kernels. In particular, we present a comprehensive overview
of a wide range of graph kernels. Furthermore, we perform an experimental
evaluation of several of those kernels on publicly available datasets, and
provide a comparative study. Finally, we discuss key applications of graph
kernels, and outline some challenges that remain to be addressed.
| stat |
Bayesian hierarchical weighting adjustment and survey inference | We combine Bayesian prediction and weighted inference as a unified approach
to survey inference. The general principles of Bayesian analysis imply that
models for survey outcomes should be conditional on all variables that affect
the probability of inclusion. We incorporate the weighting variables under the
framework of multilevel regression and poststratification, as a byproduct
generating model-based weights after smoothing. We investigate deep
interactions and introduce structured prior distributions for smoothing and
stability of estimates. The computation is done via Stan and implemented in the
open source R package "rstanarm" ready for public use. Simulation studies
illustrate that model-based prediction and weighting inference outperform
classical weighting. We apply the proposal to the New York Longitudinal Study
of Wellbeing. The new approach generates robust weights and increases
efficiency for finite population inference, especially for subsets of the
population.
| stat |
Clustering Time Series with Nonlinear Dynamics: A Bayesian
Non-Parametric and Particle-Based Approach | We propose a general statistical framework for clustering multiple time
series that exhibit nonlinear dynamics into an a-priori-unknown number of
sub-groups. Our motivation comes from neuroscience, where an important problem
is to identify, within a large assembly of neurons, subsets that respond
similarly to a stimulus or contingency. Upon modeling the multiple time series
as the output of a Dirichlet process mixture of nonlinear state-space models,
we derive a Metropolis-within-Gibbs algorithm for full Bayesian inference that
alternates between sampling cluster assignments and sampling parameter values
that form the basis of the clustering. The Metropolis step employs recent
innovations in particle-based methods. We apply the framework to clustering
time series acquired from the prefrontal cortex of mice in an experiment
designed to characterize the neural underpinnings of fear.
| stat |
Quantifying Model Complexity via Functional Decomposition for Better
Post-Hoc Interpretability | Post-hoc model-agnostic interpretation methods such as partial dependence
plots can be employed to interpret complex machine learning models. While these
interpretation methods can be applied regardless of model complexity, they can
produce misleading and verbose results if the model is too complex, especially
w.r.t. feature interactions. To quantify the complexity of arbitrary machine
learning models, we propose model-agnostic complexity measures based on
functional decomposition: number of features used, interaction strength and
main effect complexity. We show that post-hoc interpretation of models that
minimize the three measures is more reliable and compact. Furthermore, we
demonstrate the application of these measures in a multi-objective optimization
approach which simultaneously minimizes loss and complexity.
| stat |
Statistical challenges in the analysis of sequence and structure data
for the COVID-19 spike protein | As the major target of many vaccines and neutralizing antibodies against
SARS-CoV-2, the spike (S) protein is observed to mutate over time. In this
paper, we present statistical approaches to tackle some challenges associated
with the analysis of S-protein data. We build a Bayesian hierarchical model to
study the temporal and spatial evolution of S-protein sequences, after grouping
the sequences into representative clusters. We then apply sampling methods to
investigate possible changes to the S-protein's 3-D structure as a result of
commonly observed mutations. While the increasing spread of D614G variants has
been noted in other research, our results also show that the co-occurring
mutations of D614G together with S477N or A222V may spread even more rapidly,
as quantified by our model estimates.
| stat |
Cross-Fitting and Averaging for Machine Learning Estimation of
Heterogeneous Treatment Effects | We investigate the finite sample performance of sample splitting,
cross-fitting and averaging for the estimation of the conditional average
treatment effect. Recently proposed methods, so-called meta-learners, make use
of machine learning to estimate different nuisance functions and hence allow
for fewer restrictions on the underlying structure of the data. To limit a
potential overfitting bias that may result when using machine learning methods,
cross-fitting estimators have been proposed. This includes the splitting of the
data in different folds to reduce bias and averaging over folds to restore
efficiency. To the best of our knowledge, it is not yet clear how exactly the
data should be split and averaged. We employ a Monte Carlo study with different
data generation processes and consider twelve different estimators that vary in
sample-splitting, cross-fitting and averaging procedures. We investigate the
performance of each estimator independently on four different meta-learners:
the doubly-robust-learner, R-learner, T-learner and X-learner. We find that the
performance of all meta-learners heavily depends on the procedure of splitting
and averaging. The best performance in terms of mean squared error (MSE) among
the sample split estimators can be achieved when applying cross-fitting plus
taking the median over multiple different sample-splitting iterations. Some
meta-learners exhibit a high variance when the lasso is included in the ML
methods. Excluding the lasso decreases the variance and leads to robust and at
least competitive results.
| stat |
A Generalized Hosmer-Lemeshow Goodness-of-Fit Test for a Family of
Generalized Linear Models | Generalized linear models (GLMs) are used within a vast number of application
domains. However, formal goodness of fit (GOF) tests for the overall fit of the
model$-$so-called "global" tests$-$seem to be in wide use only for certain
classes of GLMs. In this paper we develop and apply a new global
goodness-of-fit test, similar to the well-known and commonly used
Hosmer-Lemeshow (HL) test, that can be used with a wide variety of GLMs. The
test statistic is a variant of the HL test statistic, but we rigorously derive
an asymptotically correct sampling distribution of the test statistic using
methods of Stute and Zhu (2002). Our new test is relatively straightforward to
implement and interpret. We demonstrate the test on a real data set, and
compare the performance of our new test with other global GOF tests for GLMs,
finding that our test provides competitive or comparable power in various
simulation settings. Our test also avoids the use of kernel-based estimators,
used in various GOF tests for regression, thereby avoiding the issues of
bandwidth selection and the curse of dimensionality. Since the asymptotic
sampling distribution is known, a bootstrap procedure for the calculation of a
p-value is also not necessary, and we therefore find that performing our test
is computationally efficient.
| stat |
Decentralized Stochastic Gradient Langevin Dynamics and Hamiltonian
Monte Carlo | Stochastic gradient Langevin dynamics (SGLD) and stochastic gradient
Hamiltonian Monte Carlo (SGHMC) are two popular Markov Chain Monte Carlo (MCMC)
algorithms for Bayesian inference that can scale to large datasets, allowing to
sample from the posterior distribution of the parameters of a statistical model
given the input data and the prior distribution over the model parameters.
However, these algorithms do not apply to the decentralized learning setting,
when a network of agents are working collaboratively to learn the parameters of
a statistical model without sharing their individual data due to privacy
reasons or communication constraints. We study two algorithms: Decentralized
SGLD (DE-SGLD) and Decentralized SGHMC (DE-SGHMC) which are adaptations of SGLD
and SGHMC methods that allow scaleable Bayesian inference in the decentralized
setting for large datasets. We show that when the posterior distribution is
strongly log-concave and smooth, the iterates of these algorithms converge
linearly to a neighborhood of the target distribution in the 2-Wasserstein
distance if their parameters are selected appropriately. We illustrate the
efficiency of our algorithms on decentralized Bayesian linear regression and
Bayesian logistic regression problems.
| stat |
Structured penalized regression for drug sensitivity prediction | Large-scale {\it in vitro} drug sensitivity screens are an important tool in
personalized oncology to predict the effectiveness of potential cancer drugs.
The prediction of the sensitivity of cancer cell lines to a panel of drugs is a
multivariate regression problem with high-dimensional heterogeneous multi-omics
data as input data and with potentially strong correlations between the outcome
variables which represent the sensitivity to the different drugs. We propose a
joint penalized regression approach with structured penalty terms which allow
us to utilize the correlation structure between drugs with group-lasso-type
penalties and at the same time address the heterogeneity between omics data
sources by introducing data-source-specific penalty factors to penalize
different data sources differently. By combining integrative penalty factors
(IPF) with tree-guided group lasso, we create the IPF-tree-lasso method. We
present a unified framework to transform more general IPF-type methods to the
original penalized method. Because the structured penalty terms have multiple
parameters, we demonstrate how the interval-search Efficient Parameter
Selection via Global Optimization (EPSGO) algorithm can be used to optimize
multiple penalty parameters efficiently. Simulation studies show that
IPF-tree-lasso can improve the prediction performance compared to other
lasso-type methods, in particular for heterogenous data sources. Finally, we
employ the new methods to analyse data from the Genomics of Drug Sensitivity in
Cancer project.
| stat |
How Fast You Can Actually Fly: A Comparative Investigation of Flight
Airborne Time in China and the U.S | Actual airborne time (AAT) is the time between wheels-off and wheels-on of a
flight. Understanding the behavior of AAT is increasingly important given the
ever growing demand for air travel and flight delays becoming more rampant. As
no research on AAT exists, this paper performs the first empirical analysis of
AAT behavior, comparatively for the U.S. and China. The focus is on how AAT is
affected by scheduled block time (SBT), origin-destination (OD) distance, and
the possible pressure to reduce AAT from other parts of flight operations.
Multiple econometric models are developed. The estimation results show that in
both countries AAT is highly correlated with SBT and OD distance. Flights in
the U.S. are faster than in China. On the other hand, facing ground delay prior
to takeoff, a flight has limited capability to speed up. The pressure from
short turnaround time after landing to reduce AAT is immaterial. Sensitivity
analysis of AAT to flight length and aircraft utilization is further conducted.
Given the more abundant airspace, flexible routing networks, and efficient ATFM
procedures, a counterfactual that the AAT behavior in the U.S. were adopted in
China is examined. We find that by doing so significant efficiency gains could
be achieved in the Chinese air traffic system. On average, 11.8 minutes of AAT
per flight would be saved, coming from both reduction in SBT and reduction in
AAT relative to the new SBT. Systemwide fuel saving would amount to over 300
million gallons with direct airline operating cost saving of nearly $1.3
billion nationwide in 2016.
| stat |
New Perspectives on Centering | Data matrix centering is an ever-present yet under-examined aspect of data
analysis. Functional data analysis (FDA) often operates with a default of
centering such that the vectors in one dimension have mean zero. We find that
centering along the other dimension identifies a novel useful mode of variation
beyond those familiar in FDA. We explore ambiguities in both matrix orientation
and nomenclature. Differences between centerings and their potential
interaction can be easily misunderstood. We propose a unified framework and new
terminology for centering operations. We clearly demonstrate the intuition
behind and consequences of each centering choice with informative graphics. We
also propose a new direction energy hypothesis test as part of a series of
diagnostics for determining which choice of centering is best for a data set.
We explore the application of these diagnostics in several FDA settings.
| stat |
Grid-Parametrize-Split (GriPS) for Improved Scalable Inference in
Spatial Big Data Analysis | Rapid advancements in spatial technologies including Geographic Information
Systems (GIS) and remote sensing have generated massive amounts of spatially
referenced data in a variety of scientific and data-driven industrial
applications. These advancements have led to a substantial, and still
expanding, literature on the modeling and analysis of spatially oriented big
data. In particular, Bayesian inferences for high-dimensional spatial processes
are being sought in a variety of remote-sensing applications including, but not
limited to, modeling next generation Light Detection and Ranging (LiDAR)
systems and other remotely sensed data. Massively scalable spatial processes,
in particular Gaussian processes (GPs), are being explored extensively for the
increasingly encountered big data settings. Recent developments include GPs
constructed from sparse Directed Acyclic Graphs (DAGs) with a limited number of
neighbors (parents) to characterize dependence across the spatial domain. The
DAG can be used to devise fast algorithms for posterior sampling of the latent
process, but these may exhibit pathological behavior in estimating covariance
parameters. While these issues are mitigated by considering marginalized
samplers that exploit the underlying sparse precision matrix, these algorithms
are slower, less flexible, and oblivious of structure in the data. The current
article introduces the Grid-Parametrize-Split (GriPS) approach for conducting
Bayesian inference in spatially oriented big data settings by a combination of
careful model construction and algorithm design to effectuate substantial
improvements in MCMC convergence. We demonstrate the effectiveness of our
proposed methods through simulation experiments and subsequently undertake the
modeling of LiDAR outcomes and production of their predictive maps using G-LiHT
and other remotely sensed variables.
| stat |
Statistical Inference for Local Granger Causality | Granger causality has been employed to investigate causality relations
between components of stationary multiple time series. Here, we generalize this
concept by developing statistical inference for local Granger causality for
multivariate locally stationary processes. Thus, our proposed local Granger
causality approach captures time-evolving causality relationships in
nonstationary processes. The proposed local Granger causality is well
represented in the frequency domain and estimated based on the parametric
time-varying spectral density matrix using the local Whittle likelihood. Under
regularity conditions, we demonstrate that the estimators converge weakly to a
Gaussian process. Additionally, the test statistic for the local Granger
causality is shown to be asymptotically distributed as a quadratic form of a
multivariate normal distribution. The finite sample performance is confirmed
with several simulation studies for multivariate time-varying VAR models. For
practical demonstration, the proposed local Granger causality method uncovered
new functional connectivity relationships between channels in brain signals.
Moreover, the method was able to identify structural changes of Granger
causality in financial data.
| stat |
Hierarchical Optimal Transport for Multimodal Distribution Alignment | In many machine learning applications, it is necessary to meaningfully
aggregate, through alignment, different but related datasets. Optimal transport
(OT)-based approaches pose alignment as a divergence minimization problem: the
aim is to transform a source dataset to match a target dataset using the
Wasserstein distance as a divergence measure. We introduce a hierarchical
formulation of OT which leverages clustered structure in data to improve
alignment in noisy, ambiguous, or multimodal settings. To solve this
numerically, we propose a distributed ADMM algorithm that also exploits the
Sinkhorn distance, thus it has an efficient computational complexity that
scales quadratically with the size of the largest cluster. When the
transformation between two datasets is unitary, we provide performance
guarantees that describe when and how well aligned cluster correspondences can
be recovered with our formulation, as well as provide worst-case dataset
geometry for such a strategy. We apply this method to synthetic datasets that
model data as mixtures of low-rank Gaussians and study the impact that
different geometric properties of the data have on alignment. Next, we applied
our approach to a neural decoding application where the goal is to predict
movement directions and instantaneous velocities from populations of neurons in
the macaque primary motor cortex. Our results demonstrate that when clustered
structure exists in datasets, and is consistent across trials or time points, a
hierarchical alignment strategy that leverages such structure can provide
significant improvements in cross-domain alignment.
| stat |
Sequential Adaptive Design for Jump Regression Estimation | Selecting input variables or design points for statistical models has been of
great interest in adaptive design and active learning. Motivated by two
scientific examples, this paper presents a strategy of selecting the design
points for a regression model when the underlying regression function is
discontinuous. The first example we undertook was for the purpose of
accelerating imaging speed in a high resolution material imaging; the second
was use of sequential design for the purpose of mapping a chemical phase
diagram. In both examples, the underlying regression functions have
discontinuities, so many of the existing design optimization approaches cannot
be applied because they mostly assume a continuous regression function.
Although some existing adaptive design strategies developed from treed
regression models can handle the discontinuities, the Bayesian approaches come
with computationally expensive Markov Chain Monte Carlo techniques for
posterior inferences and subsequent design point selections, which is not
appropriate for the first motivating example that requires computation at least
faster than the original imaging speed. In addition, the treed models are based
on the domain partitioning that are inefficient when the discontinuities occurs
over complex sub-domain boundaries. We propose a simple and effective adaptive
design strategy for a regression analysis with discontinuities: some
statistical properties with a fixed design will be presented first, and then
these properties will be used to propose a new criterion of selecting the
design points for the regression analysis. Sequential design with the new
criterion will be presented with comprehensive simulated examples, and its
application to the two motivating examples will be presented.
| stat |
Scalable Importance Tempering and Bayesian Variable Selection | We propose a Monte Carlo algorithm to sample from high dimensional
probability distributions that combines Markov chain Monte Carlo and importance
sampling. We provide a careful theoretical analysis, including guarantees on
robustness to high dimensionality, explicit comparison with standard Markov
chain Monte Carlo methods and illustrations of the potential improvements in
efficiency. Simple and concrete intuition is provided for when the novel scheme
is expected to outperform standard schemes. When applied to Bayesian
variable-selection problems, the novel algorithm is orders of magnitude more
efficient than available alternative sampling schemes and enables fast and
reliable fully Bayesian inferences with tens of thousand regressors.
| stat |
Organic fiducial inference | A substantial generalisation is put forward of the theory of subjective
fiducial inference as it was outlined in earlier papers. In particular, this
theory is extended to deal with cases where the data are discrete or
categorical rather than continuous, and cases where there was important
pre-data knowledge about some or all of the model parameters. The system for
directly expressing and then handling this pre-data knowledge, which is via
what are referred to as global and local pre-data functions for the parameters
concerned, is distinct from that which involves attempting to directly
represent this knowledge in the form of a prior distribution function over
these parameters, and then using Bayes' theorem. In this regard, the individual
attributes of what are identified as three separate types of fiducial argument,
namely the strong, moderate and weak fiducial arguments, form an integral part
of the theory that is developed. Various practical examples of the application
of this theory are presented, including examples involving binomial, Poisson
and multinomial data. The fiducial distribution functions for the parameters of
the models in these examples are interpreted in terms of a generalised
definition of subjective probability that was set out previously.
| stat |
Relationship between the melting Arctic Sea Ice Extent and North
Atlantic Oscillation Index | Arctic Sea Ice Extent (SIE) maintains the ocean circulation at the
equilibrium and provides strong feedback in the earth's climate system. When
the Arctic Sea Ice melts in the summer, it results in the oceans absorbing and
heating up the Arctic. As Arctic SIE is melting increasing rate, the oceans
absorb and heat up further. This contributes to rising sea surface temperature,
which has a larger impact on global atmospheric pressure. Thus, the climate
scientists are alarmed that global warming will cause the polar ice caps to
melt and that may lead to a "critical instability".
In our study, we construct a phase-space using the velocity and acceleration
of the Arctic Sea Ice Extent (SIE) as two variables. From the data analysis, we
show that in recent decades the melting arctic SIE resulted in increasing
phase-space volume, i.e., the phase-line distribution function has not been
constant along the trajectories. Our aim is to investigate the effect of the
melting Arctic SIE on the climate - particularly on the North Atlantic
Oscillation (NAO) index, a measure of variability in the atmospheric pressure
at sea level. Based on a Granger causal model and a conservative Bootstrap
statistical test, we find that the changing phase-plane SIE does have a
significant (at 0.01 percent) effect on the NAO index. It indicates melting SIE
has a significant effect on the changing weather pattern of the North Atlantic
region, especially in Europe and North America. In addition, we see that the
yearly median of NAO is greater than the yearly average NAO, which indicates
that the distribution of NAO is negatively skewed, even though NAO follows
nearly a mean zero stationary process. Our statistical study hints that we will
soon see a warmer climate in Eastern USA and Northern Europe. Naturally, a
warmer northern Europe would lead to a shrinking SIE, which can be a cause of
alarm.
| stat |
Resolving the Disparate Impact of Uncertainty: Affirmative Action vs.
Affirmative Information | Algorithmic risk assessments hold the promise of greatly advancing accurate
decision-making, but in practice, multiple real-world examples have been shown
to distribute errors disproportionately across demographic groups. In this
paper, we characterize why error disparities arise in the first place. We show
that predictive uncertainty often leads classifiers to systematically
disadvantage groups with lower-mean outcomes, assigning them smaller true and
false positive rates than their higher-mean counterparts. This can occur even
when prediction is group-blind. We prove that to avoid these error imbalances,
individuals in lower-mean groups must either be over-represented among positive
classifications or be assigned more accurate predictions than those in
higher-mean groups. We focus on the latter condition as a solution to bridge
error rate divides and show that data acquisition for low-mean groups can
increase access to opportunity. We call the strategy "affirmative information"
and compare it to traditional affirmative action in the classification task of
identifying creditworthy borrowers.
| stat |
EcoMem: An R package for quantifying ecological memory | Ecological processes may exhibit memory to past disturbances affecting the
resilience of ecosystems to future disturbance. Understanding the role of
ecological memory in shaping ecosystem responses to disturbance under global
change is a critical step toward developing effective adaptive management
strategies to maintain ecosystem function and biodiversity. We developed
EcoMem, an R package for quantifying ecological memory functions using common
environmental time series data (continuous, count, proportional) applying a
Bayesian hierarchical framework. The package estimates memory functions for
continuous and binary (e.g., disturbance chronology) variables making no a
priori assumption on the form of the functions. EcoMem allows users to quantify
ecological memory for a wide range of ecosystem processes and responses. The
utility of the package to advance understanding of the memory of ecosystems to
environmental drivers is demonstrated using a simulated dataset and a case
study assessing the memory of boreal tree growth to insect defoliation.
| stat |
Accounting for phenology in the analysis of animal movement | The analysis of animal tracking data provides an important source of
scientific understanding and discovery in ecology. Observations of animal
trajectories using telemetry devices provide researchers with information about
the way animals interact with their environment and each other. For many
species, specific geographical features in the landscape can have a strong
effect on behavior. Such features may correspond to a single point (e.g., dens
or kill sites), or to higher-dimensional subspaces (e.g., rivers or lakes).
Features may be relatively static in time (e.g., coastlines or home-range
centers), or may be dynamic (e.g., sea ice extent or areas of high-quality
forage for herbivores). We introduce a novel model for animal movement that
incorporates active selection for dynamic features in a landscape.
Our approach is motivated by the study of polar bear (Ursus maritimus)
movement. During the sea ice melt season, polar bears spend much of their time
on sea ice above shallow, biologically productive water where they hunt seals.
The changing distribution and characteristics of sea ice throughout the late
spring through early fall means that the location of valuable habitat is
constantly shifting. We develop a model for the movement of polar bears that
accounts for the effect of this important landscape feature. We introduce a
two-stage procedure for approximate Bayesian inference that allows us to
analyze over 300,000 observed locations of 186 polar bears from 2012--2016. We
use our proposed model to answer a particular question posed by wildlife
managers who seek to cluster polar bears from the Beaufort and Chukchi seas
into sub-populations.
| stat |
Minimal Variance Sampling in Stochastic Gradient Boosting | Stochastic Gradient Boosting (SGB) is a widely used approach to
regularization of boosting models based on decision trees. It was shown that,
in many cases, random sampling at each iteration can lead to better
generalization performance of the model and can also decrease the learning
time. Different sampling approaches were proposed, where probabilities are not
uniform, and it is not currently clear which approach is the most effective. In
this paper, we formulate the problem of randomization in SGB in terms of
optimization of sampling probabilities to maximize the estimation accuracy of
split scoring used to train decision trees. This optimization problem has a
closed-form nearly optimal solution, and it leads to a new sampling technique,
which we call Minimal Variance Sampling (MVS). The method both decreases the
number of examples needed for each iteration of boosting and increases the
quality of the model significantly as compared to the state-of-the art sampling
methods. The superiority of the algorithm was confirmed by introducing MVS as a
new default option for subsampling in CatBoost, a gradient boosting library
achieving state-of-the-art quality on various machine learning tasks.
| stat |
Combined background information for meta-analysis evaluation | Massive numbers of meta-analysis studies are being published. A Google
Scholar search of "systematic review and meta-analysis" returns about 452k hits
since 2014. The search was done on Jan 14, 2019. There is a need to have some
way to judge the reliability of a positive claim made in a meta-analysis that
uses observational studies. Our idea is to examine the quality of the
observational studies used in the meta-analysis and to examine the
heterogeneity of those studies. We provide background information and examples:
a listing of negative studies, a simulation of p-value plots, and multiple
examples of p-value plots.
| stat |
Unbiased Recommender Learning from Missing-Not-At-Random Implicit
Feedback | Recommender systems widely use implicit feedback such as click data because
of its general availability. Although the presence of clicks signals the users'
preference to some extent, the lack of such clicks does not necessarily
indicate a negative response from the users, as it is possible that the users
were not exposed to the items (positive-unlabeled problem). This leads to a
difficulty in predicting the users' preferences from implicit feedback.
Previous studies addressed the positive-unlabeled problem by uniformly
upweighting the loss for the positive feedback data or estimating the
confidence of each data having relevance information via the EM-algorithm.
However, these methods failed to address the missing-not-at-random problem in
which popular or frequently recommended items are more likely to be clicked
than other items even if a user does not have a considerable interest in them.
To overcome these limitations, we first define an ideal loss function to be
optimized to realize recommendations that maximize the relevance and propose an
unbiased estimator for the ideal loss. Subsequently, we analyze the variance of
the proposed unbiased estimator and further propose a clipped estimator that
includes the unbiased estimator as a special case. We demonstrate that the
clipped estimator is expected to improve the performance of the recommender
system, by considering the bias-variance trade-off. We conduct semi-synthetic
and real-world experiments and demonstrate that the proposed method largely
outperforms the baselines. In particular, the proposed method works better for
rare items that are less frequently observed in the training data. The findings
indicate that the proposed method can better achieve the objective of
recommending items with the highest relevance.
| stat |
Permutation-based uncertainty quantification about a mixing distribution | Nonparametric estimation of a mixing distribution based on data coming from a
mixture model is a challenging problem. Beyond estimation, there is interest in
uncertainty quantification, e.g., confidence intervals for features of the
mixing distribution. This paper focuses on estimation via the predictive
recursion algorithm, and here we take advantage of this estimator's seemingly
undesirable dependence on the data ordering to obtain a permutation-based
approximation of the sampling distribution which can be used to quantify
uncertainty. Theoretical and numerical results confirm that the proposed method
leads to valid confidence intervals, at least approximately.
| stat |
Incremental Intervention Effects in Studies with Dropout and Many
Timepoints | Modern longitudinal studies collect feature data at many timepoints, often of
the same order of sample size. Such studies are typically affected by {dropout}
and positivity violations. We tackle these problems by generalizing effects of
recent incremental interventions (which shift propensity scores rather than set
treatment values deterministically) to accommodate multiple outcomes and
subject dropout. We give an identifying expression for incremental effects when
dropout is conditionally ignorable (without requiring treatment positivity),
and derive the nonparametric efficiency bound for estimating such effects. Then
we present efficient nonparametric estimators, showing that they converge at
fast parametric rates and yield uniform inferential guarantees, even when
nuisance functions are estimated flexibly at slower rates. We also study the
efficiency of incremental effects relative to more conventional deterministic
effects in a novel infinite time horizon setting, where the number of
timepoints can grow with sample size, and show that incremental effects yield
near-exponential gains in this setup. Finally we conclude with simulations and
apply our methods in a study of the effect of low-dose aspirin on pregnancy
outcomes.
| stat |
Estimation of separable direct and indirect effects in continuous time | Many research questions involve time-to-event outcomes that can be prevented
from occurring due to competing events. In these settings, we must be careful
about the causal interpretation of classical statistical estimands. In
particular, estimands on the hazard scale, such as ratios of cause specific or
subdistribution hazards, are fundamentally hard to be interpret causally.
Estimands on the risk scale, such as contrasts of cumulative incidence
functions, do have a causal interpretation, but they only capture the total
effect of the treatment on the event of interest; that is, effects both through
and outside of the competing event. To disentangle causal treatment effects on
the event of interest and competing events, the separable direct and indirect
effects were recently introduced. Here we provide new results on the estimation
of direct and indirect separable effects in continuous time. In particular, we
derive the nonparametric influence function in continuous time and use it to
construct an estimator that has certain robustness properties. We also propose
a simple estimator based on semiparametric models for the two cause specific
hazard functions. We describe the asymptotic properties of these estimators,
and present results from simulation studies, suggesting that the estimators
behave satisfactorily in finite samples. Finally, we re-analyze the prostate
cancer trial from Stensrud et al (2020).
| stat |
Efficient Bayesian Inference of Sigmoidal Gaussian Cox Processes | We present an approximate Bayesian inference approach for estimating the
intensity of an inhomogeneous Poisson process, where the intensity function is
modelled using a Gaussian process (GP) prior via a sigmoid link function.
Augmenting the model using a latent marked Poisson process and P\'olya--Gamma
random variables we obtain a representation of the likelihood which is
conjugate to the GP prior. We estimate the posterior using a variational
free--form mean field optimisation together with the framework of sparse GPs.
Furthermore, as alternative approximation we suggest a sparse Laplace's method
for the posterior, for which an efficient expectation--maximisation algorithm
is derived to find the posterior's mode. Both algorithms compare well against
exact inference obtained by a Markov Chain Monte Carlo sampler and standard
variational Gauss approach solving the same model, while being one order of
magnitude faster. Furthermore, the performance and speed of our method is
competitive with that of another recently proposed Poisson process model based
on a quadratic link function, while not being limited to GPs with squared
exponential kernels and rectangular domains.
| stat |
A hierarchical model for estimating exposure-response curves from
multiple studies | Cookstove replacement trials have found mixed results on their impact on
respiratory health. The limited range of concentrations and small sample sizes
of individual studies are important factors that may be limiting their
statistical power. We present a hierarchical approach to modeling exposure
concentrations and pooling data from multiple studies in order to estimate a
common exposure-response curve. The exposure concentration model accommodates
temporally sparse, clustered longitudinal observations. The exposure-response
curve model provides a flexible, semi-parametric estimate of the
exposure-response relationship while accommodating heterogeneous clustered
data. We apply this model to data from three studies of cookstoves and
respiratory infections in children in Nepal, which represent three study types:
crossover trial, parallel trial, and case-control study. We find evidence of
increased odds of disease for particulate matter concentrations between 50 and
200 $\mu$g/m$^3$ and a flattening of the exposure-response curve for higher
exposure concentrations. The model we present can incorporate additional
studies and be applied to other settings.
| stat |
Kernel MMD Two-Sample Tests for Manifold Data | We present a study of kernel MMD two-sample test statistics in the manifold
setting, assuming the high-dimensional observations are close to a
low-dimensional manifold. We characterize the property of the test (level and
power) in relation to the kernel bandwidth, the number of samples, and the
intrinsic dimensionality of the manifold. Specifically, we show that when data
densities are supported on a $d$-dimensional sub-manifold $\mathcal{M}$
embedded in an $m$-dimensional space, the kernel MMD two-sample test for data
sampled from a pair of distributions $(p, q)$ that are H\"older with order
$\beta$ is consistent and powerful when the number of samples $n$ is greater
than $\delta_2(p,q)^{-2-d/\beta}$ up to certain constant, where $\delta_2$ is
the squared $\ell_2$-divergence between two distributions on manifold.
Moreover, to achieve testing consistency under this scaling of $n$, our theory
suggests that the kernel bandwidth $\gamma$ scales with $n^{-1/(d+2\beta)}$.
These results indicate that the kernel MMD two-sample test does not have a
curse-of-dimensionality when the data lie on the low-dimensional manifold. We
demonstrate the validity of our theory and the property of the MMD test for
manifold data using several numerical experiments.
| stat |
Composite local low-rank structure in learning drug sensitivity | The molecular characterization of tumor samples by multiple omics data sets
of different types or modalities (e.g. gene expression, mutation, CpG
methylation) has become an invaluable source of information for assessing the
expected performance of individual drugs and their combinations. Merging the
relevant information from the omics data modalities provides the statistical
basis for determining suitable therapies for specific cancer patients.
Different data modalities may each have their specific structures that need to
be taken into account during inference. In this paper, we assume that each
omics data modality has a low-rank structure with only few relevant features
that affect the prediction and we propose to use a composite local nuclear norm
penalization for learning drug sensitivity. Numerical results show that the
composite low-rank structure can improve the prediction performance compared to
using a global low-rank approach or elastic net regression.
| stat |
Towards causality-aware predictions in static anticausal machine
learning tasks: the linear structural causal model case | We propose a counterfactual approach to train ``causality-aware" predictive
models that are able to leverage causal information in static anticausal
machine learning tasks (i.e., prediction tasks where the outcome influences the
features). In applications plagued by confounding, the approach can be used to
generate predictions that are free from the influence of observed confounders.
In applications involving observed mediators, the approach can be used to
generate predictions that only capture the direct or the indirect causal
influences. Mechanistically, we train supervised learners on (counterfactually)
simulated features which retain only the associations generated by the causal
relations of interest. We focus on linear models, where analytical results
connecting covariances, causal effects, and prediction mean squared errors are
readily available. Quite importantly, we show that our approach does not
require knowledge of the full causal graph. It suffices to know which variables
represent potential confounders and/or mediators. We discuss the stability of
the method with respect to dataset shifts generated by selection biases and
validate the approach using synthetic data experiments.
| stat |
Risk bounds when learning infinitely many response functions by ordinary
linear regression | Consider the problem of learning a large number of response functions
simultaneously based on the same input variables. The training data consist of
a single independent random sample of the input variables drawn from a common
distribution together with the associated responses. The input variables are
mapped into a high-dimensional linear space, called the feature space, and the
response functions are modelled as linear functionals of the mapped features,
with coefficients calibrated via ordinary least squares. We provide convergence
guarantees on the worst-case excess prediction risk by controlling the
convergence rate of the excess risk uniformly in the response function. The
dimension of the feature map is allowed to tend to infinity with the sample
size. The collection of response functions, although potentiallyinfinite, is
supposed to have a finite Vapnik-Chervonenkis dimension. The bound derived can
be applied when building multiple surrogate models in a reasonable computing
time.
| stat |
Combining Unsupervised and Supervised Learning for Asset Class Failure
Prediction in Power Systems | In power systems, an asset class is a group of power equipment that has the
same function and shares similar electrical or mechanical characteristics.
Predicting failures for different asset classes is critical for electric
utilities towards developing cost-effective asset management strategies.
Previously, physical age based Weibull distribution has been widely used to
failure prediction. However, this mathematical model cannot incorporate asset
condition data such as inspection or testing results. As a result, the
prediction cannot be very specific and accurate for individual assets. To solve
this important problem, this paper proposes a novel and comprehensive
data-driven approach based on asset condition data: K-means clustering as an
unsupervised learning method is used to analyze the inner structure of
historical asset condition data and produce the asset conditional ages;
logistic regression as a supervised learning method takes in both asset
physical ages and conditional ages to classify and predict asset statuses.
Furthermore, an index called average aging rate is defined to quantify, track
and estimate the relationship between asset physical age and conditional age.
This approach was applied to an urban distribution system in West Canada to
predict medium-voltage cable failures. Case studies and comparison with
standard Weibull distribution are provided. The proposed approach demonstrates
superior performance and practicality for predicting asset class failures in
power systems.
| stat |
Low Distortion Block-Resampling with Spatially Stochastic Networks | We formalize and attack the problem of generating new images from old ones
that are as diverse as possible, only allowing them to change without
restrictions in certain parts of the image while remaining globally consistent.
This encompasses the typical situation found in generative modelling, where we
are happy with parts of the generated data, but would like to resample others
("I like this generated castle overall, but this tower looks unrealistic, I
would like a new one"). In order to attack this problem we build from the best
conditional and unconditional generative models to introduce a new network
architecture, training procedure, and algorithm for resampling parts of the
image as desired.
| stat |
Recognizing a Spatial Extreme dependence structure: A Deep Learning
approach | Understanding the behaviour of environmental extreme events is crucial for
evaluating economic losses, assessing risks, health care and many other
aspects. In the spatial context, relevant for environmental events, the
dependence structure plays a central rule, as it influence joined extreme
events and extrapolation on them. So that, recognising or at least having
preliminary informations on patterns of these dependence structures is a
valuable knowledge for understanding extreme events. In this study, we address
the question of automatic recognition of spatial Asymptotic Dependence (AD)
versus Asymptotic independence (AI), using Convolutional Neural Network (CNN).
We have designed an architecture of Convolutional Neural Network to be an
efficient classifier of the dependence structure. Upper and lower tail
dependence measures are used to train the CNN. We have tested our methodology
on simulated and real data sets: air temperature data at two meter over Iraq
land and Rainfall data in the east cost of Australia.
| stat |
Locally Private Bayesian Inference for Count Models | We present a general method for privacy-preserving Bayesian inference in
Poisson factorization, a broad class of models that includes some of the most
widely used models in the social sciences. Our method satisfies limited
precision local privacy, a generalization of local differential privacy, which
we introduce to formulate privacy guarantees appropriate for sparse count data.
We develop an MCMC algorithm that approximates the locally private posterior
over model parameters given data that has been locally privatized by the
geometric mechanism (Ghosh et al., 2012). Our solution is based on two
insights: 1) a novel reinterpretation of the geometric mechanism in terms of
the Skellam distribution (Skellam, 1946) and 2) a general theorem that relates
the Skellam to the Bessel distribution (Yuan & Kalbfleisch, 2000). We
demonstrate our method in two case studies on real-world email data in which we
show that our method consistently outperforms the commonly-used naive approach,
obtaining higher quality topics in text and more accurate link prediction in
networks. On some tasks, our privacy-preserving method even outperforms
non-private inference which conditions on the true data.
| stat |
Improved Approximations of Hedges' g* | Hedges' unbiased estimator g* has been broadly used in statistics. We propose
a sequence of polynomials to better approximate the multiplicative correction
factor of g* by incorporating analytic estimations to the ratio of gamma
functions.
| stat |
A Holistic Approach to Forecasting Wholesale Energy Market Prices | Electricity market price predictions enable energy market participants to
shape their consumption or supply while meeting their economic and
environmental objectives. By utilizing the basic properties of the
supply-demand matching process performed by grid operators, known as Optimal
Power Flow (OPF), we develop a methodology to recover energy market's structure
and predict the resulting nodal prices by using only publicly available data,
specifically grid-wide generation type mix, system load, and historical prices.
Our methodology uses the latest advancements in statistical learning to cope
with high dimensional and sparse real power grid topologies, as well as scarce,
public market data, while exploiting structural characteristics of the
underlying OPF mechanism. Rigorous validations using the Southwest Power Pool
(SPP) market data reveal a strong correlation between the grid level mix and
corresponding market prices, resulting in accurate day-ahead predictions of
real time prices. The proposed approach demonstrates remarkable proximity to
the state-of-the-art industry benchmark while assuming a fully decentralized,
market-participant perspective. Finally, we recognize the limitations of the
proposed and other evaluated methodologies in predicting large price spike
values.
| stat |
Simultaneous Differential Network Analysis and Classification for
High-dimensional Matrix-variate Data, with application to Brain Connectivity
Alteration Detection and fMRI-guided Medical Diagnoses of Alzheimer's Disease | Alzheimer's disease (AD) is the most common form of dementia, which causes
problems with memory, thinking and behavior. Growing evidence has shown that
the brain connectivity network experiences alterations for such a complex
disease. Network comparison, also known as differential network analysis, is
thus particularly powerful to reveal the disease pathologies and identify
clinical biomarkers for medical diagnoses (classification). Data from
neurophysiological measurements are multi-dimensional and in matrix-form, which
poses major challenges in brain connectivity analysis and medical diagnoses.
Naive vectorization method is not sufficient as it ignores the structural
information within the matrix. In the article, we adopt the Kronecker product
covariance matrix framework to capture both spatial and temporal correlations
of the matrix-variate data while the temporal covariance matrix is treated as a
nuisance parameter. By recognizing that the strengths of network connections
may vary across subjects, we develop an ensemble-learning procedure, which
identifies the differential interaction patterns of brain regions between the
AD group and the control group and conducts medical diagnosis (classification)
of AD simultaneously. We applied the proposed procedure to functional
connectivity analysis of fMRI dataset related with Alzheimer's disease. The hub
nodes and differential interaction patterns identified are consistent with
existing experimental studies, and satisfactory out-of-sample classification
performance is achieved for medical diagnosis of Alzheimer's disease. An R
package \SDNCMV" for implementation is available at
https://github.com/heyongstat/SDNCMV.
| stat |
Impact of COVID-19 on travel behaviour, transport, lifestyles and
location choices in Scotland | In March 2020, the UK and Scottish Governments imposed a lockdown restricting
everyday life activities to only the most essential. These Governmental
measures together with individual choices to refrain from travelling during the
COVID-19 pandemic have had a profound effect on transport related activity. In
the current investigation an online questionnaire was distributed to 994
Scottish residents in order to identify travel habits, attitudes and
preferences during the different phases of the COVID-19 pandemic outbreak and
anticipated travel habits after the pandemic. Quota constraints were enforced
for age, gender and household income to ensure the sample was representative of
the Scottish population as a whole. Perceptions of risk, trust in information
sources and compliance with COVID-19 regulations were determined together with
changes in levels of life satisfaction and modal choice following the onset of
COVID-19. In addition, survey responses were used to identify anticipated
travel mode use in the future. Consideration was also given to the effects of
COVID-19 on transport related lifestyle issues such as working from home,
online shopping and the expectations of moving residences in the future. As
part of the analysis, statistical models were developed to provide an insight
into both the relationships between the levels of non compliance with COVID-19
regulations and demographic variables and the respondent attributes which might
affect future public transport usage. In general, the study confirmed
significant reductions in traffic activity, among respondents during the
COVID-19 pandemic associated with walking, driving a car and either using a bus
or train. The respondents also indicated that they anticipated they would
continue to make less use of buses and trains at the end of the pandemic.
| stat |
It's All Relative: New Regression Paradigm for Microbiome Compositional
Data | Microbiome data are complex in nature, involving high dimensionality,
compositionally, zero inflation, and taxonomic hierarchy. Compositional data
reside in a simplex that does not admit the standard Euclidean geometry. Most
existing compositional regression methods rely on transformations that are
inadequate or even inappropriate in modeling data with excessive zeros and
taxonomic structure. We develop a novel relative-shift regression framework
that directly uses compositions as predictors. The new framework provides a
paradigm shift for compositional regression and offers a superior biological
interpretation. New equi-sparsity and taxonomy-guided regularization methods
and an efficient smoothing proximal gradient algorithm are developed to
facilitate feature aggregation and dimension reduction in regression. As a
result, the framework can automatically identify clinically relevant microbes
even if they are important at different taxonomic levels. A unified
finite-sample prediction error bound is developed for the proposed regularized
estimators. We demonstrate the efficacy of the proposed methods in extensive
simulation studies. The application to a preterm infant study reveals novel
insights of association between the gut microbiome and neurodevelopment.
| stat |
Sequential scaled sparse factor regression | Large-scale association analysis between multivariate responses and
predictors is of great practical importance, as exemplified by modern business
applications including social media marketing and crisis management. Despite
the rapid methodological advances, how to obtain scalable estimators with free
tuning of the regularization parameters remains unclear under general noise
covariance structures. In this paper, we develop a new methodology called
sequential scaled sparse factor regression (SESS) based on a new viewpoint that
the problem of recovering a jointly low-rank and sparse regression coefficient
matrix can be decomposed into several univariate response sparse regressions
through regular eigenvalue decomposition. It combines the strengths of
sequential estimation and scaled sparse regression, thus sharing the
scalability and the tuning free property for sparsity parameters inherited from
the two approaches. The stepwise convex formulation, sequential factor
regression framework, and tuning insensitiveness make SESS highly scalable for
big data applications. Comprehensive theoretical justifications with new
insights into high-dimensional multi-response regressions are also provided. We
demonstrate the scalability and effectiveness of the proposed method by
simulation studies and stock short interest data analysis.
| stat |
Efficient Bayesian Inference of General Gaussian Models on Large
Phylogenetic Trees | Phylogenetic comparative methods correct for shared evolutionary history
among a set of non-independent organisms by modeling sample traits as arising
from a diffusion process along on the branches of a possibly unknown history.
To incorporate such uncertainty, we present a scalable Bayesian inference
framework under a general Gaussian trait evolution model that exploits
Hamiltonian Monte Carlo (HMC). HMC enables efficient sampling of the
constrained model parameters and takes advantage of the tree structure for fast
likelihood and gradient computations, yielding algorithmic complexity linear in
the number of observations. This approach encompasses a wide family of
stochastic processes, including the general Ornstein-Uhlenbeck (OU) process,
with possible missing data and measurement errors. We implement inference tools
for a biologically relevant subset of all these models into the BEAST
phylogenetic software package and develop model comparison through marginal
likelihood estimation. We apply our approach to study the morphological
evolution in the superfamilly of Musteloidea (including weasels and allies) as
well as the heritability of HIV virulence. This second problem furnishes a new
measure of evolutionary heritability that demonstrates its utility through a
targeted simulation study.
| stat |
Closed-form Expressions for Maximum Mean Discrepancy with Applications
to Wasserstein Auto-Encoders | The Maximum Mean Discrepancy (MMD) has found numerous applications in
statistics and machine learning, most recently as a penalty in the Wasserstein
Auto-Encoder (WAE). In this paper we compute closed-form expressions for
estimating the Gaussian kernel based MMD between a given distribution and the
standard multivariate normal distribution. This formula reveals a connection to
the Baringhaus-Henze-Epps-Pulley (BHEP) statistic of the Henze-Zirkler test and
provides further insights about the MMD. We introduce the standardized version
of MMD as a penalty for the WAE training objective, allowing for a better
interpretability of MMD values and more compatibility across different
hyperparameter settings. Next, we propose using a version of batch
normalization at the code layer; this has the benefits of making the kernel
width selection easier, reducing the training effort, and preventing outliers
in the aggregate code distribution. Our experiments on synthetic and real data
show that the analytic formulation improves over the commonly used stochastic
approximation of the MMD, and demonstrate that code normalization provides
significant benefits when training WAEs.
| stat |
Stein's Method Meets Statistics: A Review of Some Recent Developments | Stein's method is a collection of tools for analysing distributional
comparisons through the study of a class of linear operators called Stein
operators. Originally studied in probability, Stein's method has also enabled
some important developments in statistics. This early success has led to a high
research activity in this area in recent years. The goal of this survey is to
bring together some of these developments in theoretical statistics as well as
in computational statistics and, in doing so, to stimulate further research
into the successful field of Stein's method and statistics. The topics we
discuss include: explicit error bounds for asymptotic approximations of
estimators and test statistics, a measure of prior sensitivity in Bayesian
statistics, tools to benchmark and compare sampling methods such as approximate
Markov chain Monte Carlo, deterministic alternatives to sampling methods,
control variate techniques, and goodness-of-fit testing.
| stat |
Eco-Mobility-on-Demand Fleet Control with Ride-Sharing | Shared Mobility-on-Demand using automated vehicles can reduce energy
consumption and cost for future mobility. However, its full potential in energy
saving has not been fully explored. An algorithm to minimize fleet fuel
consumption while satisfying customers travel time constraints is developed in
this paper. Numerical simulations with realistic travel demand and route choice
are performed, showing that if fuel consumption is not considered, the MOD
service can increase fleet fuel consumption due to increased empty vehicle
mileage. With fuel consumption as part of the cost function, we can reduce
total fuel consumption by 7 percent while maintaining a high level of mobility
service.
| stat |
Matrix Completion for Survey Data Prediction with Multivariate
Missingness | The National Health and Nutrition Examination Survey (NHANES) studies the
nutritional and health status over the whole U.S. population with comprehensive
physical examinations and questionnaires. However, survey data analyses become
challenging due to inevitable missingness in almost all variables. In this
paper, we develop a new imputation method to deal with multivariate missingness
at random using matrix completion. In contrast to existing imputation schemes
either conducting row-wise or column-wise imputation, we treat the data matrix
as a whole which allows exploiting both row and column patterns to impute the
missing values in the whole data matrix at one time. We adopt a
column-space-decomposition model for the population data matrix with
easy-to-obtain demographic data as covariates and a low-rank structured
residual matrix. A unique challenge arises due to lack of identification of
parameters in the sample data matrix. We propose a projection strategy to
uniquely identify the parameters and corresponding penalized estimators, which
are computationally efficient and possess desired statistical properties. The
simulation study shows that the doubly robust estimator using the proposed
matrix completion for imputation has smaller mean squared error than other
competitors. To demonstrate practical relevance, we apply the proposed method
to the 2015-2016 NHANES Questionnaire Data.
| stat |
Anderson acceleration of coordinate descent | Acceleration of first order methods is mainly obtained via inertial
techniques \`a la Nesterov, or via nonlinear extrapolation. The latter has
known a recent surge of interest, with successful applications to gradient and
proximal gradient techniques. On multiple Machine Learning problems, coordinate
descent achieves performance significantly superior to full-gradient methods.
Speeding up coordinate descent in practice is not easy: inertially accelerated
versions of coordinate descent are theoretically accelerated, but might not
always lead to practical speed-ups. We propose an accelerated version of
coordinate descent using extrapolation, showing considerable speed up in
practice, compared to inertial accelerated coordinate descent and extrapolated
(proximal) gradient descent. Experiments on least squares, Lasso, elastic net
and logistic regression validate the approach.
| stat |
Inference of modes for linear stochastic processes | For dynamical systems that can be modelled as asymptotically stable linear
systems forced by Gaussian noise, this paper develops methods to infer or
estimate their modes from observations in real time. The modes can be real or
complex. For a real mode, we wish to infer its damping rate and mode shape. For
a complex mode, we wish to infer its frequency, damping rate and (complex) mode
shape. Their amplitudes and correlations are encoded in a mode covariance
matrix. The work is motivated and illustrated by the problem of detection of
oscillations in power flow in AC electrical networks. Suggestions of other
applications are given.
| stat |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.