title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Partial Identifiability in Discrete Data With Measurement Error | When data contains measurement errors, it is necessary to make assumptions
relating the observed, erroneous data to the unobserved true phenomena of
interest. These assumptions should be justifiable on substantive grounds, but
are often motivated by mathematical convenience, for the sake of exactly
identifying the target of inference. We adopt the view that it is preferable to
present bounds under justifiable assumptions than to pursue exact
identification under dubious ones. To that end, we demonstrate how a broad
class of modeling assumptions involving discrete variables, including common
measurement error and conditional independence assumptions, can be expressed as
linear constraints on the parameters of the model. We then use linear
programming techniques to produce sharp bounds for factual and counterfactual
distributions under measurement error in such models. We additionally propose a
procedure for obtaining outer bounds on non-linear models. Our method yields
sharp bounds in a number of important settings -- such as the instrumental
variable scenario with measurement error -- for which no bounds were previously
known.
| stat |
Spatio-Temporal Mixed Models to Predict Coverage Error Rates at Local
Areas | Despite of the great efforts during the censuses, occurrence of some
nonsampling errors such as coverage error is inevitable. Coverage error which
can be classified into two types of under-count and overcount occurs when there
is no unique bijective (one-to-one) mapping between the individuals from the
census count and the target population -- individuals who usually reside in the
country (de jure residences). There are variety of reasons make the coverage
error happens including deficiencies in the census maps, errors in the field
operations or disinclination of people for participation in the undercount
situation and multiple enumeration of individuals or those who do not belong to
the scope of the census in the overcount situation. A routine practice for
estimating the net coverage error is subtracting the census count from the
estimated true population, which obtained from a dual system (or
capture-recapture) technique. Estimated coverage error usually suffers from
significant uncertainty of the direct estimate of true population or other
errors such as matching error. To rectify the above-mentioned problem and
predict a more reliable coverage error rate, we propose a set of
spatio-temporal mixed models. In an illustrative study on the 2010 census
coverage error rate of the U.S. counties with population more than 100,000, we
select the best mixed model for prediction by deviance information criteria
(DIC) and conditional predictive ordinate (CPO). Our proposed approach for
predicting coverage error rate and its measure of uncertainty is a full
Bayesian approach, which leads to a reasonable improvement over the direct
coverage error rate in terms of mean squared error (MSE) and confidence
interval (CI) as provided by the U.S. Census Bureau.
| stat |
Factor Analysis on Citation, Using a Combined Latent and Logistic
Regression Model | We propose a combined model, which integrates the latent factor model and the
logistic regression model, for the citation network. It is noticed that neither
a latent factor model nor a logistic regression model alone is sufficient to
capture the structure of the data. The proposed model has a latent (i.e.,
factor analysis) model to represents the main technological trends (a.k.a.,
factors), and adds a sparse component that captures the remaining ad-hoc
dependence. Parameter estimation is carried out through the construction of a
joint-likelihood function of edges and properly chosen penalty terms. The
convexity of the objective function allows us to develop an efficient
algorithm, while the penalty terms push towards a low-dimensional latent
component and a sparse graphical structure. Simulation results show that the
proposed method works well in practical situations. The proposed method has
been applied to a real application, which contains a citation network of
statisticians (Ji and Jin, 2016). Some interesting findings are reported.
| stat |
A Bayesian approach to regional decadal predictability: Sparse parameter
estimation in high-dimensional linear inverse models of high-latitude sea
surface temperature variability | Stochastic reduced models are an important tool in climate systems whose many
spatial and temporal scales cannot be fully discretized or underlying physics
may not be fully accounted for. One form of reduced model, the linear inverse
model (LIM), has been widely used for regional climate predictability studies -
typically focusing more on tropical or mid-latitude studies. However, most LIM
fitting techniques rely on point estimation techniques deriving from
fluctuation-dissipation theory. In this methodological study we explore the use
of Bayesian inference techniques for LIM parameter estimation of sea surface
temperature (SST), to quantify the skillful decadal predictability of Bayesian
LIM models at high latitudes. We show that Bayesian methods, when compared to
traditional point estimation methods for LIM-type models, provide better
calibrated probabilistic skill, while simultaneously providing better point
estimates due to the regularization effect of the prior distribution in
high-dimensional problems. We compare the effect of several priors, as well as
maximum likelihood estimates, on (1) estimating parameter values on a perfect
model experiment and (2) producing calibrated 1-year SST anomaly forecast
distributions using a pre-industrial control run of the Community Earth System
Model (CESM). Finally, we employ a host of probabilistic skill metrics to
determine the extent to which a LIM can forecast SST anomalies at high
latitudes. We find that the choice of prior distribution has an appreciable
impact on estimation outcomes, and priors that emphasize physically relevant
properties enhance the model's ability to capture variability of SST anomalies.
| stat |
Causal Mediation Analysis with Multiple Treatments and Latent
Confounders | Causal mediation analysis is used to evaluate direct and indirect causal
effects of a treatment on an outcome of interest through an intermediate
variable or a mediator.It is difficult to identify the direct and indirect
causal effects because the mediator cannot be randomly assigned in many real
applications. In this article, we consider a causal model including latent
confounders between the mediator and the outcome. We present sufficient
conditions for identifying the direct and indirect effects and propose an
approach for estimating them. The performance of the proposed approach is
evaluated by simulation studies. Finally, we apply the approach to a data set
of the customer loyalty survey by a telecom company.
| stat |
A COVINDEX based on a GAM beta regression model with an application to
the COVID-19 pandemic in Italy | Detecting changes in COVID-19 disease transmission over time is a key
indicator of epidemic growth. Near real-time monitoring of the pandemic growth
is crucial for policy makers and public health officials who need to make
informed decisions about whether to enforce lockdowns or allow certain
activities. The effective reproduction number Rt is the standard index used in
many countries for this goal. However, it is known that due to the delays
between infection and case registration, its use for decision making is
somewhat limited. In this paper a near real-time COVINDEX is proposed for
monitoring the evolution of the pandemic. The index is computed from
predictions obtained from a GAM beta regression for modelling the test positive
rate as a function of time. The proposal is illustrated using data on COVID-19
pandemic in Italy and compared with Rt. A simple chart is also proposed for
monitoring local and national outbreaks by policy makers and public health
officials.
| stat |
A Selective Review of Negative Control Methods in Epidemiology | Purpose of Review: Negative controls are a powerful tool to detect and adjust
for bias in epidemiological research. This paper introduces negative controls
to a broader audience and provides guidance on principled design and causal
analysis based on a formal negative control framework.
Recent Findings: We review and summarize causal and statistical assumptions,
practical strategies, and validation criteria that can be combined with subject
matter knowledge to perform negative control analyses. We also review existing
statistical methodologies for detection, reduction, and correction of
confounding bias, and briefly discuss recent advances towards nonparametric
identification of causal effects in a double negative control design.
Summary: There is great potential for valid and accurate causal inference
leveraging contemporary healthcare data in which negative controls are
routinely available. Design and analysis of observational data leveraging
negative controls is an area of growing interest in health and social sciences.
Despite these developments, further effort is needed to disseminate these novel
methods to ensure they are adopted by practicing epidemiologists.
| stat |
Classifier uncertainty: evidence, potential impact, and probabilistic
treatment | Classifiers are often tested on relatively small data sets, which should lead
to uncertain performance metrics. Nevertheless, these metrics are usually taken
at face value. We present an approach to quantify the uncertainty of
classification performance metrics, based on a probability model of the
confusion matrix. Application of our approach to classifiers from the
scientific literature and a classification competition shows that uncertainties
can be surprisingly large and limit performance evaluation. In fact, some
published classifiers are likely to be misleading. The application of our
approach is simple and requires only the confusion matrix. It is agnostic of
the underlying classifier. Our method can also be used for the estimation of
sample sizes that achieve a desired precision of a performance metric.
| stat |
Topic Diffusion Discovery Based on Deep Non-negative Autoencoder | Researchers have been overwhelmed by the explosion of research articles
published by various research communities. Many research scholarly websites,
search engines, and digital libraries have been created to help researchers
identify potential research topics and keep up with recent progress on research
of interests. However, it is still difficult for researchers to keep track of
the research topic diffusion and evolution without spending a large amount of
time reviewing numerous relevant and irrelevant articles. In this paper, we
consider a novel topic diffusion discovery technique. Specifically, we propose
using a Deep Non-negative Autoencoder with information divergence measurement
that monitors evolutionary distance of the topic diffusion to understand how
research topics change with time. The experimental results show that the
proposed approach is able to identify the evolution of research topics as well
as to discover topic diffusions in online fashions.
| stat |
Instrumental variables, spatial confounding and interference | Unobserved spatial confounding variables are prevalent in environmental and
ecological applications where the system under study is complex and the data
are often observational. Instrumental variables (IVs) are a common way to
address unobserved confounding; however, the efficacy of using IVs on spatial
confounding is largely unknown. This paper explores the effectiveness of IVs in
this situation -- with particular attention paid to the spatial scale of the
instrument. We show that, in case of spatially-dependent treatments, IVs are
most effective when they vary at a finer spatial resolution than the treatment.
We investigate IV performance in extensive simulations and apply the model in
the example of long term trends in the air pollution and cardiovascular
mortality in the United States over 1990-2010. Finally, the IV approach is also
extended to the spatial interference setting, in which treatments can affect
nearby responses.
| stat |
Nonparametric Bayes Differential Analysis for Dependent Multigroup Data
with Application to DNA Methylation Analyses in Cancer | Modern cancer genomics datasets involve widely varying sizes and scales,
measurement variables, and correlation structures. A fundamental analytical
goal in these high-throughput studies is the development of general statistical
techniques that can cleanly sift the signal from noise in identifying
disease-specific genomic signatures across a set of experimental or biological
conditions. We propose BayesDiff, a nonparametric Bayesian approach based on a
novel class of first order mixture models, called the Sticky Poisson-Dirichlet
process or multicuisine restaurant franchise. The BayesDiff methodology
flexibly utilizes information from all the measurements and adaptively
accommodates any serial dependence in the data, accounting for the inter-probe
distances, to perform simultaneous inferences on the variables. The technique
is applied to analyze a DNA methylation gastrointestinal (GI) cancer dataset,
which displays both serial correlations and complex interaction patterns. Our
analyses and results both support and complement known aspects of DNA
methylation and gene association in upper GI cancers. In simulation studies, we
demonstrate the effectiveness of the BayesDiff procedure relative to existing
techniques for differential DNA methylation.
| stat |
The Kernel Interaction Trick: Fast Bayesian Discovery of Pairwise
Interactions in High Dimensions | Discovering interaction effects on a response of interest is a fundamental
problem faced in biology, medicine, economics, and many other scientific
disciplines. In theory, Bayesian methods for discovering pairwise interactions
enjoy many benefits such as coherent uncertainty quantification, the ability to
incorporate background knowledge, and desirable shrinkage properties. In
practice, however, Bayesian methods are often computationally intractable for
even moderate-dimensional problems. Our key insight is that many hierarchical
models of practical interest admit a particular Gaussian process (GP)
representation; the GP allows us to capture the posterior with a vector of O(p)
kernel hyper-parameters rather than O(p^2) interactions and main effects. With
the implicit representation, we can run Markov chain Monte Carlo (MCMC) over
model hyper-parameters in time and memory linear in p per iteration. We focus
on sparsity-inducing models and show on datasets with a variety of covariate
behaviors that our method: (1) reduces runtime by orders of magnitude over
naive applications of MCMC, (2) provides lower Type I and Type II error
relative to state-of-the-art LASSO-based approaches, and (3) offers improved
computational scaling in high dimensions relative to existing Bayesian and
LASSO-based approaches.
| stat |
Spurious Vanishing Problem in Approximate Vanishing Ideal | Approximate vanishing ideal is a concept from computer algebra that studies
the algebraic varieties behind perturbed data points. To capture the nonlinear
structure of perturbed points, the introduction of approximation to exact
vanishing ideals plays a critical role. However, such an approximation also
gives rise to a theoretical problem---the spurious vanishing problem---in the
basis construction of approximate vanishing ideals; namely, obtained basis
polynomials can be approximately vanishing simply because of the small
coefficients. In this paper, we propose a first general method that enables
various basis construction algorithms to overcome the spurious vanishing
problem. In particular, we integrate coefficient normalization with
polynomial-based basis constructions, which do not need the proper ordering of
monomials to process for basis constructions. We further propose a method that
takes advantage of the iterative nature of basis construction so that
computationally costly operations for coefficient normalization can be
circumvented. Moreover, a coefficient truncation method is proposed for further
accelerations. From the experiments, it can be shown that the proposed method
overcomes the spurious vanishing problem, resulting in shorter feature vectors
while sustaining comparable or even lower classification error.
| stat |
Stop Explaining Black Box Machine Learning Models for High Stakes
Decisions and Use Interpretable Models Instead | Black box machine learning models are currently being used for high stakes
decision-making throughout society, causing problems throughout healthcare,
criminal justice, and in other domains. People have hoped that creating methods
for explaining these black box models will alleviate some of these problems,
but trying to \textit{explain} black box models, rather than creating models
that are \textit{interpretable} in the first place, is likely to perpetuate bad
practices and can potentially cause catastrophic harm to society. There is a
way forward -- it is to design models that are inherently interpretable. This
manuscript clarifies the chasm between explaining black boxes and using
inherently interpretable models, outlines several key reasons why explainable
black boxes should be avoided in high-stakes decisions, identifies challenges
to interpretable machine learning, and provides several example applications
where interpretable models could potentially replace black box models in
criminal justice, healthcare, and computer vision.
| stat |
Bayesian matrix completion with a spectral scaled Student prior:
theoretical guarantee and efficient sampling | We study the problem of matrix completion in this paper. A spectral scaled
Student prior is exploited to favour the underlying low-rank structure of the
data matrix. Importantly, we provide a thorough theoretical investigation for
our approach, while such an analysis is hard to obtain and limited in
theoretical understanding of Bayesian matrix completion. More precisely, we
show that our Bayesian approach enjoys a minimax-optimal oracle inequality
which guarantees that our method works well under model misspecification and
under general sampling distribution. Interestingly, we also provide efficient
gradient-based sampling implementations for our approach by using Langevin
Monte Carlo which is novel in Bayesian matrix completion. More specifically, we
show that our algorithms are significantly faster than Gibbs sampler in this
problem. To illustrate the attractive features of our inference strategy, some
numerical simulations are conducted and an application to image inpainting is
demonstrated.
| stat |
Depth Uncertainty in Neural Networks | Existing methods for estimating uncertainty in deep learning tend to require
multiple forward passes, making them unsuitable for applications where
computational resources are limited. To solve this, we perform probabilistic
reasoning over the depth of neural networks. Different depths correspond to
subnetworks which share weights and whose predictions are combined via
marginalisation, yielding model uncertainty. By exploiting the sequential
structure of feed-forward networks, we are able to both evaluate our training
objective and make predictions with a single forward pass. We validate our
approach on real-world regression and image classification tasks. Our approach
provides uncertainty calibration, robustness to dataset shift, and accuracies
competitive with more computationally expensive baselines.
| stat |
Model-based Clustering using Automatic Differentiation: Confronting
Misspecification and High-Dimensional Data | We study two practically important cases of model based clustering using
Gaussian Mixture Models: (1) when there is misspecification and (2) on high
dimensional data, in the light of recent advances in Gradient Descent (GD)
based optimization using Automatic Differentiation (AD). Our simulation studies
show that EM has better clustering performance, measured by Adjusted Rand
Index, compared to GD in cases of misspecification, whereas on high dimensional
data GD outperforms EM. We observe that both with EM and GD there are many
solutions with high likelihood but poor cluster interpretation. To address this
problem we design a new penalty term for the likelihood based on the Kullback
Leibler divergence between pairs of fitted components. Closed form expressions
for the gradients of this penalized likelihood are difficult to derive but AD
can be done effortlessly, illustrating the advantage of AD-based optimization.
Extensions of this penalty for high dimensional data and for model selection
are discussed. Numerical experiments on synthetic and real datasets demonstrate
the efficacy of clustering using the proposed penalized likelihood approach.
| stat |
The Unusual Effectiveness of Averaging in GAN Training | We examine two different techniques for parameter averaging in GAN training.
Moving Average (MA) computes the time-average of parameters, whereas
Exponential Moving Average (EMA) computes an exponentially discounted sum.
Whilst MA is known to lead to convergence in bilinear settings, we provide the
-- to our knowledge -- first theoretical arguments in support of EMA. We show
that EMA converges to limit cycles around the equilibrium with vanishing
amplitude as the discount parameter approaches one for simple bilinear games
and also enhances the stability of general GAN training. We establish
experimentally that both techniques are strikingly effective in the
non-convex-concave GAN setting as well. Both improve inception and FID scores
on different architectures and for different GAN objectives. We provide
comprehensive experimental results across a range of datasets -- mixture of
Gaussians, CIFAR-10, STL-10, CelebA and ImageNet -- to demonstrate its
effectiveness. We achieve state-of-the-art results on CIFAR-10 and produce
clean CelebA face images.\footnote{~The code is available at
\url{https://github.com/yasinyazici/EMA_GAN}}
| stat |
User-Dependent Neural Sequence Models for Continuous-Time Event Data | Continuous-time event data are common in applications such as individual
behavior data, financial transactions, and medical health records. Modeling
such data can be very challenging, in particular for applications with many
different types of events, since it requires a model to predict the event types
as well as the time of occurrence. Recurrent neural networks that parameterize
time-varying intensity functions are the current state-of-the-art for
predictive modeling with such data. These models typically assume that all
event sequences come from the same data distribution. However, in many
applications event sequences are generated by different sources, or users, and
their characteristics can be very different. In this paper, we extend the broad
class of neural marked point process models to mixtures of latent embeddings,
where each mixture component models the characteristic traits of a given user.
Our approach relies on augmenting these models with a latent variable that
encodes user characteristics, represented by a mixture model over user behavior
that is trained via amortized variational inference. We evaluate our methods on
four large real-world datasets and demonstrate systematic improvements from our
approach over existing work for a variety of predictive metrics such as
log-likelihood, next event ranking, and source-of-sequence identification.
| stat |
DINA: Estimating Heterogenous Treatment Effects in Exponential Family
and Cox Model | We propose to use the difference in natural parameters (DINA) to quantify the
heterogeneous treatment effect for the exponential family, a.k.a. the hazard
ratio for the Cox model, in contrast to the difference in means. For responses
such as binary outcome and survival time, DINA is of more practical interest
and convenient for modeling the covariates' influences on the treatment effect.
We introduce a DINA estimator that is insensitive to confounding and
non-collapsibility issues, and allows practitioners to use powerful
off-the-shelf machine learning tools for nuisance estimation. We use extensive
simulations to demonstrate the efficacy of the proposed method with various
response distributions and censoring mechanisms. We also apply the proposed
method to the SPRINT dataset to estimate the heterogeneous treatment effect,
testify the method's robustness to nuisance estimation, and conduct placebo
evaluation.
| stat |
Statistical inference for Axiom A attractors | From the climate system to the effect of the internet on society, chaotic
systems appear to have a significant role in our future. Here a method of
statistical learning for a class of chaotic systems is described along with
underlying theory that can be used not only for predicting those systems a
short time ahead, but also as a basis for statistical inference about their
dynamics. The method is applied to prediction of 3 different systems. The
statistical inference aspect can be applied to explore and enhance computer
models of such systems which in turn can provide feedback for even better
prediction and more precise inference.
| stat |
Ultimate P\'olya Gamma Samplers -- Efficient MCMC for possibly
imbalanced binary and categorical data | Modeling binary and categorical data is one of the most commonly encountered
tasks of applied statisticians and econometricians. While Bayesian methods in
this context have been available for decades now, they often require a high
level of familiarity with Bayesian statistics or suffer from issues such as low
sampling efficiency. To contribute to the accessibility of Bayesian models for
binary and categorical data, we introduce novel latent variable representations
based on P\'olya Gamma random variables for a range of commonly encountered
discrete choice models. From these latent variable representations, new Gibbs
sampling algorithms for binary, binomial and multinomial logistic regression
models are derived. All models allow for a conditionally Gaussian likelihood
representation, rendering extensions to more complex modeling frameworks such
as state space models straight-forward. However, sampling efficiency may still
be an issue in these data augmentation based estimation frameworks. To
counteract this, MCMC boosting strategies are developed and discussed in
detail. The merits of our approach are illustrated through extensive
simulations and a real data application.
| stat |
Estimating the Efficiency Gain of Covariate-Adjusted Analyses in Future
Clinical Trials Using External Data | We present a general framework for using existing data to estimate the
efficiency gain from using a covariate-adjusted estimator of a marginal
treatment effect in a future randomized trial. We describe conditions under
which it is possible to define a mapping from the distribution that generated
the existing external data to the relative efficiency of a covariate-adjusted
estimator compared to an unadjusted estimator. Under conditions, these relative
efficiencies approximate the ratio of sample size needed to achieve a desired
power. We consider two situations where the outcome is either fully or
partially observed and several treatment effect estimands that are of
particular interest in most trials. For each such estimand, we develop a
semiparametrically efficient estimator of the relative efficiency that allows
for the application of flexible statistical learning tools to estimate the
nuisance functions and an analytic form of a corresponding Wald-type confidence
interval. We also propose a double bootstrap scheme for constructing confidence
intervals. We demonstrate the performance of the proposed methods through
simulation studies and apply these methods to data to estimate the relative
efficiency of using covariate adjustment in Covid-19 therapeutic trials.
| stat |
Learning Overlapping Representations for the Estimation of
Individualized Treatment Effects | The choice of making an intervention depends on its potential benefit or harm
in comparison to alternatives. Estimating the likely outcome of alternatives
from observational data is a challenging problem as all outcomes are never
observed, and selection bias precludes the direct comparison of differently
intervened groups. Despite their empirical success, we show that algorithms
that learn domain-invariant representations of inputs (on which to make
predictions) are often inappropriate, and develop generalization bounds that
demonstrate the dependence on domain overlap and highlight the need for
invertible latent maps. Based on these results, we develop a deep kernel
regression algorithm and posterior regularization framework that substantially
outperforms the state-of-the-art on a variety of benchmarks data sets.
| stat |
Can we disregard the whole model? Omnibus non-inferiority testing for
$R^{2}$ in multivariable linear regression and $\hat{\eta}^{2}$ in ANOVA | Determining a lack of association between an outcome variable and a number of
different explanatory variables is frequently necessary in order to disregard a
proposed model (i.e., to confirm the lack of an association between an outcome
and predictors). Despite this, the literature rarely offers information about,
or technical recommendations concerning, the appropriate statistical
methodology to be used to accomplish this task. This paper introduces
non-inferiority tests for ANOVA and linear regression analyses, that correspond
to the standard widely used $F$-test for $\hat{\eta}^2$ and $R^{2}$,
respectively. A simulation study is conducted to examine the type I error rates
and statistical power of the tests, and a comparison is made with an
alternative Bayesian testing approach. The results indicate that the proposed
non-inferiority test is a potentially useful tool for 'testing the null.'
| stat |
Designing group sequential clinical trials when a delayed effect is
anticipated: A practical guidance | A common feature of many recent trials evaluating the effects of
immunotherapy on survival is that non-proportional hazards can be anticipated
at the design stage. This raises the possibility to use a statistical method
tailored towards testing the purported long-term benefit, rather than applying
the more standard log-rank test and/or Cox model. Many such proposals have been
made in recent years, but there remains a lack of practical guidance on
implementation, particularly in the context of group-sequential designs. In
this article, we aim to fill this gap. We discuss how the POPLAR trial, which
compared immunotherapy versus chemotherapy in non-small-cell lung cancer, might
have been re-designed to be more robust to the presence of a delayed effect. We
then provide step-by-step instructions on how to analyse a hypothetical
realisation of the trial, based on this new design. Basic theory on weighted
log-rank tests and group-sequential methods is covered, and an accompanying R
package (including vignette) is provided.
| stat |
Meta-Learning Deep Energy-Based Memory Models | We study the problem of learning associative memory -- a system which is able
to retrieve a remembered pattern based on its distorted or incomplete version.
Attractor networks provide a sound model of associative memory: patterns are
stored as attractors of the network dynamics and associative retrieval is
performed by running the dynamics starting from a query pattern until it
converges to an attractor. In such models the dynamics are often implemented as
an optimization procedure that minimizes an energy function, such as in the
classical Hopfield network. In general it is difficult to derive a writing rule
for a given dynamics and energy that is both compressive and fast. Thus, most
research in energy-based memory has been limited either to tractable energy
models not expressive enough to handle complex high-dimensional objects such as
natural images, or to models that do not offer fast writing. We present a novel
meta-learning approach to energy-based memory models (EBMM) that allows one to
use an arbitrary neural architecture as an energy model and quickly store
patterns in its weights. We demonstrate experimentally that our EBMM approach
can build compressed memories for synthetic and natural data, and is capable of
associative retrieval that outperforms existing memory systems in terms of the
reconstruction error and compression rate.
| stat |
varstan: An R package for Bayesian analysis of structured time series
models with Stan | varstan is an \proglang{R} package for Bayesian analysis of time series
models using \proglang{Stan}. The package offers a dynamic way to choose a
model, define priors in a wide range of distributions, check model's fit, and
forecast with the m-steps ahead predictive distribution. The users can widely
choose between implemented models such as \textit{multiplicative seasonal
ARIMA, dynamic regression, random walks, GARCH, dynamic harmonic
regressions,VARMA, stochastic Volatility Models, and generalized t-student with
unknown degree freedom GARCH models}. Every model constructor in \pkg{varstan}
defines weakly informative priors, but prior specifications can be changed in a
dynamic and flexible way, so the prior distributions reflect the parameter's
initial beliefs. For model selection, the package offers the classical
information criteria: AIC, AICc, BIC, DIC, Bayes factor. And more recent
criteria such as Widely-applicable information criteria (\textit{WAIC}), and
the Bayesian leave one out cross-validation (\textit{loo}). In addition, a
Bayesian version for automatic order selection in seasonal ARIMA and dynamic
regression models can be used as an initial step for the time series analysis.
| stat |
TRAP: A Predictive Framework for Trail Running Assessment of Performance | Trail running is an endurance sport in which athletes face severe physical
challenges. Due to the growing number of participants, the organization of
limited staff, equipment, and medical support in these races now plays a key
role. Monitoring runner's performance is a difficult task that requires
knowledge of the terrain and of the runner's ability. In the past, choices were
solely based on the organizers' experience without reliance on data. However,
this approach is neither scalable nor transferable. Instead, we propose a firm
statistical methodology to perform this task, both before and during the race.
Our proposed framework, Trail Running Assessment of Performance (TRAP), studies
(1) the the assessment of the runner's ability to reach the next checkpoint,
(2) the prediction of the runner's expected passage time at the next
checkpoint, and (3) corresponding prediction intervals for the passage time. To
obtain data on the ability of runners, we introduce a Python package,
ScrapITRA, to access the race history of runners from the International Trail
Running Association (ITRA). We apply our methodology, using the ITRA data along
with checkpoint and terrain-level information, to the "holy grail" of
ultra-trail running, the Ultra-Trail du Mont-Blanc (UTMB) race, demonstrating
the predictive power of our methodology.
| stat |
Bayesian Evidential Deep Learning with PAC Regularization | We propose a novel method for closed-form predictive distribution modeling
with neural nets. In quantifying prediction uncertainty, we build on Evidential
Deep Learning, which has been impactful as being both simple to implement and
giving closed-form access to predictive uncertainty. We employ it to model
aleatoric uncertainty and extend it to account also for epistemic uncertainty
by converting it to a Bayesian Neural Net. While extending its uncertainty
quantification capabilities, we maintain its analytically accessible predictive
distribution model by performing progressive moment matching for the first time
for approximate weight marginalization. The eventual model introduces a
prohibitively large number of hyperparameters for stable training. We overcome
this drawback by deriving a vacuous PAC bound that comprises the marginal
likelihood of the predictor and a complexity penalty. We observe on regression,
classification, and out-of-domain detection benchmarks that our method improves
model fit and uncertainty quantification.
| stat |
Implicit Hamiltonian Monte Carlo for Sampling Multiscale Distributions | Hamiltonian Monte Carlo (HMC) has been widely adopted in the statistics
community because of its ability to sample high-dimensional distributions much
more efficiently than other Metropolis-based methods. Despite this, HMC often
performs sub-optimally on distributions with high correlations or marginal
variances on multiple scales because the resulting stiffness forces the
leapfrog integrator in HMC to take an unreasonably small stepsize. We provide
intuition as well as a formal analysis showing how these multiscale
distributions limit the stepsize of leapfrog and we show how the implicit
midpoint method can be used, together with Newton-Krylov iteration, to
circumvent this limitation and achieve major efficiency gains. Furthermore, we
offer practical guidelines for when to choose between implicit midpoint and
leapfrog and what stepsize to use for each method, depending on the
distribution being sampled. Unlike previous modifications to HMC, our method is
generally applicable to highly non-Gaussian distributions exhibiting multiple
scales. We illustrate how our method can provide a dramatic speedup over
leapfrog in the context of the No-U-Turn sampler (NUTS) applied to several
examples.
| stat |
How much baseline correction do we need in ERP research? Extended GLM
model can replace baseline correction while lifting its limits | Baseline correction plays an important role in past and current
methodological debates in ERP research (e.g. the Tanner v. Maess debate in
Journal of Neuroscience Methods), serving as a potential alternative to strong
highpass filtering. However, the very assumptions that underlie traditional
baseline also undermine it, making it statistically unnecessary and even
undesirable and reducing signal-to-noise ratio. Including the baseline interval
as a predictor in a GLM-based statistical approach allows the data to determine
how much baseline correction is needed, including both full traditional and no
baseline correction as subcases, while reducing the amount of variance in the
residual error term and thus potentially increasing statistical power.
| stat |
Functional Peaks-over-threshold Analysis | Peaks-over-threshold analysis using the generalized Pareto distribution is
widely applied in modelling tails of univariate random variables, but much
information may be lost when complex extreme events are studied using
univariate results. In this paper, we extend peaks-over-threshold analysis to
extremes of functional data. Threshold exceedances defined using a functional
$r$ are modelled by the generalized $r$-Pareto process, a functional
generalization of the generalized Pareto distribution that covers the three
classical regimes for the decay of tail probabilities, and that is the only
possible continuous limit for $r$-exceedances of a properly rescaled process.
We give construction rules, simulation algorithms and inference procedures for
generalized $r$-Pareto processes, discuss model validation, and use the new
methodology to study extreme European windstorms and heavy spatial rainfall.
| stat |
Bridging Parametric and Nonparametric Methods in Cognitive Diagnosis | A number of parametric and nonparametric methods for estimating cognitive
diagnosis models (CDMs) have been developed and applied in a wide range of
contexts. However, in the literature, a wide chasm exists between these two
families of methods, and their relationship to each other is not well
understood. In this paper, we propose a unified estimation framework to bridge
the divide between parametric and nonparametric methods in cognitive diagnosis
to better understand their relationship. We also develop iterative joint
estimation algorithms and establish consistency properties within the proposed
framework. Lastly, we present comprehensive simulation results to compare
different methods, and provide practical recommendations on the appropriate use
of the proposed framework in various CDM contexts.
| stat |
Context-specific volume-delay curves by combining crowd-sourced traffic
data with Automated Traffic Counters (ATC): a case study for London | Traffic congestion across the world has reached chronic levels. Despite many
technological disruptions, one of the most fundamental and widely used
functions within traffic modelling, the volume delay function, has seen little
in the way of change since it was developed in the 1960's. Traditionally
macroscopic methods have been employed to relate traffic volume to vehicular
journey time. The general nature of these functions enables their ease of use
and gives widespread applicability. However, they lack the ability to consider
individual road characteristics (i.e. geometry, presence of traffic furniture,
road quality and surrounding environment). This research investigates the
feasibility to reconstruct the model using two different data sources, namely
the traffic speed from Google Maps' Directions Application Programming
Interface (API) and traffic volume data from automated traffic counters (ATC).
Google's traffic speed data are crowd-sourced from the smartphone Global
Positioning System (GPS) of road users, able to reflect real-time,
context-specific traffic condition of a road. On the other hand, the ATCs
enable the harvesting of the vehicle volume data over equally fine temporal
resolutions (hourly or less). By combining them for different road types in
London, new context-specific volume-delay functions can be generated. This
method shows promise in selected locations with the generation of robust
functions. In other locations it highlights the need to better understand other
influencing factors, such as the presence of on road parking or weather events.
| stat |
Efficient Designs of SLOPE Penalty Sequences in Finite Dimension | In linear regression, SLOPE is a new convex analysis method that generalizes
the Lasso via the sorted L1 penalty: larger fitted coefficients are penalized
more heavily. This magnitude-dependent regularization requires an input of
penalty sequence $\lambda$, instead of a scalar penalty as in the Lasso case,
thus making the design extremely expensive in computation. In this paper, we
propose two efficient algorithms to design the possibly high-dimensional SLOPE
penalty, in order to minimize the mean squared error. For Gaussian data
matrices, we propose a first order Projected Gradient Descent (PGD) under the
Approximate Message Passing regime. For general data matrices, we present a
zero-th order Coordinate Descent (CD) to design a sub-class of SLOPE, referred
to as the k-level SLOPE. Our CD allows a useful trade-off between the accuracy
and the computation speed. We demonstrate the performance of SLOPE with our
designs via extensive experiments on synthetic data and real-world datasets.
| stat |
Preparing Weather Data for Real-Time Building Energy Simulation | This study introduces a framework for quality control of measured weather
data, including anomaly detection, and infilling missing values. Weather data
is a fundamental input to building performance simulations, in which anomalous
values defect the results while missing data lead to an unexpected termination
of the simulation process. Traditionally, infilling missing values in weather
data is performed through periodic or linear interpolations. However, when
missing values exceed many consecutive hours, the accuracy of traditional
methods is subject to debate. This study demonstrates how Neural Networks can
increase the accuracy of data imputation when compared to other supervised
learning methods. The framework is validated by predicting missing temperature
and relative humidity data for an observation site, through a network of nearby
weather stations in Milan, Italy. Results show that the proposed method can
facilitate real-time building simulations with accurate and rapid quality
control.
| stat |
Forward Stability and Model Path Selection | Most scientific publications follow the familiar recipe of (i) obtain data,
(ii) fit a model, and (iii) comment on the scientific relevance of the effects
of particular covariates in that model. This approach, however, ignores the
fact that there may exist a multitude of similarly-accurate models in which the
implied effects of individual covariates may be vastly different. This problem
of finding an entire collection of plausible models has also received
relatively little attention in the statistics community, with nearly all of the
proposed methodologies being narrowly tailored to a particular model class
and/or requiring an exhaustive search over all possible models, making them
largely infeasible in the current big data era. This work develops the idea of
forward stability and proposes a novel, computationally-efficient approach to
finding collections of accurate models we refer to as model path selection
(MPS). MPS builds up a plausible model collection via a forward selection
approach and is entirely agnostic to the model class and loss function
employed. The resulting model collection can be displayed in a simple and
intuitive graphical fashion, easily allowing practitioners to visualize whether
some covariates can be swapped for others with minimal loss.
| stat |
Integrated causal-predictive machine learning models for tropical
cyclone epidemiology | Strategic preparedness has been shown to reduce the adverse health impacts of
hurricanes and tropical storms, referred to collectively as tropical cyclones
(TCs), but its protective impact could be enhanced by a more comprehensive and
rigorous characterization of TC epidemiology. To generate the insights and
tools necessary for high-precision TC preparedness, we develop and apply a
novel Bayesian machine learning approach that standardizes estimation of
historic TC health impacts, discovers common patterns and sources of
heterogeneity in those health impacts, and enables identification of
communities at highest health risk for future TCs. The model integrates (1) a
causal inference component to quantify the immediate health impacts of recent
historic TCs at high spatial resolution and (2) a predictive component that
captures how TC meteorological features and socioeconomic/demographic
characteristics of impacted communities are associated with health impacts. We
apply it to a rich data platform containing detailed historic TC exposure
information and Medicare claims data. The health outcomes used in our analyses
are all-cause mortality and cardiovascular- and respiratory-related
hospitalizations. We report a high degree of heterogeneity in the acute health
impacts of historic TCs at both the TC level and the community level, with
substantial increases in respiratory hospitalizations, on average, during a
two-week period surrounding TCs. TC sustained windspeeds are found to be the
primary driver of increased mortality and respiratory risk. Our modeling
approach has broader utility for predicting the health impacts of many types of
extreme climate events.
| stat |
Large Scale Tensor Regression using Kernels and Variational Inference | We outline an inherent weakness of tensor factorization models when latent
factors are expressed as a function of side information and propose a novel
method to mitigate this weakness. We coin our method \textit{Kernel Fried
Tensor}(KFT) and present it as a large scale forecasting tool for high
dimensional data. Our results show superior performance against
\textit{LightGBM} and \textit{Field Aware Factorization Machines}(FFM), two
algorithms with proven track records widely used in industrial forecasting. We
also develop a variational inference framework for KFT and associate our
forecasts with calibrated uncertainty estimates on three large scale datasets.
Furthermore, KFT is empirically shown to be robust against uninformative side
information in terms of constants and Gaussian noise.
| stat |
A Spatially Discrete Approximation to Log-Gaussian Cox Processes for
Modelling Aggregated Disease Count Data | In this paper, we develop a computationally efficient discrete approximation
to log-Gaussian Cox process (LGCP) models for the analysis of spatially
aggregated disease count data. Our approach overcomes an inherent limitation of
spatial models based on Markov structures, namely that each such model is tied
to a specific partition of the study area, and allows for spatially continuous
prediction. We compare the predictive performance of our modelling approach
with LGCP through a simulation study and an application to primary biliary
cirrhosis incidence data in Newcastle-Upon-Tyne, UK. Our results suggest that
when disease risk is assumed to be a spatially continuous process, the proposed
approximation to LGCP provides reliable estimates of disease risk both on
spatially continuous and aggregated scales. The proposed methodology is
implemented in the open-source R package SDALGCP.
| stat |
Bayesian Neural Networks at Finite Temperature | We recapitulate the Bayesian formulation of neural network based classifiers
and show that, while sampling from the posterior does indeed lead to better
generalisation than is obtained by standard optimisation of the cost function,
even better performance can in general be achieved by sampling finite
temperature ($T$) distributions derived from the posterior. Taking the example
of two different deep (3 hidden layers) classifiers for MNIST data, we find
quite different $T$ values to be appropriate in each case. In particular, for a
typical neural network classifier a clear minimum of the test error is observed
at $T>0$. This suggests an early stopping criterion for full batch simulated
annealing: cool until the average validation error starts to increase, then
revert to the parameters with the lowest validation error. As $T$ is increased
classifiers transition from accurate classifiers to classifiers that have
higher training error than assigning equal probability to each class. Efficient
studies of these temperature-induced effects are enabled using a
replica-exchange Hamiltonian Monte Carlo simulation technique. Finally, we show
how thermodynamic integration can be used to perform model selection for deep
neural networks. Similar to the Laplace approximation, this approach assumes
that the posterior is dominated by a single mode. Crucially, however, no
assumption is made about the shape of that mode and it is not required to
precisely compute and invert the Hessian.
| stat |
Modeling semi-competing risks data as a longitudinal bivariate process | The Adult Changes in Thought (ACT) study is a long-running prospective study
of incident all-cause dementia and Alzheimer's disease (AD). As the cohort
ages, death (a terminal event) is a prominent competing risk for AD (a
non-terminal event), although the reverse is not the case. As such, analyses of
data from ACT can be placed within the semi-competing risks framework. Central
to semi-competing risks, and in contrast to standard competing risks, is that
one can learn about the dependence structure between the two events. To-date,
however, most methods for semi-competing risks treat dependence as a nuisance
and not a potential source of new clinical knowledge. We propose a novel
regression-based framework that views the two time-to-event outcomes through
the lens of a longitudinal bivariate process on a partition of the time scale.
A key innovation of the framework is that dependence is represented in two
distinct forms, $\textit{local}$ and $\textit{global}$ dependence, both of
which have intuitive clinical interpretations. Estimation and inference are
performed via penalized maximum likelihood, and can accommodate right
censoring, left truncation and time-varying covariates. The framework is used
to investigate the role of gender and having $\ge$1 APOE-$\epsilon4$ allele on
the joint risk of AD and death.
| stat |
Sufficient Dimension Reduction for Average Causal Effect Estimation | Having a large number of covariates can have a negative impact on the quality
of causal effect estimation since confounding adjustment becomes unreliable
when the number of covariates is large relative to the samples available.
Propensity score is a common way to deal with a large covariate set, but the
accuracy of propensity score estimation (normally done by logistic regression)
is also challenged by large number of covariates. In this paper, we prove that
a large covariate set can be reduced to a lower dimensional representation
which captures the complete information for adjustment in causal effect
estimation. The theoretical result enables effective data-driven algorithms for
causal effect estimation. We develop an algorithm which employs a supervised
kernel dimension reduction method to search for a lower dimensional
representation for the original covariates, and then utilizes nearest neighbor
matching in the reduced covariate space to impute the counterfactual outcomes
to avoid large-sized covariate set problem. The proposed algorithm is evaluated
on two semi-synthetic and three real-world datasets and the results have
demonstrated the effectiveness of the algorithm.
| stat |
Multiplicity for a Group Sequential Trial with Biomarker Subpopulations | Biomarker subpopulations have become increasingly important for drug
development in targeted therapies. The use of biomarkers has the potential to
facilitate more effective outcomes by guiding patient selection appropriately,
thus enhancing the benefit-risk profile and improving trial power. Studying a
broad population simultaneously with a more targeted one allows the trial to
determine the population for which a treatment is effective and allows a goal
of making approved regulatory labeling as inclusive as is appropriate. We
examine new methods accounting for the complete correlation structure in group
sequential designs with hypotheses in nested subgroups. The designs provide
full control of family-wise Type I error rate. This extension of previous
methods accounting for either group sequential design or correlation between
subgroups improves efficiency (power or sample size) over a typical Bonferroni
approach for testing nested populations.
| stat |
Learning the Hypotheses Space from data Part II: Convergence and
Feasibility | In part \textit{I} we proposed a structure for a general Hypotheses Space
$\mathcal{H}$, the Learning Space $\mathbb{L}(\mathcal{H})$, which can be
employed to avoid \textit{overfitting} when estimating in a complex space with
relative shortage of examples. Also, we presented the U-curve property, which
can be taken advantage of in order to select a Hypotheses Space without
exhaustively searching $\mathbb{L}(\mathcal{H})$. In this paper, we carry
further our agenda, by showing the consistency of a model selection framework
based on Learning Spaces, in which one selects from data the Hypotheses Space
on which to learn. The method developed in this paper adds to the
state-of-the-art in model selection, by extending Vapnik-Chervonenkis Theory to
\textit{random} Hypotheses Spaces, i.e., Hypotheses Spaces learned from data.
In this framework, one estimates a random subspace $\hat{\mathcal{M}} \in
\mathbb{L}(\mathcal{H})$ which converges with probability one to a target
Hypotheses Space $\mathcal{M}^{\star} \in \mathbb{L}(\mathcal{H})$ with desired
properties. As the convergence implies asymptotic unbiased estimators, we have
a consistent framework for model selection, showing that it is feasible to
learn the Hypotheses Space from data. Furthermore, we show that the
generalization errors of learning on $\hat{\mathcal{M}}$ are lesser than those
we commit when learning on $\mathcal{H}$, so it is more efficient to learn on a
subspace learned from data.
| stat |
A pragmatic adaptive enrichment design for selecting the right target
population for cancer immunotherapies | One of the challenges in the design of confirmatory trials is to deal with
uncertainties regarding the optimal target population for a novel drug.
Adaptive enrichment designs (AED) which allow for a data-driven selection of
one or more pre-specified biomarker subpopulations at an interim analysis have
been proposed in this setting but practical case studies of AEDs are still
relatively rare. We present the design of an AED with a binary endpoint in the
highly dynamic setting of cancer immunotherapy. The trial was initiated as a
conventional trial in early triple-negative breast cancer but amended to an AED
based on emerging data external to the trial suggesting that PD-L1 status could
be a predictive biomarker. Operating characteristics are discussed including
the concept of a minimal detectable difference, that is, the smallest observed
treatment effect that would lead to a statistically significant result in at
least one of the target populations at the interim or the final analysis,
respectively, in the setting of AED.
| stat |
A Particle Method for Solving Fredholm Equations of the First Kind | Fredholm integral equations of the first kind are the prototypical example of
ill-posed linear inverse problems. They model, among other things,
reconstruction of distorted noisy observations and indirect density estimation
and also appear in instrumental variable regression. However, their numerical
solution remains a challenging problem. Many techniques currently available
require a preliminary discretization of the domain of the solution and make
strong assumptions about its regularity. For example, the popular expectation
maximization smoothing (EMS) scheme requires the assumption of piecewise
constant solutions which is inappropriate for most applications. We propose
here a novel particle method that circumvents these two issues. This algorithm
can be thought of as a Monte Carlo approximation of the EMS scheme which not
only performs an adaptive stochastic discretization of the domain but also
results in smooth approximate solutions. We analyze the theoretical properties
of the EMS iteration and of the corresponding particle algorithm. Compared to
standard EMS, we show experimentally that our novel particle method provides
state-of-the-art performance for realistic systems, including motion deblurring
and reconstruction of cross-section images of the brain from positron emission
tomography.
| stat |
Binacox: automatic cut-point detection in high-dimensional Cox model
with applications in genetics | We introduce the binacox, a prognostic method to deal with the problem of
detecting multiple cut-points per features in a multivariate setting where a
large number of continuous features are available. The method is based on the
Cox model and combines one-hot encoding with the binarsity penalty, which uses
total-variation regularization together with an extra linear constraint, and
enables feature selection. Original nonasymptotic oracle inequalities for
prediction (in terms of Kullback-Leibler divergence) and estimation with a fast
rate of convergence are established. The statistical performance of the method
is examined in an extensive Monte Carlo simulation study, and then illustrated
on three publicly available genetic cancer datasets. On these high-dimensional
datasets, our proposed method significantly outperforms state-of-the-art
survival models regarding risk prediction in terms of the C-index, with a
computing time orders of magnitude faster. In addition, it provides powerful
interpretability from a clinical perspective by automatically pinpointing
significant cut-points in relevant variables.
| stat |
Maximum likelihood estimation of regularisation parameters in
high-dimensional inverse problems: an empirical Bayesian approach. Part I:
Methodology and Experiments | Many imaging problems require solving an inverse problem that is
ill-conditioned or ill-posed. Imaging methods typically address this difficulty
by regularising the estimation problem to make it well-posed. This often
requires setting the value of the so-called regularisation parameters that
control the amount of regularisation enforced. These parameters are notoriously
difficult to set a priori, and can have a dramatic impact on the recovered
estimates. In this work, we propose a general empirical Bayesian method for
setting regularisation parameters in imaging problems that are convex w.r.t.
the unknown image. Our method calibrates regularisation parameters directly
from the observed data by maximum marginal likelihood estimation, and can
simultaneously estimate multiple regularisation parameters. Furthermore, the
proposed algorithm uses the same basic operators as proximal optimisation
algorithms, namely gradient and proximal operators, and it is therefore
straightforward to apply to problems that are currently solved by using
proximal optimisation techniques. Our methodology is demonstrated with a range
of experiments and comparisons with alternative approaches from the literature.
The considered experiments include image denoising, non-blind image
deconvolution, and hyperspectral unmixing, using synthesis and analysis priors
involving the L1, total-variation, total-variation and L1, and
total-generalised-variation pseudo-norms. A detailed theoretical analysis of
the proposed method is presented in the companion paper arXiv:2008.05793.
| stat |
Optimal Stopping via Randomized Neural Networks | This paper presents new machine learning approaches to approximate the
solution of optimal stopping problems. The key idea of these methods is to use
neural networks, where the hidden layers are generated randomly and only the
last layer is trained, in order to approximate the continuation value. Our
approaches are applicable for high dimensional problems where the existing
approaches become increasingly impractical. In addition, since our approaches
can be optimized using a simple linear regression, they are very easy to
implement and theoretical guarantees can be provided. In Markovian examples our
randomized reinforcement learning approach and in non-Markovian examples our
randomized recurrent neural network approach outperform the state-of-the-art
and other relevant machine learning approaches.
| stat |
Robust Empirical Bayes Small Area Estimation with Density Power
Divergence | A two-stage normal hierarchical model called the Fay--Herriot model and the
empirical Bayes estimator are widely used to provide indirect and model-based
estimates of means in small areas. However, the performance of the empirical
Bayes estimator might be poor when the assumed normal distribution is
misspecified. In this article, we propose a simple modification by using
density power divergence and suggest a new robust empirical Bayes small area
estimator. The mean squared error and estimated mean squared error of the
proposed estimator are derived based on the asymptotic properties of the robust
estimator of the model parameters. We investigate the numerical performance of
the proposed method through simulations and an application to survey data.
| stat |
Fast ABC with joint generative modelling and subset simulation | We propose a novel approach for solving inverse-problems with
high-dimensional inputs and an expensive forward mapping. It leverages joint
deep generative modelling to transfer the original problem spaces to a lower
dimensional latent space. By jointly modelling input and output variables and
endowing the latent with a prior distribution, the fitted probabilistic model
indirectly gives access to the approximate conditional distributions of
interest. Since model error and observational noise with unknown distributions
are common in practice, we resort to likelihood-free inference with Approximate
Bayesian Computation (ABC). Our method calls on ABC by Subset Simulation to
explore the regions of the latent space with dissimilarities between generated
and observed outputs below prescribed thresholds. We diagnose the diversity of
approximate posterior solutions by monitoring the probability content of these
regions as a function of the threshold. We further analyze the curvature of the
resulting diagnostic curve to propose an adequate ABC threshold. When applied
to a cross-borehole tomography example from geophysics, our approach delivers
promising performance without using prior knowledge of the forward nor of the
noise distribution.
| stat |
Integrating overlapping datasets using bivariate causal discovery | Causal knowledge is vital for effective reasoning in science, as causal
relations, unlike correlations, allow one to reason about the outcomes of
interventions. Algorithms that can discover causal relations from observational
data are based on the assumption that all variables have been jointly measured
in a single dataset. In many cases this assumption fails. Previous approaches
to overcoming this shortcoming devised algorithms that returned all joint
causal structures consistent with the conditional independence information
contained in each individual dataset. But, as conditional independence tests
only determine causal structure up to Markov equivalence, the number of
consistent joint structures returned by these approaches can be quite large.
The last decade has seen the development of elegant algorithms for discovering
causal relations beyond conditional independence, which can distinguish among
Markov equivalent structures. In this work we adapt and extend these so-called
bivariate causal discovery algorithms to the problem of learning consistent
causal structures from multiple datasets with overlapping variables belonging
to the same generating process, providing a sound and complete algorithm that
outperforms previous approaches on synthetic and real data.
| stat |
Approximating multivariate posterior distribution functions from Monte
Carlo samples for sequential Bayesian inference | An important feature of Bayesian statistics is the opportunity to do
sequential inference: the posterior distribution obtained after seeing a
dataset can be used as prior for a second inference. However, when Monte Carlo
sampling methods are used for inference, we only have a set of samples from the
posterior distribution. To do sequential inference, we then either have to
evaluate the second posterior at only these locations and reweight the samples
accordingly, or we can estimate a functional description of the posterior
probability distribution from the samples and use that as prior for the second
inference. Here, we investigated to what extent we can obtain an accurate joint
posterior from two datasets if the inference is done sequentially rather than
jointly, under the condition that each inference step is done using Monte Carlo
sampling. To test this, we evaluated the accuracy of kernel density estimates,
Gaussian mixtures, vine copulas and Gaussian processes in approximating
posterior distributions, and then tested whether these approximations can be
used in sequential inference. In low dimensionality, Gaussian processes are
more accurate, whereas in higher dimensionality Gaussian mixtures or vine
copulas perform better. In our test cases, posterior approximations are
preferable over direct sample reweighting, although joint inference is still
preferable over sequential inference. Since the performance is case-specific,
we provide an R package mvdens with a unified interface for the density
approximation methods.
| stat |
Anomaly Detection in Stationary Settings: A Permutation-Based Higher
Criticism Approach | Anomaly detection when observing a large number of data streams is essential
in a variety of applications, ranging from epidemiological studies to
monitoring of complex systems. High-dimensional scenarios are usually tackled
with scan-statistics and related methods, requiring stringent modeling
assumptions for proper calibration. In this work we take a non-parametric
stance, and propose a permutation-based variant of the higher criticism
statistic not requiring knowledge of the null distribution. This results in an
exact test in finite samples which is asymptotically optimal in the wide class
of exponential models. We demonstrate the power loss in finite samples is
minimal with respect to the oracle test. Furthermore, since the proposed
statistic does not rely on asymptotic approximations it typically performs
better than popular variants of higher criticism that rely on such
approximations. We include recommendations such that the test can be readily
applied in practice, and demonstrate its applicability in monitoring the daily
number of COVID-19 cases in the Netherlands.
| stat |
MCMC for Bayesian uncertainty quantification from time-series data | Many problems in science and engineering require uncertainty quantification
that accounts for observed data. For example, in computational neuroscience,
Neural Population Models (NPMs) are mechanistic models that describe brain
physiology in a range of different states. Within computational neuroscience
there is growing interest in the inverse problem of inferring NPM parameters
from recordings such as the EEG (Electroencephalogram). Uncertainty
quantification is essential in this application area in order to infer the
mechanistic effect of interventions such as anaesthesia. This paper presents
C++ software for Bayesian uncertainty quantification in the parameters of NPMs
from approximately stationary data using Markov Chain Monte Carlo (MCMC).
Modern MCMC methods require first order (and in some cases higher order)
derivatives of the posterior density. The software presented offers two
distinct methods of evaluating derivatives: finite differences and exact
derivatives obtained through Algorithmic Differentiation (AD). For AD, two
different implementations are used: the open source Stan Math Library and the
commercially licenced dco/c++ tool distributed by NAG (Numerical Algorithms
Group). The use of derivative information in MCMC sampling is demonstrated
through a simple example, the noise-driven harmonic oscillator. And different
methods for computing derivatives are compared. The software is written in a
modular object-oriented way such that it can be extended to derivative based
MCMC for other scientific domains.
| stat |
Multivariate Convolutional Sparse Coding with Low Rank Tensor | This paper introduces a new multivariate convolutional sparse coding based on
tensor algebra with a general model enforcing both element-wise sparsity and
low-rankness of the activations tensors. By using the CP decomposition, this
model achieves a significantly more efficient encoding of the multivariate
signal-particularly in the high order/ dimension setting-resulting in better
performance. We prove that our model is closely related to the Kruskal tensor
regression problem, offering interesting theoretical guarantees to our setting.
Furthermore, we provide an efficient optimization algorithm based on
alternating optimization to solve this model. Finally, we evaluate our
algorithm with a large range of experiments, highlighting its advantages and
limitations.
| stat |
Approximate Factor Models with Strongly Correlated Idiosyncratic Errors | We consider the estimation of approximate factor models for time series data,
where strong serial and cross-sectional correlations amongst the idiosyncratic
component are present. This setting comes up naturally in many applications,
but existing approaches in the literature rely on the assumption that such
correlations are weak, leading to mis-specification of the number of factors
selected and consequently inaccurate inference. In this paper, we explicitly
incorporate the dependent structure present in the idiosyncratic component
through lagged values of the observed multivariate time series. We formulate a
constrained optimization problem to estimate the factor space and the
transition matrices of the lagged values {\em simultaneously}, wherein the
constraints reflect the low rank nature of the common factors and the sparsity
of the transition matrices. We establish theoretical properties of the obtained
estimates, and introduce an easy-to-implement computational procedure for
empirical work. The performance of the model and the implementation procedure
is evaluated on synthetic data and compared with competing approaches, and
further illustrated on a data set involving weekly log-returns of 75 US large
financial institutions for the 2001-2016 period.
| stat |
A note of feature screening via rank-based coefficient of correlation | Feature screening is useful and popular to detect informative predictors for
ultrahigh-dimensional data before developing proceeding statistical analysis or
constructing statistical models. While a large body of feature screening
procedures has been developed, most of them are restricted on examining either
continuous or discrete responses. Moreover, even though many model-free feature
screening methods have been proposed, additional assumptions are imposed in
those methods to ensure their theoretical results. To address those
difficulties and provide simple implementation, in this paper we extend the
rank-based coefficient of correlation proposed by Chatterjee (2020) to develop
feature screening procedure. We show that this new screening criterion is able
to deal with continuous and discrete responses. Theoretically, sure screening
property is established to justify the proposed method. Simulation studies
demonstrate that the predictors with nonlinear and oscillatory trajectory are
successfully detected regardless of the distribution of the response.
| stat |
Disentangled Attribution Curves for Interpreting Random Forests and
Boosted Trees | Tree ensembles, such as random forests and AdaBoost, are ubiquitous machine
learning models known for achieving strong predictive performance across a wide
variety of domains. However, this strong performance comes at the cost of
interpretability (i.e. users are unable to understand the relationships a
trained random forest has learned and why it is making its predictions). In
particular, it is challenging to understand how the contribution of a
particular feature, or group of features, varies as their value changes. To
address this, we introduce Disentangled Attribution Curves (DAC), a method to
provide interpretations of tree ensemble methods in the form of (multivariate)
feature importance curves. For a given variable, or group of variables, DAC
plots the importance of a variable(s) as their value changes. We validate DAC
on real data by showing that the curves can be used to increase the accuracy of
logistic regression while maintaining interpretability, by including DAC as an
additional feature. In simulation studies, DAC is shown to out-perform
competing methods in the recovery of conditional expectations. Finally, through
a case-study on the bike-sharing dataset, we demonstrate the use of DAC to
uncover novel insights into a dataset.
| stat |
Deep Gaussian Markov Random Fields | Gaussian Markov random fields (GMRFs) are probabilistic graphical models
widely used in spatial statistics and related fields to model dependencies over
spatial structures. We establish a formal connection between GMRFs and
convolutional neural networks (CNNs). Common GMRFs are special cases of a
generative model where the inverse mapping from data to latent variables is
given by a 1-layer linear CNN. This connection allows us to generalize GMRFs to
multi-layer CNN architectures, effectively increasing the order of the
corresponding GMRF in a way which has favorable computational scaling. We
describe how well-established tools, such as autodiff and variational
inference, can be used for simple and efficient inference and learning of the
deep GMRF. We demonstrate the flexibility of the proposed model and show that
it outperforms the state-of-the-art on a dataset of satellite temperatures, in
terms of prediction and predictive uncertainty.
| stat |
SQL-Rank: A Listwise Approach to Collaborative Ranking | In this paper, we propose a listwise approach for constructing user-specific
rankings in recommendation systems in a collaborative fashion. We contrast the
listwise approach to previous pointwise and pairwise approaches, which are
based on treating either each rating or each pairwise comparison as an
independent instance respectively. By extending the work of (Cao et al. 2007),
we cast listwise collaborative ranking as maximum likelihood under a
permutation model which applies probability mass to permutations based on a low
rank latent score matrix. We present a novel algorithm called SQL-Rank, which
can accommodate ties and missing data and can run in linear time. We develop a
theoretical framework for analyzing listwise ranking methods based on a novel
representation theory for the permutation model. Applying this framework to
collaborative ranking, we derive asymptotic statistical rates as the number of
users and items grow together. We conclude by demonstrating that our SQL-Rank
method often outperforms current state-of-the-art algorithms for implicit
feedback such as Weighted-MF and BPR and achieve favorable results when
compared to explicit feedback algorithms such as matrix factorization and
collaborative ranking.
| stat |
Approximation of Functions over Manifolds: A Moving Least-Squares
Approach | We present an algorithm for approximating a function defined over a
$d$-dimensional manifold utilizing only noisy function values at locations
sampled from the manifold with noise. To produce the approximation we do not
require any knowledge regarding the manifold other than its dimension $d$. We
use the Manifold Moving Least-Squares approach of (Sober and Levin 2016) to
reconstruct the atlas of charts and the approximation is built on-top of those
charts. The resulting approximant is shown to be a function defined over a
neighborhood of a manifold, approximating the originally sampled manifold. In
other words, given a new point, located near the manifold, the approximation
can be evaluated directly on that point. We prove that our construction yields
a smooth function, and in case of noiseless samples the approximation order is
$\mathcal{O}(h^{m+1})$, where $h$ is a local density of sample parameter (i.e.,
the fill distance) and $m$ is the degree of a local polynomial approximation,
used in our algorithm. In addition, the proposed algorithm has linear time
complexity with respect to the ambient-space's dimension. Thus, we are able to
avoid the computational complexity, commonly encountered in high dimensional
approximations, without having to perform non-linear dimension reduction, which
inevitably introduces distortions to the geometry of the data. Additionaly, we
show numerical experiments that the proposed approach compares favorably to
statistical approaches for regression over manifolds and show its potential.
| stat |
Robust real-time monitoring of high-dimensional data streams | Robust real-time monitoring of high-dimensional data streams has many
important real-world applications such as industrial quality control, signal
detection, biosurveillance, but unfortunately it is highly non-trivial to
develop efficient schemes due to two challenges: (1) the unknown sparse number
or subset of affected data streams and (2) the uncertainty of model
specification for high-dimensional data. In this article, motivated by the
detection of smaller persistent changes in the presence of larger transient
outliers, we develop a family of efficient real-time robust detection schemes
for high-dimensional data streams through monitoring feature spaces such as PCA
or wavelet coefficients when the feature coefficients are from Tukey-Huber's
gross error models with outliers. We propose to construct a new local detection
statistic for each feature called $L_{\alpha}$-CUSUM statistic that can reduce
the effect of outliers by using the Box-Cox transformation of the likelihood
function, and then raise a global alarm based upon the sum of the
soft-thresholding transformation of these local $L_{\alpha}$-CUSUM statistics
so that to filter out unaffected features. In addition, we propose a new
concept called false alarm breakdown point to measure the robustness of online
monitoring schemes, and also characterize the breakdown point of our proposed
schemes. Asymptotic analysis, extensive numerical simulations and case study of
nonlinear profile monitoring are conducted to illustrate the robustness and
usefulness of our proposed schemes.
| stat |
A folded model for compositional data analysis | A folded type model is developed for analyzing compositional data. The
proposed model involves an extension of the $\alpha$-transformation for
compositional data and provides a new and flexible class of distributions for
modeling data defined on the simplex sample space. Despite its rather seemingly
complex structure, employment of the EM algorithm guarantees efficient
parameter estimation. The model is validated through simulation studies and
examples which illustrate that the proposed model performs better in terms of
capturing the data structure, when compared to the popular logistic normal
distribution, and can be advantageous over a similar model without folding.
| stat |
Beyond the Signs: Nonparametric Tensor Completion via Sign Series | We consider the problem of tensor estimation from noisy observations with
possibly missing entries. A nonparametric approach to tensor completion is
developed based on a new model which we coin as sign representable tensors. The
model represents the signal tensor of interest using a series of structured
sign tensors. Unlike earlier methods, the sign series representation
effectively addresses both low- and high-rank signals, while encompassing many
existing tensor models -- including CP models, Tucker models, single index
models, several hypergraphon models -- as special cases. We show that the sign
tensor series is theoretically characterized, and computationally estimable,
via classification tasks with carefully-specified weights. Excess risk bounds,
estimation error rates, and sample complexities are established. We demonstrate
the outperformance of our approach over previous methods on two datasets, one
on human brain connectivity networks and the other on topic data mining.
| stat |
Black-Box Inference for Non-Linear Latent Force Models | Latent force models are systems whereby there is a mechanistic model
describing the dynamics of the system state, with some unknown forcing term
that is approximated with a Gaussian process. If such dynamics are non-linear,
it can be difficult to estimate the posterior state and forcing term jointly,
particularly when there are system parameters that also need estimating. This
paper uses black-box variational inference to jointly estimate the posterior,
designing a multivariate extension to local inverse autoregressive flows as a
flexible approximater of the system. We compare estimates on systems where the
posterior is known, demonstrating the effectiveness of the approximation, and
apply to problems with non-linear dynamics, multi-output systems and models
with non-Gaussian likelihoods.
| stat |
MetaFun: Meta-Learning with Iterative Functional Updates | We develop a functional encoder-decoder approach to supervised meta-learning,
where labeled data is encoded into an infinite-dimensional functional
representation rather than a finite-dimensional one. Furthermore, rather than
directly producing the representation, we learn a neural update rule resembling
functional gradient descent which iteratively improves the representation. The
final representation is used to condition the decoder to make predictions on
unlabeled data. Our approach is the first to demonstrates the success of
encoder-decoder style meta-learning methods like conditional neural processes
on large-scale few-shot classification benchmarks such as miniImageNet and
tieredImageNet, where it achieves state-of-the-art performance.
| stat |
Bias in the estimation of cumulative viremia in cohort studies of
HIV-infected individuals | Purpose: The use of cumulative measures of exposure to raised HIV viral load
(viremia copy-years) is an increasingly common in HIV prevention and treatment
epidemiology due to the high biological plausibility. We sought to estimate the
magnitude and direction of bias in a cumulative measure of viremia caused by
different frequency of sampling and duration of follow-up.
Methods: We simulated longitudinal viral load measures and reanalysed cohort
study datasets with longitudinal viral load measurements under different
sampling strategies to estimate cumulative viremia.
Results: In both simulated and observed data, estimates of cumulative viremia
by the trapezoidal rule show systematic upward bias when there are fewer
sampling time points and/or increased duration between sampling time points,
compared to estimation of full time series. Absolute values of cumulative
viremia vary appreciably by the patterns of viral load over time, even after
adjustment for total duration of follow up.
Conclusions: Sampling bias due to differential frequency of sampling appears
extensive and of meaningful magnitude in measures of cumulative viremia.
Cumulative measures of viremia should be used only in studies with sufficient
frequency of viral load measures and always as relative measures.
| stat |
Detecting Abrupt Changes in the Presence of Local Fluctuations and
Autocorrelated Noise | Whilst there are a plethora of algorithms for detecting changes in mean in
univariate time-series, almost all struggle in real applications where there is
autocorrelated noise or where the mean fluctuates locally between the abrupt
changes that one wishes to detect. In these cases, default implementations,
which are often based on assumptions of a constant mean between changes and
independent noise, can lead to substantial over-estimation of the number of
changes. We propose a principled approach to detect such abrupt changes that
models local fluctuations as a random walk process and autocorrelated noise via
an AR(1) process. We then estimate the number and location of changepoints by
minimising a penalised cost based on this model. We develop a novel and
efficient dynamic programming algorithm, DeCAFS, that can solve this
minimisation problem; despite the additional challenge of dependence across
segments, due to the autocorrelated noise, which makes existing algorithms
inapplicable. Theory and empirical results show that our approach has greater
power at detecting abrupt changes than existing approaches. We apply our method
to measuring gene expression levels in bacteria.
| stat |
Bayes Calculations from Quantile Implied Likelihood | In statistical practice, a realistic Bayesian model for a given data set can
be defined by a likelihood function that is analytically or computationally
intractable, due to large data sample size, high parameter dimensionality, or
complex likelihood functional form. This in turn poses challenges to the
computation and inference of the posterior distribution of the model
parameters. For such a model, a tractable likelihood function is introduced
which approximates the exact likelihood through its quantile function. It is
defined by an asymptotic chi-square confidence distribution for a pivotal
quantity, which is generated by the asymptotic normal distribution of the
sample quantiles given model parameters. This Quantile Implied Likelihood (QIL)
gives rise to an approximate posterior distribution which can be estimated by
using penalized log-likelihood maximization or any suitable Monte Carlo
algorithm. The QIL approach to Bayesian Computation is illustrated through the
Bayesian analysis of simulated and real data sets having sample sizes that
reach the millions. The analyses involve various models for univariate or
multivariate iid or non-iid data, with low or high parameter dimensionality,
many of which are defined by intractable likelihoods. The probability models
include the Student's t, g-and-h, and g-and-k distributions; the Bayesian logit
regression model with many covariates; exponential random graph model, a
doubly-intractable model for networks; the multivariate skew normal model, for
robust inference of the inverse-covariance matrix when it is large relative to
the sample size; and the Wallenius distribution model.
| stat |
Reflection on modern methods: Good practices for applied statistical
learning in epidemiology | Statistical learning (SL) includes methods that extract knowledge from
complex data. SL methods beyond generalized linear models are being
increasingly implemented in public health research and epidemiology because
they can perform better in instances with complex or high-dimensional
data---settings when traditional statistical methods fail. These novel methods,
however, often include random sampling which may induce variability in results.
Best practices in data science can help to ensure robustness. As a case study,
we included four SL models that have been applied previously to analyze the
relationship between environmental mixtures and health outcomes. We ran each
model across 100 initializing values for random number generation, or "seeds,"
and assessed variability in resulting estimation and inference. All methods
exhibited some seed-dependent variability in results. The degree of variability
differed across methods and exposure of interest. Any SL method reliant on a
random seed will exhibit some degree of seed sensitivity. We recommend that
researchers repeat their analysis with various seeds as a sensitivity analysis
when implementing these methods to enhance interpretability and robustness of
results.
| stat |
The role of regularization in classification of high-dimensional noisy
Gaussian mixture | We consider a high-dimensional mixture of two Gaussians in the noisy regime
where even an oracle knowing the centers of the clusters misclassifies a small
but finite fraction of the points. We provide a rigorous analysis of the
generalization error of regularized convex classifiers, including ridge, hinge
and logistic regression, in the high-dimensional limit where the number $n$ of
samples and their dimension $d$ go to infinity while their ratio is fixed to
$\alpha= n/d$. We discuss surprising effects of the regularization that in some
cases allows to reach the Bayes-optimal performances. We also illustrate the
interpolation peak at low regularization, and analyze the role of the
respective sizes of the two clusters.
| stat |
Estimating regression errors without ground truth values | Regression analysis is a standard supervised machine learning method used to
model an outcome variable in terms of a set of predictor variables. In most
real-world applications we do not know the true value of the outcome variable
being predicted outside the training data, i.e., the ground truth is unknown.
It is hence not straightforward to directly observe when the estimate from a
model potentially is wrong, due to phenomena such as overfitting and concept
drift. In this paper we present an efficient framework for estimating the
generalization error of regression functions, applicable to any family of
regression functions when the ground truth is unknown. We present a theoretical
derivation of the framework and empirically evaluate its strengths and
limitations. We find that it performs robustly and is useful for detecting
concept drift in datasets in several real-world domains.
| stat |
Screening Rules and its Complexity for Active Set Identification | Screening rules were recently introduced as a technique for explicitly
identifying active structures such as sparsity, in optimization problem arising
in machine learning. This has led to new methods of acceleration based on a
substantial dimension reduction. We show that screening rules stem from a
combination of natural properties of subdifferential sets and optimality
conditions, and can hence be understood in a unified way. Under mild
assumptions, we analyze the number of iterations needed to identify the optimal
active set for any converging algorithm. We show that it only depends on its
convergence rate.
| stat |
A Brief Note on the Convergence of Langevin Monte Carlo in Chi-Square
Divergence | We study sampling from a target distribution $\nu_* \propto e^{-f}$ using the
unadjusted Langevin Monte Carlo (LMC) algorithm when the target $\nu_*$
satisfies the Poincar\'e inequality, and the potential $f$ is first-order
smooth and dissipative. Under an opaque uniform warmness condition on the LMC
iterates, we establish that $\widetilde{\mathcal{O}}(\epsilon^{-1})$ steps are
sufficient for LMC to reach $\epsilon$ neighborhood of the target in Chi-square
divergence. We hope that this note serves as a step towards establishing a
complete convergence analysis of LMC under Chi-square divergence.
| stat |
An adequacy approach for deciding the number of clusters for OTRIMLE
robust Gaussian mixture based clustering | We introduce a new approach to deciding the number of clusters. The approach
is applied to Optimally Tuned Robust Improper Maximum Likelihood Estimation
(OTRIMLE; Coretto and Hennig 2016) of a Gaussian mixture model allowing for
observations to be classified as "noise", but it can be applied to other
clustering methods as well. The quality of a clustering is assessed by a
statistic $Q$ that measures how close the within-cluster distributions are to
elliptical unimodal distributions that have the only mode in the mean. This
nonparametric measure allows for non-Gaussian clusters as long as they have a
good quality according to $Q$. The simplicity of a model is assessed by a
measure $S$ that prefers a smaller number of clusters unless additional
clusters can reduce the estimated noise proportion substantially. The simplest
model is then chosen that is adequate for the data in the sense that its
observed value of $Q$ is not significantly larger than what is expected for
data truly generated from the fitted model, as can be assessed by parametric
bootstrap. The approach is compared with model-based clustering using the
Bayesian Information Criterion (BIC) and the Integrated Complete Likelihood
(ICL) in a simulation study and on two datasets of scientific interest.
Keywords: parametric bootstrap; noise component; unimodality; model-based
clustering
| stat |
Solving Optimal Experimental Design with Sequential Quadratic
Programming and Chebyshev Interpolation | We propose an optimization algorithm to compute the optimal sensor locations
in experimental design in the formulation of Bayesian inverse problems, where
the parameter-to-observable mapping is described through an integral equation
and its discretization results in a continuously indexed matrix whose size
depends on the mesh size n. By approximating the gradient and Hessian of the
objective design criterion from Chebyshev interpolation, we solve a sequence of
quadratic programs and achieve the complexity $\mathcal{O}(n\log^2(n))$. An
error analysis guarantees the integrality gap shrinks to zero as $n\to\infty$,
and we apply the algorithm on a two-dimensional advection-diffusion equation,
to determine the LIDAR's optimal sensing directions for data collection.
| stat |
Stochastic Variance-Reduced Hamilton Monte Carlo Methods | We propose a fast stochastic Hamilton Monte Carlo (HMC) method, for sampling
from a smooth and strongly log-concave distribution. At the core of our
proposed method is a variance reduction technique inspired by the recent
advance in stochastic optimization. We show that, to achieve $\epsilon$
accuracy in 2-Wasserstein distance, our algorithm achieves $\tilde
O(n+\kappa^{2}d^{1/2}/\epsilon+\kappa^{4/3}d^{1/3}n^{2/3}/\epsilon^{2/3})$
gradient complexity (i.e., number of component gradient evaluations), which
outperforms the state-of-the-art HMC and stochastic gradient HMC methods in a
wide regime. We also extend our algorithm for sampling from smooth and general
log-concave distributions, and prove the corresponding gradient complexity as
well. Experiments on both synthetic and real data demonstrate the superior
performance of our algorithm.
| stat |
A Bayesian semi-parametric approach for inference on the population
partly conditional mean from longitudinal data with dropout | Studies of memory trajectories using longitudinal data often result in highly
non-representative samples due to selective study enrollment and attrition. An
additional bias comes from practice effects that result in improved or
maintained performance due to familiarity with test content or context. These
challenges may bias study findings and severely distort the ability to
generalize to the target population. In this study we propose an approach for
estimating the finite population mean of a longitudinal outcome conditioning on
being alive at a specific time point. We develop a flexible Bayesian
semi-parametric predictive estimator for population inference when longitudinal
auxiliary information is known for the target population. We evaluate
sensitivity of the results to untestable assumptions and further compare our
approach to other methods used for population inference in a simulation study.
The proposed approach is motivated by 15-year longitudinal data from the Betula
longitudinal cohort study. We apply our approach to estimate lifespan
trajectories in episodic memory, with the aim to generalize findings to a
target population.
| stat |
Active Learning for Deep Gaussian Process Surrogates | Deep Gaussian processes (DGPs) are increasingly popular as predictive models
in machine learning (ML) for their non-stationary flexibility and ability to
cope with abrupt regime changes in training data. Here we explore DGPs as
surrogates for computer simulation experiments whose response surfaces exhibit
similar characteristics. In particular, we transport a DGP's automatic warping
of the input space and full uncertainty quantification (UQ), via a novel
elliptical slice sampling (ESS) Bayesian posterior inferential scheme, through
to active learning (AL) strategies that distribute runs non-uniformly in the
input space -- something an ordinary (stationary) GP could not do. Building up
the design sequentially in this way allows smaller training sets, limiting both
expensive evaluation of the simulator code and mitigating cubic costs of DGP
inference. When training data sizes are kept small through careful acquisition,
and with parsimonious layout of latent layers, the framework can be both
effective and computationally tractable. Our methods are illustrated on
simulation data and two real computer experiments of varying input
dimensionality. We provide an open source implementation in the "deepgp"
package on CRAN.
| stat |
Evaluation of Logistic Regression Applied to Respondent-Driven Samples:
Simulated and Real Data | Objective: To investigate the impact of different logistic regression
estimators applied to RDS samples obtained by simulation and real data.
Methods: Four simulated populations were created combining different
connectivity models, levels of clusterization and infection processes. Each
subject in the population received two attributes, only one of them related to
the infection process. From each population, RDS samples with different sizes
were obtained. Similarly, RDS samples were obtained from a real-world dataset.
Three logistic regression estimators were applied to assess the association
between the attributes and the infection status, and subsequently the observed
coverage of each was measured. Results: The type of connectivity had more
impact on estimators performance than the clusterization level. In simulated
datasets, unweighted logistic regression estimators emerged as the best option,
although all estimators showed a fairly good performance. In the real dataset,
the performance of weighted estimators presented some instabilities, making
them a risky option. Conclusion: An unweighted logistic regression estimator is
a reliable option to be applied to RDS samples, with similar performance to
random samples and, therefore, should be the preferred option.
| stat |
Six-Day Footraces in the Post-Pedestrianism Era | In a six-day footrace, competitors accumulate as much distance as possible on
foot over 144 consecutive hours by circumambulating a loop course. Now an
obscure event on the fringe of ultra running and contested by amateurs, six-day
races and the associated sport of pedestrianism used to be a lucrative
professional athletic endeavor. Indeed, pedestrianism was the most popular
spectator sport in America c. 1874-c. 1881. We analyzed data from 277 six-day
races spanning 37 years in the post-pedestrianism era (1981-2018). Men
outnumber women 3:1 in six-day race participation. The men's (women's) six-day
world record is 644.2 (549.1) miles and the top 4% achieve 500 (450) miles.
Adopting the forecasting model of Godsey (2012), we predict a 53% (21%)
probability that the men's (women's) world record will be broken within the
next decade.
| stat |
Using Bayesian deep learning approaches for uncertainty-aware building
energy surrogate models | Fast machine learning-based surrogate models are trained to emulate slow,
high-fidelity engineering simulation models to accelerate engineering design
tasks. This introduces uncertainty as the surrogate is only an approximation of
the original model.
Bayesian methods can quantify that uncertainty, and deep learning models
exist that follow the Bayesian paradigm. These models, namely Bayesian neural
networks and Gaussian process models, enable us to give predictions together
with an estimate of the model's uncertainty. As a result we can derive
uncertainty-aware surrogate models that can automatically suspect unseen design
samples that cause large emulation errors. For these samples, the high-fidelity
model can be queried instead. This outlines how the Bayesian paradigm allows us
to hybridize fast, but approximate, and slow, but accurate models.
In this paper, we train two types of Bayesian models, dropout neural networks
and stochastic variational Gaussian Process models, to emulate a complex high
dimensional building energy performance simulation problem. The surrogate model
processes 35 building design parameters (inputs) to estimate 12 different
performance metrics (outputs). We benchmark both approaches, prove their
accuracy to be competitive, and show that errors can be reduced by up to 30%
when the 10% of samples with the highest uncertainty are transferred to the
high-fidelity model.
| stat |
A Bayesian Mixture Modelling of Stop Signal Reaction Time Distributions | The distribution of single Stop Signal Reaction Times (SSRT) in the stop
signal task (SST) as a measurement of the latency of the unobservable stopping
process has been modeled with a nonparametric method by Hans Colonius (1990)
and with a Bayesian parametric method by Eric-Jan Wagenmakers and colleagues
(2012). These methods assume equal impact of the preceding trial type (go/stop)
in the SST trials on the SSRT distributional estimation without addressing the
case of the violated assumption. This study presents the required model by
considering two-state mixture model for the SSRT distribution. It then compares
the Bayesian parametric single SSRT and mixture SSRT distributions in the usual
stochastic order at the individual and the population level under the
ex-Gaussian distributional format. It shows that compared to a single SSRT
distribution, the mixture SSRT distribution is more diverse, more positively
skewed, more leptokurtic, and larger in stochastic order. The size of the
disparities in the results also depends on the choice of weights in the mixture
SSRT distribution. This study confirms that mixture SSRT indices as a constant
or distribution are significantly larger than their single SSRT counterparts in
the related order. This offers a vital improvement in the SSRT estimations.
| stat |
Bipartisan politics and poverty as a risk factor for contagion and
mortality from SARS-CoV-2 virus in the United States of America | In the United States, from the start of the COVID-19 pandemic to December 31,
2020, 341,199 deaths and more than 19,663,976 infections were recorded. Recent
literature establishes that communities with poverty-related health problems,
such as obesity, cardiovascular disease, diabetes, and hypertension, are more
susceptible to mortality from SARS-CoV-2 infection. Additionally, controversial
public health policies implemented by the nation's political leaders have
highlighted the socioeconomic inequalities of minorities. Therefore, through
multivariate correlational analysis using machine learning techniques and
structural equations, we measure whether social determinants are associated
with increased infection and death from COVID-19 disease. The PLS least squares
regression analysis allowed identifying a significant impact between social
determinants and COVID-19 disease through a predictive value of R2 = .916,
\b{eta} = .836, p =. 000 (t-value = 66,137) shows that for each unit of
increase in social determinants, COVID-19 disease increases by 91.6%. The
clustering index used for correlational analysis generated a new data set
comprising three groups: C1 Republicans, C2 and C3 Democrats from California,
New York, Texas, and Florida. This analysis made it possible to identify the
poverty variable as the main risk factor related to the high rates of infection
in Republican states and a high positive correlation between the population not
insured with a medical plan and high levels of virus contagion in the states of
group C3. These findings explain the argument that poverty and lack of economic
security put the public or private health system at risk and calamity.
| stat |
Getting a CLUE: A Method for Explaining Uncertainty Estimates | Both uncertainty estimation and interpretability are important factors for
trustworthy machine learning systems. However, there is little work at the
intersection of these two areas. We address this gap by proposing a novel
method for interpreting uncertainty estimates from differentiable probabilistic
models, like Bayesian Neural Networks (BNNs). Our method, Counterfactual Latent
Uncertainty Explanations (CLUE), indicates how to change an input, while
keeping it on the data manifold, such that a BNN becomes more confident about
the input's prediction. We validate CLUE through 1) a novel framework for
evaluating counterfactual explanations of uncertainty, 2) a series of ablation
experiments, and 3) a user study. Our experiments show that CLUE outperforms
baselines and enables practitioners to better understand which input patterns
are responsible for predictive uncertainty.
| stat |
An Optimal Policy for Dynamic Assortment Planning Under Uncapacitated
Multinomial Logit Models | We study the dynamic assortment planning problem, where for each arriving
customer, the seller offers an assortment of substitutable products and
customer makes the purchase among offered products according to an
uncapacitated multinomial logit (MNL) model. Since all the utility parameters
of MNL are unknown, the seller needs to simultaneously learn customers' choice
behavior and make dynamic decisions on assortments based on the current
knowledge. The goal of the seller is to maximize the expected revenue, or
equivalently, to minimize the expected regret. Although dynamic assortment
planning problem has received an increasing attention in revenue management,
most existing policies require the estimation of mean utility for each product
and the final regret usually involves the number of products $N$. The optimal
regret of the dynamic assortment planning problem under the most basic and
popular choice model---MNL model is still open. By carefully analyzing a
revenue potential function, we develop a trisection based policy combined with
adaptive confidence bound construction, which achieves an {item-independent}
regret bound of $O(\sqrt{T})$, where $T$ is the length of selling horizon. We
further establish the matching lower bound result to show the optimality of our
policy. There are two major advantages of the proposed policy. First, the
regret of all our policies has no dependence on $N$. Second, our policies are
almost assumption free: there is no assumption on mean utility nor any
"separability" condition on the expected revenues for different assortments.
Our result also extends the unimodal bandit literature.
| stat |
An Approximate Bayesian Approach to Model-assisted Survey Estimation
with Many Auxiliary Variables | Model-assisted estimation with complex survey data is an important practical
problem in survey sampling. When there are many auxiliary variables, selecting
significant variables associated with the study variable would be necessary to
achieve efficient estimation of population parameters of interest. In this
paper, we formulate a regularized regression estimator in the framework of
Bayesian inference using the penalty function as the shrinkage prior for model
selection. The proposed Bayesian approach enables us to get not only efficient
point estimates but also reasonable credible intervals. Results from two
limited simulation studies are presented to facilitate comparison with existing
frequentist methods.
| stat |
Debiased Sinkhorn barycenters | Entropy regularization in optimal transport (OT) has been the driver of many
recent interests for Wasserstein metrics and barycenters in machine learning.
It allows to keep the appealing geometrical properties of the unregularized
Wasserstein distance while having a significantly lower complexity thanks to
Sinkhorn's algorithm. However, entropy brings some inherent smoothing bias,
resulting for example in blurred barycenters. This side effect has prompted an
increasing temptation in the community to settle for a slower algorithm such as
log-domain stabilized Sinkhorn which breaks the parallel structure that can be
leveraged on GPUs, or even go back to unregularized OT. Here we show how this
bias is tightly linked to the reference measure that defines the entropy
regularizer and propose debiased Wasserstein barycenters that preserve the best
of both worlds: fast Sinkhorn-like iterations without entropy smoothing.
Theoretically, we prove that the entropic OT barycenter of univariate Gaussians
is a Gaussian and quantify its variance bias. This result is obtained by
extending the differentiability and convexity of entropic OT to sub-Gaussian
measures with unbounded supports. Empirically, we illustrate the reduced
blurring and the computational advantage on various applications.
| stat |
Preprocessing noisy functional data using factor models | We consider functional data which are measured on a discrete set of
observation points. Often such data are measured with noise, and then the
target is to recover the underlying signal. Most commonly, practitioners use
some smoothing approach, e.g.,\ kernel smoothing or spline fitting towards this
goal. The drawback of such curve fitting techniques is that they act function
by function, and don't take into account information from the entire sample. In
this paper we argue that signal and noise can be naturally represented as the
common and idiosyncratic component, respectively, of a factor model.
Accordingly, we propose to an estimation scheme which is based on factor
models. The purpose of this paper is to explain the reasoning behind our
approach and to compare its performance on simulated and on real data to
competing methods.
| stat |
A review of Bayesian perspectives on sample size derivation for
confirmatory trials | Sample size derivation is a crucial element of the planning phase of any
confirmatory trial. A sample size is typically derived based on constraints on
the maximal acceptable type I error rate and a minimal desired power. Here,
power depends on the unknown true effect size. In practice, power is typically
calculated either for the smallest relevant effect size or a likely point
alternative. The former might be problematic if the minimal relevant effect is
close to the null, thus requiring an excessively large sample size. The latter
is dubious since it does not account for the a priori uncertainty about the
likely alternative effect size. A Bayesian perspective on the sample size
derivation for a frequentist trial naturally emerges as a way of reconciling
arguments about the relative a priori plausibility of alternative effect sizes
with ideas based on the relevance of effect sizes. Many suggestions as to how
such `hybrid' approaches could be implemented in practice have been put forward
in the literature. However, key quantities such as assurance, probability of
success, or expected power are often defined in subtly different ways in the
literature. Starting from the traditional and entirely frequentist approach to
sample size derivation, we derive consistent definitions for the most commonly
used `hybrid' quantities and highlight connections, before discussing and
demonstrating their use in the context of sample size derivation for clinical
trials.
| stat |
Protocol for an Observational Study on the Effects of Early-Life
Participation in Contact Sports on Later-Life Cognition in a Sample of
Monozygotic and Dizygotic Swedish Twins Reared Together and Twins Reared
Apart | A large body of work links traumatic brain injury (TBI) in adulthood to the
onset of Alzheimer's disease (AD). AD is the chief cause of dementia, leading
to reduced cognitive capacity and autonomy and increased mortality risk. More
recently, researchers have sought to investigate whether TBI experienced in
early-life may influence trajectories of cognitive dysfunction in adulthood. It
has been speculated that early-life participation in collision sports may lead
to poor cognitive and mental health outcomes. However, to date, the few studies
to investigate this relationship have produced mixed results. We propose to
extend this literature by conducting a prospective study on the effects of
early-life participation in collision sports on later-life cognitive health
using the Swedish Adoption/Twin Study on Aging (SATSA). The SATSA is unique in
its sampling of monozygotic and dizygotic twins reared together (respectively
MZT, DZT) and twins reared apart (respectively MZA, DZA). The proposed analysis
is a prospective study of 660 individuals comprised of 270 twin pairs and 120
singletons. Seventy-eight (11.8% individuals reported participation in
collision sports. Our primary outcome will be an indicator of cognitive
impairment determined by scores on the Mini-Mental State Examination (MMSE). We
will also consider several secondary cognitive outcomes including verbal and
spatial ability, memory, and processing speed. Our sample will be restricted to
individuals with at least one MMSE score out of seven repeated assessments
spaced approximately three years apart. We will adjust for age, sex, and
education in each of our models.
| stat |
On choosing mixture components via non-local priors | Choosing the number of mixture components remains an elusive challenge. Model
selection criteria can be either overly liberal or conservative and return
poorly-separated components of limited practical use. We formalize non-local
priors (NLPs) for mixtures and show how they lead to well-separated components
with non-negligible weight, interpretable as distinct subpopulations. We also
propose an estimator for posterior model probabilities under local and
non-local priors, showing that Bayes factors are ratios of posterior to prior
empty-cluster probabilities. The estimator is widely applicable and helps set
thresholds to drop unoccupied components in overfitted mixtures. We suggest
default prior parameters based on multi-modality for Normal/T mixtures and
minimal informativeness for categorical outcomes. We characterise theoretically
the NLP-induced sparsity, derive tractable expressions and algorithms. We fully
develop Normal, Binomial and product Binomial mixtures but the theory,
computation and principles hold more generally. We observed a serious lack of
sensitivity of the Bayesian information criterion (BIC), insufficient parsimony
of the AIC and a local prior, and a mixed behavior of the singular BIC. We also
considered overfitted mixtures, their performance was competitive but depended
on tuning parameters. Under our default prior elicitation NLPs offered a good
compromise between sparsity and power to detect meaningfully-separated
components.
| stat |
Combining Cox Regressions Across a Heterogeneous Distributed Research
Network Facing Small and Zero Counts | Studies of the effects of medical interventions increasingly take place in
distributed research settings using data from multiple clinical data sources
including electronic health records and administrative claims. In such
settings, privacy concerns typically prohibit sharing of individual patient
data, and instead, analyses can only utilize summary statistics from the
individual databases. In the specific but very common context of the Cox
proportional hazards model, we show that standard meta analysis methods then
lead to substantial bias when outcome counts are small. This bias derives
primarily from the normal approximations that the methods utilize. Here we
propose and evaluate methods that eschew normal approximations in favor of
three more flexible approximations: a skew-normal, a one-dimensional grid, and
a custom parametric function that mimics the behavior of the Cox likelihood
function. In extensive simulation studies we demonstrate how these
approximations impact bias in the context of both fixed-effects and (Bayesian)
random-effects models. We then apply these approaches to three real-world
studies of the comparative safety of antidepressants, each using data from four
observational healthcare databases.
| stat |
Approximation Algorithms for D-optimal Design | Experimental design is a classical statistics problem and its aim is to
estimate an unknown $m$-dimensional vector $\beta$ from linear measurements
where a Gaussian noise is introduced in each measurement. For the combinatorial
experimental design problem, the goal is to pick $k$ out of the given $n$
experiments so as to make the most accurate estimate of the unknown parameters,
denoted as $\hat{\beta}$. In this paper, we will study one of the most robust
measures of error estimation - $D$-optimality criterion, which corresponds to
minimizing the volume of the confidence ellipsoid for the estimation error
$\beta-\hat{\beta}$. The problem gives rise to two natural variants depending
on whether repetitions of experiments are allowed or not. We first propose an
approximation algorithm with a $\frac1e$-approximation for the $D$-optimal
design problem with and without repetitions, giving the first constant factor
approximation for the problem. We then analyze another sampling approximation
algorithm and prove that it is $(1-\epsilon)$-approximation if $k\geq
\frac{4m}{\epsilon}+\frac{12}{\epsilon^2}\log(\frac{1}{\epsilon})$ for any
$\epsilon \in (0,1)$. Finally, for $D$-optimal design with repetitions, we
study a different algorithm proposed by literature and show that it can improve
this asymptotic approximation ratio.
| stat |
ODE$^2$VAE: Deep generative second order ODEs with Bayesian neural
networks | We present Ordinary Differential Equation Variational Auto-Encoder
(ODE$^2$VAE), a latent second order ODE model for high-dimensional sequential
data. Leveraging the advances in deep generative models, ODE$^2$VAE can
simultaneously learn the embedding of high dimensional trajectories and infer
arbitrarily complex continuous-time latent dynamics. Our model explicitly
decomposes the latent space into momentum and position components and solves a
second order ODE system, which is in contrast to recurrent neural network (RNN)
based time series models and recently proposed black-box ODE techniques. In
order to account for uncertainty, we propose probabilistic latent ODE dynamics
parameterized by deep Bayesian neural networks. We demonstrate our approach on
motion capture, image rotation and bouncing balls datasets. We achieve
state-of-the-art performance in long term motion prediction and imputation
tasks.
| stat |
Yield forecasting with machine learning and small data: what gains for
grains? | Forecasting crop yields is important for food security, in particular to
predict where crop production is likely to drop. Climate records and
remotely-sensed data have become instrumental sources of data for crop yield
forecasting systems. Similarly, machine learning methods are increasingly used
to process big Earth observation data. However, access to data necessary to
train such algorithms is often limited in food-insecure countries. Here, we
evaluate the performance of machine learning algorithms and small data to
forecast yield on a monthly basis between the start and the end of the growing
season. To do so, we developed a robust and automated machine-learning pipeline
which selects the best features and model for prediction. Taking Algeria as
case study, we predicted national yields for barley, soft wheat and durum wheat
with an accuracy of 0.16-0.2 t/ha (13-14 % of mean yield) within the season.
The best machine-learning models always outperformed simple benchmark models.
This was confirmed in low-yielding years, which is particularly relevant for
early warning. Nonetheless, the differences in accuracy between machine
learning and benchmark models were not always of practical significance.
Besides, the benchmark models outperformed up to 60% of the machine learning
models that were tested, which stresses the importance of proper model
calibration and selection. For crop yield forecasting, like for many
application domains, machine learning has delivered significant improvement in
predictive power. Nonetheless, superiority over simple benchmarks is often
fully achieved after extensive calibration, especially when dealing with small
data.
| stat |
Bayesian Analysis of Social Influence | The network influence model is a model for binary outcome variables that
accounts for dependencies between outcomes for units that are relationally
tied. The basic influence model was previously extended to afford a suite of
new dependence assumptions and because of its relation to traditional Markov
random field models it is often referred to as the auto logistic
actor-attribute model (ALAAM). We extend on current approaches for fitting
ALAAMs by presenting a comprehensive Bayesian inference scheme that supports
testing of dependencies across subsets of data and the presence of missing
data. We illustrate different aspects of the procedures through three empirical
examples: masculinity attitudes in an all-male Australian school class,
educational progression in Swedish schools, and un-employment among adults in a
community sample in Australia.
| 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.