title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Latent variable modeling with random features | Gaussian process-based latent variable models are flexible and theoretically
grounded tools for nonlinear dimension reduction, but generalizing to
non-Gaussian data likelihoods within this nonlinear framework is statistically
challenging. Here, we use random features to develop a family of nonlinear
dimension reduction models that are easily extensible to non-Gaussian data
likelihoods; we call these random feature latent variable models (RFLVMs). By
approximating a nonlinear relationship between the latent space and the
observations with a function that is linear with respect to random features, we
induce closed-form gradients of the posterior distribution with respect to the
latent variable. This allows the RFLVM framework to support computationally
tractable nonlinear latent variable models for a variety of data likelihoods in
the exponential family without specialized derivations. Our generalized RFLVMs
produce results comparable with other state-of-the-art dimension reduction
methods on diverse types of data, including neural spike train recordings,
images, and text data.
| stat |
Complementary-Label Learning for Arbitrary Losses and Models | In contrast to the standard classification paradigm where the true class is
given to each training pattern, complementary-label learning only uses training
patterns each equipped with a complementary label, which only specifies one of
the classes that the pattern does not belong to. The goal of this paper is to
derive a novel framework of complementary-label learning with an unbiased
estimator of the classification risk, for arbitrary losses and models---all
existing methods have failed to achieve this goal. Not only is this beneficial
for the learning stage, it also makes model/hyper-parameter selection (through
cross-validation) possible without the need of any ordinarily labeled
validation data, while using any linear/non-linear models or convex/non-convex
loss functions. We further improve the risk estimator by a non-negative
correction and gradient ascent trick, and demonstrate its superiority through
experiments.
| stat |
Ensemble Kalman filter based Sequential Monte Carlo Sampler for
sequential Bayesian inference | Many real-world problems require one to estimate parameters of interest, in a
Bayesian framework, from data that are collected sequentially in time.
Conventional methods for sampling from posterior distributions, such as {Markov
Chain Monte Carlo} can not efficiently address such problems as they do not
take advantage of the data's sequential structure. To this end, sequential
methods which seek to update the posterior distribution whenever a new
collection of data become available are often used to solve these types of
problems. Two popular choices of sequential method are the Ensemble Kalman
filter (EnKF) and the sequential Monte Carlo sampler (SMCS). An advantage of
the SMCS method is that, unlike the EnKF method that only computes a Gaussian
approximation of the posterior distribution, SMCS can draw samples directly
from the posterior. Its performance, however, depends critically upon the
kernels that are used. In this work, we present a method that constructs the
kernels of SMCS using an EnKF formulation, and we demonstrate the performance
of the method with numerical examples.
| stat |
Exact and computationally efficient Bayesian inference for generalized
Markov modulated Poisson processes | Statistical modeling of point patterns is an important and common problem in
several areas. The Poisson process is the most common process used for this
purpose, in particular, its generalization that considers the intensity
function to be stochastic. This is called a Cox process and different choices
to model the dynamics of the intensity gives rise to a wide range of models. We
present a new class of unidimensional Cox process models in which the intensity
function assumes parametric functional forms that switch among them according
to a continuous-time Markov chain. A novel methodology is proposed to perform
exact Bayesian inference based on MCMC algorithms. The term exact refers to the
fact that no discrete time approximation is used and Monte Carlo error is the
only source of inaccuracy. The reliability of the algorithms depends on a
variety of specifications which are carefully addressed, resulting in a
computationally efficient (in terms of computing time) algorithm and enabling
its use with large data sets. Simulated and real examples are presented to
illustrate the efficiency and applicability of the proposed methodology. A
specific model to fit epidemic curves is proposed and used to analyze data from
Dengue Fever in Brazil and COVID-19 in some countries.
| stat |
A response to critiques of "The reproducibility of research and the
misinterpretation of p-values" | I proposed (8, 1, 3) that p values should be supplemented by an estimate of
the false positive risk (FPR). FPR was defined as the probability that, if you
claim that there is a real effect on the basis of p value from a single
unbiased experiment, that you will be mistaken and the result has occurred by
chance. This is a Bayesian quantity and that means that there is an infinitude
of ways to calculate it. My choice of a way to estimate FPR was, therefore,
arbitrary. I maintain that it is a reasonable way, and has the advantage of
being mathematically simpler than other proposals and easier to understand than
other methods. This might make it more easily accepted by users. As always, not
every statistician agrees. This paper is a response to a critique of my 2017
paper (1) by Arandjelovic (2)
| stat |
Loss Rate Forecasting Framework Based on Macroeconomic Changes:
Application to US Credit Card Industry | A major part of the balance sheets of the largest US banks consists of credit
card portfolios. Hence, managing the charge-off rates is a vital task for the
profitability of the credit card industry. Different macroeconomic conditions
affect individuals' behavior in paying down their debts. In this paper, we
propose an expert system for loss forecasting in the credit card industry using
macroeconomic indicators. We select the indicators based on a thorough review
of the literature and experts' opinions covering all aspects of the economy,
consumer, business, and government sectors. The state of the art machine
learning models are used to develop the proposed expert system framework. We
develop two versions of the forecasting expert system, which utilize different
approaches to select between the lags added to each indicator. Among 19
macroeconomic indicators that were used as the input, six were used in the
model with optimal lags, and seven indicators were selected by the model using
all lags. The features that were selected by each of these models covered all
three sectors of the economy. Using the charge-off data for the top 100 US
banks ranked by assets from the first quarter of 1985 to the second quarter of
2019, we achieve mean squared error values of 1.15E-03 and 1.04E-03 using the
model with optimal lags and the model with all lags, respectively. The proposed
expert system gives a holistic view of the economy to the practitioners in the
credit card industry and helps them to see the impact of different
macroeconomic conditions on their future loss.
| stat |
Measurement error induced by locational uncertainty when estimating
discrete choice models with a distance as a regressor | Spatial microeconometric studies typically suffer from various forms of
inaccuracies that are not present when dealing with the classical regional
spatial econometrics models. Among those, missing data, locational errors,
sampling without a formal sample design, measurement errors and misalignment
are the typical sources of inaccuracy that can affects the results in a spatial
microeconometric analysis. In this paper, we have examined the effects of
measurement error introduced in a logistic model by random geo-masking, when
distances are used as predictors. Extending the classical results on the
measurement error in a linear regression model, our MC experiment on hospital
choices showed that the higher the distortion produced by the geo-masking, the
higher is the downward bias in absolute value towards zero of the coefficient
associated to the distance in a regression model.
| stat |
Conditional independence testing: a predictive perspective | Conditional independence testing is a key problem required by many machine
learning and statistics tools. In particular, it is one way of evaluating the
usefulness of some features on a supervised prediction problem. We propose a
novel conditional independence test in a predictive setting, and show that it
achieves better power than competing approaches in several settings. Our
approach consists in deriving a p-value using a permutation test where the
predictive power using the unpermuted dataset is compared with the predictive
power of using dataset where the feature(s) of interest are permuted. We
conclude that the method achives sensible results on simulated and real
datasets.
| stat |
A Unified Computational and Theoretical Framework for High-Dimensional
Bayesian Additive Models | We introduce a general framework for estimation and variable selection in
high-dimensional Bayesian generalized additive models where the response
belongs to the overdispersed exponential family. Our framework subsumes popular
models such as binomial regression, Poisson regression, Gaussian regression,
and negative binomial regression, and encompasses both canonical and
non-canonical link functions. Our method can be implemented with a highly
efficient EM algorithm, allowing us to rapidly attain estimates of the
significant functions while thresholding out insignificant ones. Under mild
regularity conditions, we establish posterior contraction rates and model
selection consistency when the number of covariates grows at nearly exponential
rate with sample size. We illustrate our method on both synthetic and real data
sets.
| stat |
Bayesian Inference with the l1-ball Prior: Solving Combinatorial
Problems with Exact Zeros | The l1-regularization is very popular in high dimensional statistics -- it
changes a combinatorial problem of choosing which subset of the parameter are
zero, into a simple continuous optimization. Using a continuous prior
concentrated near zero, the Bayesian counterparts are successful in quantifying
the uncertainty in the variable selection problems; nevertheless, the lack of
exact zeros makes it difficult for broader problems such as the change-point
detection and rank selection. Inspired by the duality of the l1-regularization
as a constraint onto an l1-ball, we propose a new prior by projecting a
continuous distribution onto the l1-ball. This creates a positive probability
on the ball boundary, which contains both continuous elements and exact zeros.
Unlike the spike-and-slab prior, this l1-ball projection is continuous and
differentiable almost surely, making the posterior estimation amenable to the
Hamiltonian Monte Carlo algorithm. We examine the properties, such as the
volume change due to the projection, the connection to the combinatorial prior,
the minimax concentration rate in the linear problem. We demonstrate the
usefulness of exact zeros that simplify the combinatorial problems, such as the
change-point detection in time series, the dimension selection of mixture model
and the low-rank-plus-sparse change detection in the medical images.
| stat |
High-dimensional macroeconomic forecasting using message passing
algorithms | This paper proposes two distinct contributions to econometric analysis of
large information sets and structural instabilities. First, it treats a
regression model with time-varying coefficients, stochastic volatility and
exogenous predictors, as an equivalent high-dimensional static regression
problem with thousands of covariates. Inference in this specification proceeds
using Bayesian hierarchical priors that shrink the high-dimensional vector of
coefficients either towards zero or time-invariance. Second, it introduces the
frameworks of factor graphs and message passing as a means of designing
efficient Bayesian estimation algorithms. In particular, a Generalized
Approximate Message Passing (GAMP) algorithm is derived that has low
algorithmic complexity and is trivially parallelizable. The result is a
comprehensive methodology that can be used to estimate time-varying parameter
regressions with arbitrarily large number of exogenous predictors. In a
forecasting exercise for U.S. price inflation this methodology is shown to work
very well.
| stat |
Improve Orthogonal GARCH with Hidden Markov Model | Orthogonal Generalized Autoregressive Conditional Heteroskedasticity model
(OGARCH) is widely used in finance industry to produce volatility and
correlation forecasts. We show that the classic OGARCH model, nevertheless,
tends to be too slow in reflecting sudden changes in market condition due to
excessive persistence of the integral univariate GARCH processes. To obtain
more flexibility to accommodate abrupt market changes, e.g. financial crisis,
we extend classic OGARCH model by incorporating a two-state Markov
regime-switching GARCH process. This novel construction allows us to capture
recurrent systemic regime shifts. Empirical results show that this
generalization resolves the problem of excessive persistency effectively and
greatly enhances OGARCH's ability to adapt to sudden market breaks while
preserving OGARCH's most attractive features such as dimension reduction and
multi-step ahead forecasting. By constructing a global minimum variance
portfolio (GMVP), we are able to demonstrate significant outperformance of the
extended model over the classic OGARCH model and the commonly used
Exponentially Weighted Moving Average (EWMA) model. In addition, we show that
the extended model is superior to OGARCH and EWMA in terms of predictive
accuracy.
| stat |
Duality between Approximate Bayesian Methods and Prior Robustness | In this paper we show that there is a link between approximate Bayesian
methods and prior robustness. We show that what is typically recognized as an
approximation to the likelihood, either due to the simulated data as in the
Approximate Bayesian Computation (ABC) methods or due to the functional
approximation to the likelihood, can instead also be viewed upon as an implicit
exercise in prior robustness. We first define two new classes of priors for the
cases where the sufficient statistics is available, establish their
mathematical properties and show, for a simple illustrative example, that these
classes of priors can also be used to obtain the posterior distribution that
would be obtained by implementing ABC. We then generalize and define two
further classes of priors that are applicable in very general scenarios; one
where the sufficient statistics is not available and another where the
likelihood is approximated using a functional approximation. We then discuss
the interpretation and elicitation aspects of the classes proposed here as well
as their potential applications and possible computational benefits. These
classes establish the duality between approximate Bayesian inference and prior
robustness for a wide category of Bayesian inference methods.
| stat |
Optimisation des parcours patients pour lutter contre l'errance de
diagnostic des patients atteints de maladies rares | A patient suffering from a rare disease in France has to wait an average of
two years before being diagnosed. This medical wandering is highly detrimental
both for the health system and for patients whose pathology may worsen. There
exists an efficient network of Centres of Reference for Rare Diseases (CRMR),
but patients are often referred to these structures too late. We are
considering a probabilistic modelling of the patient pathway in order to create
a simulator that will allow us to create an alert system that detects wandering
patients and refers them to a CRMR while considering the potential additional
costs associated with these decisions.
| stat |
Selecting the Metric in Hamiltonian Monte Carlo | We present a selection criterion for the Euclidean metric adapted during
warmup in a Hamiltonian Monte Carlo sampler that makes it possible for a
sampler to automatically pick the metric based on the model and the
availability of warmup draws. Additionally, we present a new adaptation
inspired by the selection criterion that requires significantly fewer warmup
draws to be effective. The effectiveness of the selection criterion and
adaptation are demonstrated on a number of applied problems. An implementation
for the Stan probabilistic programming language is provided.
| stat |
A Convergence Diagnostic for Bayesian Clustering | In many applications of Bayesian clustering, posterior sampling on the
discrete state space of cluster allocations is achieved via Markov chain Monte
Carlo (MCMC) techniques. As it is typically challenging to design transition
kernels to explore this state space efficiently, MCMC convergence diagnostics
for clustering applications is especially important. For general MCMC problems,
state-of-the-art convergence diagnostics involve comparisons across multiple
chains. However, single-chain alternatives can be appealing for computationally
intensive and slowly-mixing MCMC, as is typically the case for Bayesian
clustering. Thus, we propose here a single-chain convergence diagnostic
specifically tailored to discrete-space MCMC. Namely, we consider a
Hotelling-type statistic on the highest probability states, and use
regenerative sampling theory to derive its equilibrium distribution. By
leveraging information from the unnormalized posterior, our diagnostic protects
against seemingly convergent chains in which the relative frequency of visited
states is incorrect. The methodology is illustrated with a Bayesian clustering
analysis of genetic mutants of the flowering plant Arabidopsis thaliana.
| stat |
COVID-19 Policy Impact Evaluation: A guide to common design issues | Policy responses to COVID-19, particularly those related to
non-pharmaceutical interventions, are unprecedented in scale and scope.
Epidemiologists are more involved in policy decisions and evidence generation
than ever before. However, policy impact evaluations always require a complex
combination of circumstance, study design, data, statistics, and analysis.
Beyond the issues that are faced for any policy, evaluation of COVID-19
policies is complicated by additional challenges related to infectious disease
dynamics and lags, lack of direct observation of key outcomes, and a
multiplicity of interventions occurring on an accelerated time scale. The
methods needed for policy-level impact evaluation are not often used or taught
in epidemiology, and differ in important ways that may not be obvious. The
volume and speed, and methodological complications of policy evaluations can
make it difficult for decision-makers and researchers to synthesize and
evaluate strength of evidence in COVID-19 health policy papers.
In this paper, we (1) introduce the basic suite of policy impact evaluation
designs for observational data, including cross-sectional analyses, pre/post,
interrupted time-series, and difference-in-differences analysis, (2)
demonstrate key ways in which the requirements and assumptions underlying these
designs are often violated in the context of COVID-19, and (3) provide
decision-makers and reviewers a conceptual and graphical guide to identifying
these key violations. The overall goal of this paper is to help
epidemiologists, policy-makers, journal editors, journalists, researchers, and
other research consumers understand and weigh the strengths and limitations of
evidence that is essential to decision-making.
| stat |
Using activation histograms to bound the number of affine regions in
ReLU feed-forward neural networks | Several current bounds on the maximal number of affine regions of a ReLU
feed-forward neural network are special cases of the framework [1] which relies
on layer-wise activation histogram bounds. We analyze and partially solve a
problem in algebraic topology the solution of which would fully exploit this
framework. Our partial solution already induces slightly tighter bounds and
suggests insight in how parameter initialization methods can affect the number
of regions. Furthermore, we extend the framework to allow the composition of
subnetwork instead of layer-wise activation histogram bounds to reduce the
number of required compositions which negatively affect the tightness of the
resulting bound.
| stat |
Post hoc false discovery proportion inference under a Hidden Markov
Model | We address the multiple testing problem under the assumption that the
true/false hypotheses are driven by a Hidden Markov Model (HMM), which is
recognized as a fundamental setting to model multiple testing under dependence
since the seminal work of \citet{sun2009large}. While previous work has
concentrated on deriving specific procedures with a controlled False Discovery
Rate (FDR) under this model, following a recent trend in selective inference,
we consider the problem of establishing confidence bounds on the false
discovery proportion (FDP), for a user-selected set of hypotheses that can
depend on the observed data in an arbitrary way. We develop a methodology to
construct such confidence bounds first when the HMM model is known, then when
its parameters are unknown and estimated, including the data distribution under
the null and the alternative, using a nonparametric approach. In the latter
case, we propose a bootstrap-based methodology to take into account the effect
of parameter estimation error. We show that taking advantage of the assumed HMM
structure allows for a substantial improvement of confidence bound sharpness
over existing agnostic (structure-free) methods, as witnessed both via
numerical experiments and real data examples.
| stat |
A unified treatment of multiple testing with prior knowledge using the
p-filter | There is a significant literature on methods for incorporating knowledge into
multiple testing procedures so as to improve their power and precision. Some
common forms of prior knowledge include (a) beliefs about which hypotheses are
null, modeled by non-uniform prior weights; (b) differing importances of
hypotheses, modeled by differing penalties for false discoveries; (c) multiple
arbitrary partitions of the hypotheses into (possibly overlapping) groups; and
(d) knowledge of independence, positive or arbitrary dependence between
hypotheses or groups, suggesting the use of more aggressive or conservative
procedures. We present a unified algorithmic framework called p-filter for
global null testing and false discovery rate (FDR) control that allows the
scientist to incorporate all four types of prior knowledge (a)-(d)
simultaneously, recovering a variety of known algorithms as special cases.
| stat |
On the Non-Monotonicity of a Non-Differentially Mismeasured Binary
Confounder | Suppose that we are interested in the average causal effect of a binary
treatment on an outcome when this relationship is confounded by a binary
confounder. Suppose that the confounder is unobserved but a non-differential
binary proxy of it is observed. We identify conditions under which adjusting
for the proxy comes closer to the incomputable true average causal effect than
not adjusting at all. Unlike other works, we do not assume that the average
causal effect of the confounder on the outcome is in the same direction among
treated and untreated.
| stat |
COBI-GRINE: A Tool for Visualization and Advanced Evaluation of
Communities in Mass Channel Similarity Graphs | The detection of groups of molecules that co-localize with histopathological
patterns or sub-structures is an important step to combine the rich
high-dimensional content of mass spectrometry imaging (MSI) with classic
histopathological staining. Here we present the evolution of GRINE to
COBI-GRINE, an interactive web tool that maps MSI data onto a graph structure
to detect communities of laterally similar distributed molecules and
co-visualizes the communities with Hematoxylin and Eosin (HE) stained images.
Thereby the tool enables biologists and pathologists to examine the MSI image
graph in a target-oriented manner and links molecular co-localization to
pathology. Another feature is the manual optimization of cluster results with
the assist of graph statistics in order to improve the community results. As
the graphs can become very complex, those statistics provide good heuristics to
support and accelerate the detection of sub-clusters and misclusterings. This
kind of edited cluster optimization allows the integration of expert background
knowledge into the clustering result and a more precise analysis of links
between molecular co-localization and pathology.
| stat |
COVID-19 Clinical footprint to infer about mortality | Information of 1.6 million patients identified as SARS-CoV-2 positive in
Mexico is used to understand the relationship between comorbidities, symptoms,
hospitalizations and deaths due to the COVID-19 disease. Using the presence or
absence of these latter variables a clinical footprint for each patient is
created. The risk, expected mortality and the prediction of death outcomes,
among other relevant quantities, are obtained and analyzed by means of a
multivariate Bernoulli distribution. The proposal considers all possible
footprint combinations resulting in a robust model suitable for Bayesian
inference.
| stat |
Clustering of Arrivals in Queueing Systems: Autoregressive Conditional
Duration Approach | Arrivals in queueing systems are typically assumed to be independent and
exponentially distributed. Our analysis of an online bookshop, however, shows
that there is an autocorrelation structure present. First, we adjust the
inter-arrival times for diurnal and seasonal patterns. Second, we model
adjusted inter-arrival times by the generalized autoregressive score (GAS)
model based on the generalized gamma distribution in the spirit of the
autoregressive conditional duration (ACD) models. Third, in a simulation study,
we investigate the effects of the dynamic arrival model on the number of
customers, the busy period, and the response time in queueing systems with
single and multiple servers. We find that ignoring the autocorrelation
structure leads to significantly underestimated performance measures and
consequently suboptimal decisions. The proposed approach serves as a general
methodology for the treatment of arrivals clustering in practice.
| stat |
Scalable Statistical Inference of Photometric Redshift via Data
Subsampling | Handling big data has largely been a major bottleneck in traditional
statistical models. Consequently, when accurate point prediction is the primary
target, machine learning models are often preferred over their statistical
counterparts for bigger problems. But full probabilistic statistical models
often outperform other models in quantifying uncertainties associated with
model predictions. We develop a data-driven statistical modeling framework that
combines the uncertainties from an ensemble of statistical models learned on
smaller subsets of data carefully chosen to account for imbalances in the input
space. We demonstrate this method on a photometric redshift estimation problem
in cosmology, which seeks to infer a distribution of the redshift -- the
stretching effect in observing the light of far-away galaxies -- given
multivariate color information observed for an object in the sky. Our proposed
method performs balanced partitioning, graph-based data subsampling across the
partitions, and training of an ensemble of Gaussian process models.
| stat |
Containerized Analyses Enable Interactive and Reproducible Statistics | In recent decades the analysis of data has become increasingly computational.
Correspondingly, this has changed how scientific and statistical work is
shared. For example, it is now commonplace for underlying analysis code and
data to be proffered alongside journal publications and conference talks.
Unfortunately, sharing code faces several challenges. First, it is often
difficult to take code from one computer and run it on another. Code
configuration, version, and dependency issues often make this challenging.
Secondly, even if the code runs, it is often hard to understand or interact
with the analysis. This makes it difficult to assess the code and its findings,
for example, in a peer review process. In this paper we advocate for two
practical approaches to help make sharing interactive and reproducible analyses
easy: (1) analysis containerization, a technology that fully encapsulates an
analysis, data, code and dependencies into a shareable format, and (2) code
notebooks, an accessible format for interacting with third-party analyses. We
will demonstrate that the combination of these two technologies is powerful and
that containerizing interactive code notebooks can help make it easy for
statisticians to share code, analyses, and ideas.
| stat |
Translation-invariant functional clustering on COVID-19 deaths adjusted
on population risk factors | The COVID-19 pandemic has taken the world by storm with its high infection
rate. Investigating its geographical disparities has paramount interest in
order to gauge its relationships with political decisions, economic indicators,
or mental health. This paper focuses on clustering the daily death rates
reported in several regions of Europe and the United States over eight months.
Several methods have been developed to cluster such functional data. However,
these methods are not translation-invariant and thus cannot handle different
times of arrivals of the disease, nor can they consider external covariates and
so are unable to adjust for the population risk factors of each region. We
propose a novel three-step clustering method to circumvent these issues. As a
first step, feature extraction is performed by translation-invariant wavelet
decomposition which permits to deal with the different onsets. As a second
step, single-index regression is used to neutralize disparities caused by
population risk factors. As a third step, a nonparametric mixture is fitted on
the regression residuals to achieve the region clustering. Supplementary
materials for this article, including a standardized description of the
materials available for reproducing the work, are available online.
| stat |
Bayesian Inference for Regression Copulas | We propose a new semi-parametric distributional regression smoother that is
based on a copula decomposition of the joint distribution of the vector of
response values. The copula is high-dimensional and constructed by inversion of
a pseudo regression, where the conditional mean and variance are
semi-parametric functions of covariates modeled using regularized basis
functions. By integrating out the basis coefficients, an implicit copula
process on the covariate space is obtained, which we call a `regression
copula'. We combine this with a non-parametric margin to define a copula model,
where the entire distribution - including the mean and variance - of the
response is a smooth semi-parametric function of the covariates. The copula is
estimated using both Hamiltonian Monte Carlo and variational Bayes; the latter
of which is scalable to high dimensions. Using real data examples and a
simulation study we illustrate the efficacy of these estimators and the copula
model. In a substantive example, we estimate the distribution of half-hourly
electricity spot prices as a function of demand and two time covariates using
radial bases and horseshoe regularization. The copula model produces
distributional estimates that are locally adaptive with respect to the
covariates, and predictions that are more accurate than those from benchmark
models.
| stat |
Gaussian Mixture Models for Stochastic Block Models with Non-Vanishing
Noise | Community detection tasks have received a lot of attention across statistics,
machine learning, and information theory with a large body of work
concentrating on theoretical guarantees for the stochastic block model. One
line of recent work has focused on modeling the spectral embedding of a network
using Gaussian mixture models (GMMs) in scaling regimes where the ability to
detect community memberships improves with the size of the network. However,
these regimes are not very realistic. This paper provides tractable methodology
motivated by new theoretical results for networks with non-vanishing noise. We
present a procedure for community detection using GMMs that incorporates
certain truncation and shrinkage effects that arise in the non-vanishing noise
regime. We provide empirical validation of this new representation using both
simulated and real-world data.
| stat |
Nonlinear quantile mixed models | In regression applications, the presence of nonlinearity and correlation
among observations offer computational challenges not only in traditional
settings such as least squares regression, but also (and especially) when the
objective function is non-smooth as in the case of quantile regression. In this
paper, we develop methods for the modeling and estimation of nonlinear
conditional quantile functions when data are clustered within two-level nested
designs. This work represents an extension of the linear quantile mixed models
of Geraci and Bottai (2014, Statistics and Computing). We develop a novel
algorithm which is a blend of a smoothing algorithm for quantile regression and
a second order Laplacian approximation for nonlinear mixed models. To assess
the proposed methods, we present a simulation study and two applications, one
in pharmacokinetics and one related to growth curve modeling in agriculture.
| stat |
Dissecting the statistical properties of the Linear Extrapolation Method
of determining protein stability | When protein stability is measured by denaturant induced unfolding the linear
extrapolation method is usually used to analyse the data. This method is based
on the observation that the change in Gibbs free energy associated with
unfolding, $\Delta_rG$, is often found to be a linear function of the
denaturant concentration, $D$. The free energy change of unfolding in the
absence of denaturant, $\Delta_rG_0$, is estimated by extrapolation from this
linear relationship. Data analysis is generally done by nonlinear least-squares
regression to obtain estimates of the parameters as well as confidence
intervals. We have compared different methods for calculating confidence
intervals of the parameters and found that a simple method based on linear
theory gives as good, if not better, results than more advanced methods. We
have also compared three different parameterizations of the linear
extrapolation method and show that one of the forms, $\Delta_rG(D) =
\Delta_rG_0 - mD$, is problematic since the value of $\Delta_rG_0$ and that of
the $m$-value are correlated in the nonlinear least-squares analysis. Parameter
correlation can in some cases cause problems in the estimation of
confidence-intervals and -regions and should be avoided when possible. Two
alternative parameterizations, $\Delta_rG(D) = -m(D-D_{50})$ and $\Delta_rG(D)
= \Delta_rG_0(1-D/D_{50})$, where $D_{50}$ is the midpoint of the transition
region show much less correlation between parameters.
| stat |
Model-Assisted Uniformly Honest Inference for Optimal Treatment Regimes
in High Dimension | This paper develops new tools to quantify uncertainty in optimal decision
making and to gain insight into which variables one should collect information
about given the potential cost of measuring a large number of variables. We
investigate simultaneous inference to determine if a group of variables is
relevant for estimating an optimal decision rule in a high-dimensional
semiparametric framework. The unknown link function permits flexible modeling
of the interactions between the treatment and the covariates, but leads to
nonconvex estimation in high dimension and imposes significant challenges for
inference. We first establish that a local restricted strong convexity
condition holds with high probability and that any feasible local sparse
solution of the estimation problem can achieve the near-oracle estimation error
bound. We further rigorously verify that a wild bootstrap procedure based on a
debiased version of the local solution can provide asymptotically honest
uniform inference for the effect of a group of variables on optimal decision
making. The advantage of honest inference is that it does not require the
initial estimator to achieve perfect model selection and does not require the
zero and nonzero effects to be well-separated. We also propose an efficient
algorithm for estimation. Our simulations suggest satisfactory performance. An
example from a diabetes study illustrates the real application.
| stat |
Resampled Priors for Variational Autoencoders | We propose Learned Accept/Reject Sampling (LARS), a method for constructing
richer priors using rejection sampling with a learned acceptance function. This
work is motivated by recent analyses of the VAE objective, which pointed out
that commonly used simple priors can lead to underfitting. As the distribution
induced by LARS involves an intractable normalizing constant, we show how to
estimate it and its gradients efficiently. We demonstrate that LARS priors
improve VAE performance on several standard datasets both when they are learned
jointly with the rest of the model and when they are fitted to a pretrained
model. Finally, we show that LARS can be combined with existing methods for
defining flexible priors for an additional boost in performance.
| stat |
Neural Likelihoods via Cumulative Distribution Functions | We leverage neural networks as universal approximators of monotonic functions
to build a parameterization of conditional cumulative distribution functions
(CDFs). By the application of automatic differentiation with respect to
response variables and then to parameters of this CDF representation, we are
able to build black box CDF and density estimators. A suite of families is
introduced as alternative constructions for the multivariate case. At one
extreme, the simplest construction is a competitive density estimator against
state-of-the-art deep learning methods, although it does not provide an easily
computable representation of multivariate CDFs. At the other extreme, we have a
flexible construction from which multivariate CDF evaluations and
marginalizations can be obtained by a simple forward pass in a deep neural net,
but where the computation of the likelihood scales exponentially with
dimensionality. Alternatives in between the extremes are discussed. We evaluate
the different representations empirically on a variety of tasks involving tail
area probabilities, tail dependence and (partial) density estimation.
| stat |
Generation of new exciting regressors for consistent on-line estimation
for a scalar parameter | In this paper the problem of estimation of a single parameter from a linear
regression equation in the absence of sufficient excitation in the regressor is
addressed. A novel procedure to generate a new exciting regressor is proposed.
The superior performance of a classical gradient estimator using this new
regressor, instead of the original one, is illustrated with comprehensive
simulations.
| stat |
Inference in the stochastic Cox-Ingersol-Ross diffusion process with
continuous sampling: Computational aspects and simulation | In this paper, we consider a stochastic model based on the Cox- Ingersoll-
Ross model (CIR). The stochastic model is parameterized analytically by
applying It\^o's calculus and the trend functions of the proposed process is
calculated. The parameter estimators are then derived by means of two
procedures: the first is used to estimate the parameters in the drift
coefficient by the maximum likelihood (ML) method, based on continuous
sampling, and the second procedure approximates the diffusion coefficient by
two methods. Finally, a simulation of the process is presented. Thus, a typical
simulated trajectory of the process and its estimators is obtained.
| stat |
Random Matrix Based Extended Target Tracking with Orientation: A New
Model and Inference | In this study, we propose a novel extended target tracking algorithm which is
capable of representing the extent of dynamic objects as an ellipsoid with a
time-varying orientation angle. A diagonal positive semi-definite matrix is
defined to model objects' extent within the random matrix framework where the
diagonal elements have inverse-Gamma priors. The resulting measurement equation
is non-linear in the state variables, and it is not possible to find a
closed-form analytical expression for the true posterior because of the absence
of conjugacy. We use the variational Bayes technique to perform approximate
inference, where the Kullback-Leibler divergence between the true and the
approximate posterior is minimized by performing fixed-point iterations. The
update equations are easy to implement, and the algorithm can be used in
real-time tracking applications. We illustrate the performance of the method in
simulations and experiments with real data. The proposed method outperforms the
state-of-the-art methods when compared with respect to accuracy and robustness.
| stat |
Principal Regression for High Dimensional Covariance Matrices | This manuscript presents an approach to perform generalized linear regression
with multiple high dimensional covariance matrices as the outcome. Model
parameters are proposed to be estimated by maximizing a pseudo-likelihood. When
the data are high dimensional, the normal likelihood function is ill-posed as
the sample covariance matrix is rank-deficient. Thus, a well-conditioned linear
shrinkage estimator of the covariance matrix is introduced. With multiple
covariance matrices, the shrinkage coefficients are proposed to be common
across matrices. Theoretical studies demonstrate that the proposed covariance
matrix estimator is optimal achieving the uniformly minimum quadratic loss
asymptotically among all linear combinations of the identity matrix and the
sample covariance matrix. Under regularity conditions, the proposed estimator
of the model parameters is consistent. The superior performance of the proposed
approach over existing methods is illustrated through simulation studies.
Implemented to a resting-state functional magnetic resonance imaging study
acquired from the Alzheimer's Disease Neuroimaging Initiative, the proposed
approach identified a brain network within which functional connectivity is
significantly associated with Apolipoprotein E $\varepsilon$4, a strong genetic
marker for Alzheimer's disease.
| stat |
Bipartite mixed membership stochastic blockmodel | Mixed membership problem for undirected network has been well studied in
network analysis recent years. However, the more general case of mixed
membership for directed network remains a challenge. Here, we propose an
interpretable model: bipartite mixed membership stochastic blockmodel (BiMMSB
for short) for directed mixed membership networks. BiMMSB allows that row nodes
and column nodes of the adjacency matrix can be different and these nodes may
have distinct community structure in a directed network. We also develop an
efficient spectral algorithm called BiMPCA to estimate the mixed memberships
for both row nodes and column nodes in a directed network. We show that the
approach is asymptotically consistent under BiMMSB. We demonstrate the
advantages of BiMMSB with applications to a small-scale simulation study, the
directed Political blogs network and the Papers Citations network.
| stat |
Large-Scale Online Experimentation with Quantile Metrics | Online experimentation (or A/B testing) has been widely adopted in industry
as the gold standard for measuring product impacts. Despite the wide adoption,
few literatures discuss A/B testing with quantile metrics. Quantile metrics,
such as 90th percentile page load time, are crucial to A/B testing as many key
performance metrics including site speed and service latency are defined as
quantiles. However, with LinkedIn's data size, quantile metric A/B testing is
extremely challenging because there is no statistically valid and scalable
variance estimator for the quantile of dependent samples: the bootstrap
estimator is statistically valid, but takes days to compute; the standard
asymptotic variance estimate is scalable but results in order-of-magnitude
underestimation. In this paper, we present a statistically valid and scalable
methodology for A/B testing with quantiles that is fully generalizable to other
A/B testing platforms. It achieves over 500 times speed up compared to
bootstrap and has only $2\%$ chance to differ from bootstrap estimates. Beyond
methodology, we also share the implementation of a data pipeline using this
methodology and insights on pipeline optimization.
| stat |
Clustering Longitudinal Life-Course Sequences using Mixtures of
Exponential-Distance Models | Sequence analysis is an increasingly popular approach for the analysis of
life courses represented by an ordered collection of activities experienced by
subjects over a given time period. Several criteria exist for measuring
pairwise dissimilarities among sequences. Typically, dissimilarity matrices are
employed as input to heuristic clustering algorithms, with the aim of
identifying the most relevant patterns in the data.
Here, we propose a model-based clustering approach for categorical sequence
data. The technique is applied to a survey data set containing information on
the career trajectories of a cohort of Northern Irish youths tracked between
the ages of 16 and 22.
Specifically, we develop a family of methods for clustering sequences
directly, based on mixtures of exponential-distance models, which we call
MEDseq. The use of the Hamming distance or weighted variants thereof as the
distance metrics permits closed-form expressions for the normalising constant,
thereby facilitating the development of an ECM algorithm for model fitting.
Additionally, MEDseq models allow the probability of component membership to
depend on fixed covariates. Sampling weights, which are often associated with
life-course data arising from surveys, are also accommodated. Simultaneously
including weights and covariates in the clustering process yields new insights
on the Northern Irish data.
| stat |
Isotonic regression for metallic microstructure data: estimation and
testing under order restrictions | Investigating the main determinants of the mechanical performance of metals
is not a simple task. Already known physical inspired qualitative relations
between 2D microstructure characteristics and 3D mechanical properties can act
as the starting point of the investigation. Isotonic regression allows to take
into account ordering relations and leads to more efficient and accurate
results when the underlying assumptions actually hold. The main goal in this
paper is to test order relations in a model inspired by a materials science
application. The statistical estimation procedure is described considering
three different scenarios according to the knowledge of the variances: known
variance ratio, completely unknown variances, variances under order
restrictions. New likelihood ratio tests are developed in the last two cases.
Both parametric and non-parametric bootstrap approaches are developed for
finding the distribution of the test statistics under the null hypothesis.
Finally an application on the relation between Geometrically Necessary
Dislocations and number of observed microstructure precipitations is shown.
| stat |
Assimilation of semi-qualitative sea ice thickness data with the EnKF-SQ | A newly introduced stochastic data assimilation method, the Ensemble Kalman
Filter Semi-Qualitative (EnKF-SQ) is applied to a realistic coupled ice-ocean
model of the Arctic, the TOPAZ4 configuration, in a twin experiment framework.
The method is shown to add value to range-limited thin ice thickness
measurements, as obtained from passive microwave remote sensing, with respect
to more trivial solutions like neglecting the out-of-range values or
assimilating climatology instead. Some known properties inherent to the EnKF-SQ
are evaluated: the tendency to draw the solution closer to the thickness
threshold, the skewness of the resulting analysis ensemble and the potential
appearance of outliers. The experiments show that none of these properties
prove deleterious in light of the other sub-optimal characters of the sea ice
data assimilation system used here (non-linearities, non-Gaussian variables,
lack of strong coupling). The EnKF-SQ has a single tuning parameter that is
adjusted for best performance of the system at hand. The sensitivity tests
reveal that the results do not depend critically on the choice of this tuning
parameter. The EnKF-SQ makes overall a valid approach for assimilating
semi-qualitative observations into high-dimensional nonlinear systems.
| stat |
Forecasting Fertility with Parametric Mixture Models | This paper sets out a forecasting method that employs a mixture of parametric
functions to capture the pattern of fertility with respect to age. The overall
level of cohort fertility is decomposed over the range of fertile ages using a
mixture of parametric density functions. The level of fertility and the
parameters describing the shape of the fertility curve are projected foward
using time series methods. The model is estimated within a Bayesian framework,
allowing predictive distributions of future fertility rates to be produced that
naturally incorporate both time series and parametric uncertainty. A number of
choices are possible for the precise form of the functions used in the
two-component mixtures. The performance of several model variants is tested on
data from four countries; England and Wales, the USA, Sweden and France. The
former two countries exhibit multi-modality in their fertility rate curves as a
function of age, while the latter two are largely uni-modal. The models are
estimated using Hamiltonian Monte Carlo and the `stan` software package on data
covering the period up to 2006, with the period 2007-2016 held back for
assessment purposes. Forecasting performance is found to be comparable to other
models identified as producing accurate fertility forecasts in the literature.
| stat |
Flexible Mixture Modeling on Constrained Spaces | This paper addresses challenges in flexibly modeling multimodal data that lie
on constrained spaces. Such data are commonly found in spatial applications,
such as climatology and criminology, where measurements are restricted to a
geographical area. Other settings include domains where unsuitable recordings
are discarded, such as flow-cytometry measurements. A simple approach to
modeling such data is through the use of mixture models, especially
nonparametric mixture models. Mixture models, while flexible and theoretically
well-understood, are unsuitable for settings involving complicated constraints,
leading to difficulties in specifying the component distributions and in
evaluating normalization constants. Bayesian inference over the parameters of
these models results in posterior distributions that are doubly-intractable. We
address this problem via an algorithm based on rejection sampling and data
augmentation. We view samples from a truncated distribution as outcomes of a
rejection sampling scheme, where proposals are made from a simple mixture model
and are rejected if they violate the constraints. Our scheme proceeds by
imputing the rejected samples given mixture parameters and then resampling
parameters given all samples. We study two modeling approaches: mixtures of
truncated Gaussians and truncated mixtures of Gaussians, along with their
associated Markov chain Monte Carlo sampling algorithms. We also discuss
variations of the models, as well as approximations to improve mixing, reduce
computational cost, and lower variance. We present results on simulated data
and apply our algorithms to two problems; one involving flow-cytometry data,
and the other, crime recorded in the city of Chicago.
| stat |
Empirical Likelihood Test for Diagonal Symmetry | Energy distance is a statistical distance between the distributions of random
variables, which characterizes the equality of the distributions. Utilizing the
energy distance, we develop a nonparametric test for the diagonal symmetry,
which is consistent against any fixed alternatives. The test statistic
developed in this paper is based on the difference of two $U$-statistics. By
applying the jackknife empirical likelihood approach, the standard limiting
chi-square distribution with degree freedom of one is established and is used
to determine critical value and $p$-value of the test. Simulation studies show
that our method is competitive in terms of empirical sizes and empirical
powers.
| stat |
Peak Electricity Demand and Global Warming in the Industrial and
Residential areas of Pune : An Extreme Value Approach | Industrial and residential activities respond distinctly to electricity
demand on temperature. Due to increasing temperature trend on account of global
warming, its impact on peak electricity demand is a proxy for effective
management of electricity infrastructure. Few studies explore the relationship
between electricity demand and temperature changes in industrial areas in India
mainly due to the limitation of data. The precise role of industrial and
residential activities response to the temperature is not explored in
sub-tropical humid climate of India. Here, we show the temperature sensitivity
of industrial and residential areas in the city of Pune, Maharashtra by keeping
other influencing variables on electricity demand as constant. The study seeks
to estimate the behaviour of peak electricity demand with the apparent
temperature (AT) using the Extreme Value Theory. Our analysis shows that
industrial activities are not much influenced by the temperature whereas
residential activities show around 1.5-2% change in average electricity demand
with 1 degree rise in AT. Further, we show that peak electricity demand in
residential areas, performed using stationary and non-stationary GEV models,
are significantly influenced by the rise in temperature. The study shows that
with the improvement in data collection, better planning for the future
development, accounting for the climate change effects, will enhance the
effectiveness of electricity distribution system. The study is limited to the
geographical area of Pune. However, the methods are useful in estimating the
peak power load attributed to climate change to other geographical regions
located in subtropical and humid climate.
| stat |
Testing relevant hypotheses in functional time series via
self-normalization | In this paper we develop methodology for testing relevant hypotheses about
functional time series in a tuning-free way. Instead of testing for exact
equality, for example for the equality of two mean functions from two
independent time series, we propose to test the null hypothesis of no relevant
deviation. In the two sample problem this means that an $L^2$-distance between
the two mean functions is smaller than a pre-specified threshold. For such
hypotheses self-normalization, which was introduced by Shao (2010) and Shao and
Zhang (2010) and is commonly used to avoid the estimation of nuisance
parameters, is not directly applicable. We develop new self-normalized
procedures for testing relevant hypotheses in the one sample, two sample and
change point problem and investigate their asymptotic properties. Finite sample
properties of the proposed tests are illustrated by means of a simulation study
and data examples. Our main focus is on functional time series, but extensions
to other settings are also briefly discussed.
| stat |
Importance Weighted Generative Networks | Deep generative networks can simulate from a complex target distribution, by
minimizing a loss with respect to samples from that distribution. However,
often we do not have direct access to our target distribution - our data may be
subject to sample selection bias, or may be from a different but related
distribution. We present methods based on importance weighting that can
estimate the loss with respect to a target distribution, even if we cannot
access that distribution directly, in a variety of settings. These estimators,
which differentially weight the contribution of data to the loss function,
offer both theoretical guarantees and impressive empirical performance.
| stat |
MREC: a fast and versatile framework for aligning and matching point
clouds with applications to single cell molecular data | Comparing and aligning large datasets is a pervasive problem occurring across
many different knowledge domains. We introduce and study MREC, a recursive
decomposition algorithm for computing matchings between data sets. The basic
idea is to partition the data, match the partitions, and then recursively match
the points within each pair of identified partitions. The matching itself is
done using black box matching procedures that are too expensive to run on the
entire data set. Using an absolute measure of the quality of a matching, the
framework supports optimization over parameters including partitioning
procedures and matching algorithms. By design, MREC can be applied to extremely
large data sets. We analyze the procedure to describe when we can expect it to
work well and demonstrate its flexibility and power by applying it to a number
of alignment problems arising in the analysis of single cell molecular data.
| stat |
Temporal Wasserstein non-negative matrix factorization for non-rigid
motion segmentation and spatiotemporal deconvolution | Motion segmentation for natural images commonly relies on dense optic flow to
yield point trajectories which can be grouped into clusters through various
means including spectral clustering or minimum cost multicuts. However, in
biological imaging scenarios, such as fluorescence microscopy or calcium
imaging, where the signal to noise ratio is compromised and intensity
fluctuations occur, optical flow may be difficult to approximate. To this end,
we propose an alternative paradigm for motion segmentation based on optimal
transport which models the video frames as time-varying mass represented as
histograms. Thus, we cast motion segmentation as a temporal non-linear matrix
factorization problem with Wasserstein metric loss. The dictionary elements of
this factorization yield segmentation of motion into coherent objects while the
loading coefficients allow for time-varying intensity signal of the moving
objects to be captured. We demonstrate the use of the proposed paradigm on a
simulated multielectrode drift scenario, as well as calcium indicating
fluorescence microscopy videos of the nematode Caenorhabditis elegans (C.
elegans). The latter application has the added utility of extracting neural
activity of the animal in freely conducted behavior.
| stat |
Robust Factor Number Specification for Large-dimensional Elliptical
Factor Model | The accurate specification of the number of factors is critical to the
validity of factor models and the topic almost occupies the central position in
factor analysis. Plenty of estimators are available under the restrictive
condition that the fourth moments of the factors and idiosyncratic errors are
bounded. In this paper we propose efficient and robust estimators for the
factor number via considering a more general static Elliptical Factor Model
(EFM) framework. We innovatively propose to exploit the multivariate Kendall's
tau matrix, which captures the correlation structure of elliptical random
vectors. Theoretically we show that the proposed estimators are consistent
without exerting any moment condition when both cross-sections N and time
dimensions T go to infinity. Simulation study shows that the new estimators
perform much better in heavy-tailed data setting while performing comparably
with the state-of-the-art methods in the light-tailed Gaussian setting. At
last, a real macroeconomic data example is given to illustrate its empirical
advantages and usefulness.
| stat |
Explicit agreement extremes for a $2\times2$ table with given marginals | The problem of maximizing (or minimizing) the agreement between clusterings,
subject to given marginals, can be formally posed under a common framework for
several agreement measures. Until now, it was possible to find its solution
only through numerical algorithms. Here, an explicit solution is shown for the
case where the two clusterings have two clusters each.
| stat |
On Some Integral Means | Harmonic, Geometric, Arithmetic, Heronian and Contraharmonic means have been
studied by many mathematicians. In 2003, H. Evens studied these means from
geometrical point of view and established some of the inequalities between them
in using a circle and its radius. In 1961, E. Beckenback and R. Bellman
introduced several inequalities corresponding to means. In this paper, we will
introduce the concept of mean functions and integral means and give bounds on
some of these mean functions and integral means.
| stat |
Model Testing for Generalized Scalar-on-Function Linear Models | Scalar-on-function linear models are commonly used to regress functional
predictors on a scalar response. However, functional models are more difficult
to estimate and interpret than traditional linear models, and may be
unnecessarily complex for a data application. Hypothesis testing can be used to
guide model selection by determining if a functional predictor is necessary.
Using a mixed effects representation with penalized splines and variance
component tests, we propose a framework for testing functional linear models
with responses from exponential family distributions. The proposed method can
accommodate dense and sparse functional data, and be used to test functional
predictors for no effect and form of the effect. We show via simulation study
that the proposed method achieves the nominal level and has high power, and we
demonstrate its utility with two data applications.
| stat |
Does the winning team always covers the point spread? A study in
professional basketball since 1990 | In sports betting it is easier to predict the winner of a game match than the
team that covers the bet. Since, a winner team might not cover a bet.
This study focuses on the relation of the variable win to the betting
variable cover the point spread. The study is performed with data from
professional basketball (betting lines and scores) and tries to answer the
question: Does the winning team always covers the point spread?. In order to
answer this question, a regression analysis is performed taking into account
the most and less winning teams, together with their betting variables since
the 1990-1991 NBA season. The regression results are inserted in the SPXS
expert system revealing an indirect factor analysis that correlates betting
variables with teams winning percentages.
| stat |
Fast Stable Parameter Estimation for Linear Dynamical Systems | Dynamical systems describe the changes in processes that arise naturally from
their underlying physical principles, such as the laws of motion or the
conservation of mass, energy or momentum. These models facilitate a causal
explanation for the drivers and impediments of the processes. But do they
describe the behaviour of the observed data? And how can we quantify the
models' parameters that cannot be measured directly? This paper addresses these
two questions by providing a methodology for estimating the solution; and the
parameters of linear dynamical systems from incomplete and noisy observations
of the processes.
The proposed procedure builds on the parameter cascading approach, where a
linear combination of basis functions approximates the implicitly defined
solution of the dynamical system. The systems' parameters are then estimated so
that this approximating solution adheres to the data. By taking advantage of
the linearity of the system, we have simplified the parameter cascading
estimation procedure, and by developing a new iterative scheme, we achieve fast
and stable computation.
We illustrate our approach by obtaining a linear differential equation that
represents real data from biomechanics. Comparing our approach with popular
methods for estimating the parameters of linear dynamical systems, namely, the
non-linear least-squares approach, simulated annealing, parameter cascading and
smooth functional tempering reveals a considerable reduction in computation and
an improved bias and sampling variance.
| stat |
Mid-Epidemic Forecasts of COVID-19 Cases and Deaths: A Bivariate Model
applied to the UK | The evolution of the COVID-19 epidemic has been accompanied by accumulating
evidence on the underlying epidemiological parameters. Hence there is potential
for models providing mid-term forecasts of the epidemic trajectory using such
information. The effectiveness of lockdown interventions can also be assessed
by modelling later epidemic stages, possibly using a multiphase epidemic model.
Commonly applied methods to analyze epidemic trajectories include
phenomenological growth models (e.g. the Richards), and variants of the
susceptible-infected-recovered (SIR) compartment model. Here we focus on a
practical forecasting approach, applied to interim UK COVID data, using a
bivariate Reynolds model (cases and deaths). We show the utility of informative
priors in developing and estimating the model, and compare error densities
(Poisson-gamma, Poisson-lognormal, Poisson-logStudent) for overdispersed data
on new cases and deaths. We use cross-validation to assess medium term
forecasts. We also consider the longer term post-lockdown epidemic profile to
assess epidemic containment, using a two phase model. Fit to mid-epidemic data
shows better fit to training data and better cross validation performance for a
Poisson-logStudent model. Estimation of longer term epidemic data after
lockdown relaxation, characterised by protracted slow downturn and then upturn
in cases, casts doubt on effective containment. Many applications of
phenomenological models have been to complete epidemics. However, evaluation of
such models based simply on their fit to observed data may give only a partial
picture, and cross-validation against actual trends is also useful. Similarly,
it may be preferable to model incidence rather than cumulative data, though
this raises questions about suitable error densities for modelling often
erratic fluctuations. Hence there may be utility in evaluating alternative
error assumptions.
| stat |
A partition-based similarity for classification distributions | Herein we define a measure of similarity between classification distributions
that is both principled from the perspective of statistical pattern recognition
and useful from the perspective of machine learning practitioners. In
particular, we propose a novel similarity on classification distributions,
dubbed task similarity, that quantifies how an optimally-transformed optimal
representation for a source distribution performs when applied to inference
related to a target distribution. The definition of task similarity allows for
natural definitions of adversarial and orthogonal distributions. We highlight
limiting properties of representations induced by (universally) consistent
decision rules and demonstrate in simulation that an empirical estimate of task
similarity is a function of the decision rule deployed for inference. We
demonstrate that for a given target distribution, both transfer efficiency and
semantic similarity of candidate source distributions correlate with empirical
task similarity.
| stat |
Bayesian quantification for coherent anti-Stokes Raman scattering
spectroscopy | We propose a Bayesian statistical model for analyzing coherent anti-Stokes
Raman scattering (CARS) spectra. Our quantitative analysis includes statistical
estimation of constituent line-shape parameters, underlying Raman signal,
error-corrected CARS spectrum, and the measured CARS spectrum. As such, this
work enables extensive uncertainty quantification in the context of CARS
spectroscopy. Furthermore, we present an unsupervised method for improving
spectral resolution of Raman-like spectra requiring little to no \textit{a
priori} information. Finally, the recently-proposed wavelet prism method for
correcting the experimental artefacts in CARS is enhanced by using
interpolation techniques for wavelets. The method is validated using CARS
spectra of adenosine mono-, di-, and triphosphate in water, as well as,
equimolar aqueous solutions of D-fructose, D-glucose, and their disaccharide
combination sucrose.
| stat |
A class of two-sample nonparametric statistics for binary and
time-to-event outcomes | We propose a class of two-sample statistics for testing the equality of
proportions and the equality of survival functions. We build our proposal on a
weighted combination of a score test for the difference in proportions and a
Weighted Kaplan-Meier statistic-based test for the difference of survival
functions. The proposed statistics are fully non-parametric and do not rely on
the proportional hazards assumption for the survival outcome. We present the
asymptotic distribution of these statistics, propose a variance estimator and
show their asymptotic properties under fixed and local alternatives. We discuss
different choices of weights including those that control the relative
relevance of each outcome and emphasize the type of difference to be detected
in the survival outcome. We evaluate the performance of these statistics with a
simulation study, and illustrate their use with a randomized phase III cancer
vaccine trial. We have implemented the proposed statistics in the R package
SurvBin, available on GitHub (https://github.com/MartaBofillRoig/SurvBin).
| stat |
Noisy multi-label semi-supervised dimensionality reduction | Noisy labeled data represent a rich source of information that often are
easily accessible and cheap to obtain, but label noise might also have many
negative consequences if not accounted for. How to fully utilize noisy labels
has been studied extensively within the framework of standard supervised
machine learning over a period of several decades. However, very little
research has been conducted on solving the challenge posed by noisy labels in
non-standard settings. This includes situations where only a fraction of the
samples are labeled (semi-supervised) and each high-dimensional sample is
associated with multiple labels. In this work, we present a novel
semi-supervised and multi-label dimensionality reduction method that
effectively utilizes information from both noisy multi-labels and unlabeled
data. With the proposed Noisy multi-label semi-supervised dimensionality
reduction (NMLSDR) method, the noisy multi-labels are denoised and unlabeled
data are labeled simultaneously via a specially designed label propagation
algorithm. NMLSDR then learns a projection matrix for reducing the
dimensionality by maximizing the dependence between the enlarged and denoised
multi-label space and the features in the projected space. Extensive
experiments on synthetic data, benchmark datasets, as well as a real-world case
study, demonstrate the effectiveness of the proposed algorithm and show that it
outperforms state-of-the-art multi-label feature extraction algorithms.
| stat |
High-Dimensional Spatial Quantile Function-on-Scalar Regression | This paper develops a novel spatial quantile function-on-scalar regression
model, which studies the conditional spatial distribution of a high-dimensional
functional response given scalar predictors. With the strength of both quantile
regression and copula modeling, we are able to explicitly characterize the
conditional distribution of the functional or image response on the whole
spatial domain. Our method provides a comprehensive understanding of the effect
of scalar covariates on functional responses across different quantile levels
and also gives a practical way to generate new images for given covariate
values. Theoretically, we establish the minimax rates of convergence for
estimating coefficient functions under both fixed and random designs. We
further develop an efficient primal-dual algorithm to handle high-dimensional
image data. Simulations and real data analysis are conducted to examine the
finite-sample performance.
| stat |
A fast algorithm with minimax optimal guarantees for topic models with
an unknown number of topics | We propose a new method of estimation in topic models, that is not a
variation on the existing simplex finding algorithms, and that estimates the
number of topics K from the observed data. We derive new finite sample minimax
lower bounds for the estimation of A, as well as new upper bounds for our
proposed estimator. We describe the scenarios where our estimator is minimax
adaptive. Our finite sample analysis is valid for any number of documents (n),
individual document length (N_i), dictionary size (p) and number of topics (K),
and both p and K are allowed to increase with n, a situation not handled well
by previous analyses. We complement our theoretical results with a detailed
simulation study. We illustrate that the new algorithm is faster and more
accurate than the current ones, although we start out with a computational and
theoretical disadvantage of not knowing the correct number of topics K, while
we provide the competing methods with the correct value in our simulations.
| stat |
Re-weighting of Vector-weighted Mechanisms for Utility Maximization
under Differential Privacy | We present practical aspects of implementing a pseudo posterior synthesizer
for microdata dissemination under a new re-weighting strategy for utility
maximization. Our re-weighting strategy applies to any vector-weighting
approach under which a vector of observation-indexed weight are used to
downweight likelihood contributions for high disclosure risk records. We
demonstrate our method on two different vector-weighted schemes that target
high-risk records by exponentiating each of their likelihood contributions with
a record-indexed weight, $\alpha_i \in [0,1]$ for record $i \in (1,\ldots,n)$.
We compute the overall Lipschitz bound, $\Delta_{\bm{\alpha},\mathbf{x}}$, for
the database $\mathbf{x}$, under each vector-weighted scheme where a local
$\epsilon_{x} = 2\Delta_{\bm{\alpha},\mathbf{x}}$. Our new method for
constructing record-indexed downeighting maximizes the data utility under any
privacy budget for the vector-weighted synthesizers by adjusting the by-record
weights, $(\alpha_{i})_{i = 1}^{n}$, such that their individual Lipschitz
bounds, $\Delta_{\bm{\alpha},x_{i}}$, approach the bound for the entire
database, $\Delta_{\bm{\alpha},\mathbf{x}}$. We illustrate our methods using
simulated count data with and without over-dispersion-induced skewness and
compare the results to a scalar-weighted synthesizer under the Exponential
Mechanism (EM). We demonstrate our pDP result in a simulation study and our
methods on a sample of the Survey of Doctorate Recipients.
| stat |
Fast and frugal time series forecasting | Over the years, families of forecasting models, such as the exponential
smoothing family and Autoregressive Integrated Moving Average, have expanded to
contain multiple possible forms and forecasting profiles. In this paper, we
question the need to consider such large families of models. We argue that
parsimoniously identifying suitable subsets of models will not decrease the
forecasting accuracy nor will it reduce the ability to estimate the forecast
uncertainty. We propose a framework that balances forecasting performance
versus computational cost, resulting in a set of reduced families of models and
empirically demonstrate this trade-offs. We translate computational benefits to
monetary cost savings and discuss the implications of our results in the
context of large retailers.
| stat |
Estimation of the number of spiked eigenvalues in a covariance matrix by
bulk eigenvalue matching analysis | The spiked covariance model has gained increasing popularity in
high-dimensional data analysis. A fundamental problem is determination of the
number of spiked eigenvalues, $K$. For estimation of $K$, most attention has
focused on the use of $top$ eigenvalues of sample covariance matrix, and there
is little investigation into proper ways of utilizing $bulk$ eigenvalues to
estimate $K$. We propose a principled approach to incorporating bulk
eigenvalues in the estimation of $K$. Our method imposes a working model on the
residual covariance matrix, which is assumed to be a diagonal matrix whose
entries are drawn from a gamma distribution. Under this model, the bulk
eigenvalues are asymptotically close to the quantiles of a fixed parametric
distribution. This motivates us to propose a two-step method: the first step
uses bulk eigenvalues to estimate parameters of this distribution, and the
second step leverages these parameters to assist the estimation of $K$. The
resulting estimator $\hat{K}$ aggregates information in a large number of bulk
eigenvalues. We show the consistency of $\hat{K}$ under a standard spiked
covariance model. We also propose a confidence interval estimate for $K$. Our
extensive simulation studies show that the proposed method is robust and
outperforms the existing methods in a range of scenarios. We apply the proposed
method to analysis of a lung cancer microarray data set and the 1000 Genomes
data set.
| stat |
The rule of conditional probability is valid in quantum theory [Comment
on Gelman & Yao's "Holes in Bayesian Statistics"] | In a recent manuscript, Gelman & Yao (2020) claim that "the usual rules of
conditional probability fail in the quantum realm" and that "probability theory
isn't true (quantum physics)" and purport to support these statements with the
example of a quantum double-slit experiment. The present comment recalls some
relevant literature in quantum theory and shows that (i) Gelman & Yao's
statements are false; in fact, the quantum example confirms the rules of
probability theory; (ii) the particular inequality found in the quantum example
can be shown to appear also in very non-quantum examples, such as drawing from
an urn; thus there is nothing peculiar to quantum theory in this matter. A
couple of wrong or imprecise statements about quantum theory in the cited
manuscript are also corrected.
| stat |
Bayesian Posterior Interval Calibration to Improve the Interpretability
of Observational Studies | Observational healthcare data offer the potential to estimate causal effects
of medical products on a large scale. However, the confidence intervals and
p-values produced by observational studies only account for random error and
fail to account for systematic error. As a consequence, operating
characteristics such as confidence interval coverage and Type I error rates
often deviate sharply from their nominal values and render interpretation
impossible. While there is longstanding awareness of systematic error in
observational studies, analytic approaches to empirically account for
systematic error are relatively new. Several authors have proposed approaches
using negative controls (also known as "falsification hypotheses") and positive
controls. The basic idea is to adjust confidence intervals and p-values in
light of the bias (if any) detected in the analyses of the negative and
positive control. In this work, we propose a Bayesian statistical procedure for
posterior interval calibration that uses negative and positive controls. We
show that the posterior interval calibration procedure restores nominal
characteristics, such as 95% coverage of the true effect size by the 95%
posterior interval.
| stat |
Simulation-based Value-at-Risk for Nonlinear Portfolios | Value-at-risk (VaR) has been playing the role of a standard risk measure
since its introduction. In practice, the delta-normal approach is usually
adopted to approximate the VaR of portfolios with option positions. Its
effectiveness, however, substantially diminishes when the portfolios concerned
involve a high dimension of derivative positions with nonlinear payoffs; lack
of closed form pricing solution for these potentially highly correlated,
American-style derivatives further complicates the problem. This paper proposes
a generic simulation-based algorithm for VaR estimation that can be easily
applied to any existing procedures. Our proposal leverages cross-sectional
information and applies variable selection techniques to simplify the existing
simulation framework. Asymptotic properties of the new approach demonstrate
faster convergence due to the additional model selection component introduced.
We have also performed sets of numerical results that verify the effectiveness
of our approach in comparison with some existing strategies.
| stat |
Estimating crop yields with remote sensing and deep learning | Increasing the accuracy of crop yield estimates may allow improvements in the
whole crop production chain, allowing farmers to better plan for harvest, and
for insurers to better understand risks of production, to name a few
advantages. To perform their predictions, most current machine learning models
use NDVI data, which can be hard to use, due to the presence of clouds and
their shadows in acquired images, and due to the absence of reliable crop masks
for large areas, especially in developing countries. In this paper, we present
a deep learning model able to perform pre-season and in-season predictions for
five different crops. Our model uses crop calendars, easy-to-obtain remote
sensing data and weather forecast information to provide accurate yield
estimates.
| stat |
Tree Space Prototypes: Another Look at Making Tree Ensembles
Interpretable | Ensembles of decision trees perform well on many problems, but are not
interpretable. In contrast to existing approaches in interpretability that
focus on explaining relationships between features and predictions, we propose
an alternative approach to interpret tree ensemble classifiers by surfacing
representative points for each class -- prototypes. We introduce a new distance
for Gradient Boosted Tree models, and propose new, adaptive prototype selection
methods with theoretical guarantees, with the flexibility to choose a different
number of prototypes in each class. We demonstrate our methods on random
forests and gradient boosted trees, showing that the prototypes can perform as
well as or even better than the original tree ensemble when used as a
nearest-prototype classifier. In a user study, humans were better at predicting
the output of a tree ensemble classifier when using prototypes than when using
Shapley values, a popular feature attribution method. Hence, prototypes present
a viable alternative to feature-based explanations for tree ensembles.
| stat |
Confounding and Regression Adjustment in Difference-in-Differences | Difference-in-differences (diff-in-diff) is a study design that compares
outcomes of two groups (treated and comparison) at two time points (pre- and
post-treatment) and is widely used in evaluating new policy implementations.
For instance, diff-in-diff has been used to estimate the effect that increasing
minimum wage has on employment rates and to assess the Affordable Care Act's
effect on health outcomes. Although diff-in-diff appears simple, potential
pitfalls lurk. In this paper, we discuss one such complication: time-varying
confounding. We provide rigorous definitions for confounders in diff-in-diff
studies and explore regression strategies to adjust for confounding. In
simulations, we show how and when regression adjustment can ameliorate
confounding for both time-invariant and time-varying covariates. We compare our
regression approach to those models commonly fit in applied literature, which
often fail to address the time-varying nature of confounding in diff-in-diff.
| stat |
Predicting replicability -- analysis of survey and prediction market
data from large-scale forecasting projects | The reproducibility of published research has become an important topic in
science policy. A number of large-scale replication projects have been
conducted to gauge the overall reproducibility in specific academic fields.
Here, we present an analysis of data from four studies which sought to forecast
the outcomes of replication projects in the social and behavioural sciences,
using human experts who participated in prediction markets and answered
surveys. Because the number of findings replicated and predicted in each
individual study was small, pooling the data offers an opportunity to evaluate
hypotheses regarding the performance of prediction markets and surveys at a
higher power. In total, peer beliefs were elicited for the replication outcomes
of 103 published findings. We find there is information within the scientific
community about the replicability of scientific findings, and that both surveys
and prediction markets can be used to elicit and aggregate this information.
Our results show prediction markets can determine the outcomes of direct
replications with 73% accuracy (n=103). Both the prediction market prices and
the average survey responses are correlated with outcomes (0.581 and 0.564
respectively, both p < .001). We also found a significant relationship between
p-values of the original findings and replication outcomes. The dataset is made
available through the R package pooledmaRket and can be used to further study
community beliefs towards replications outcomes as elicited in the surveys and
prediction markets.
| stat |
Intelligent Surveillance of World Health Organization (WHO) Integrated
Disease Surveillance and Response (IDSR) Data in Cameroon Using Multivariate
Cross-Correlation | As developing countries continue to face challenges associated with
infectious diseases, the need to improve infrastructure to systematically
collect data which can be used to understand their outbreak patterns becomes
more critical. The World Health Organization (WHO) Integrated Disease
Surveillance and Response (IDSR) strategy seeks to drive the systematic
collection of surveillance data to strengthen district-level reporting and to
translate them into public health actions. Since the analysis of this
surveillance data at the central levels of government in many developing
nations has traditionally not included advanced analytics, there are
opportunities for the development and exploration of computational approaches
that can provide proactive insights and improve general health outcomes of
infectious disease outbreaks. We propose and demonstrate a multivariate time
series cross-correlation analysis as a foundational step towards gaining
insight on infectious disease patterns via the pairwise computation of weighted
cross-correlation scores for a specified disease across different health
districts using surveillance data from Cameroon. Following the computation of
weighted cross-correlation scores, we apply an anomaly detection algorithm to
assess how outbreak alarm patterns align in highly correlated health districts.
We demonstrate how multivariate cross-correlation analysis of weekly
surveillance data can provide insight into infectious disease incidence
patterns in Cameroon by identifying highly correlated health districts for a
given disease. We further demonstrate scenarios in which identification of
highly correlated districts aligns with alarms flagged using a standard anomaly
detection algorithm, hinting at the potential of end to end solutions combining
anomaly detection algorithms for flagging alarms in combination with
multivariate cross-correlation analysis.
| stat |
Graph cuts always find a global optimum (with a catch) | We prove that the alpha-expansion algorithm for MAP inference always returns
a globally optimal assignment for Markov Random Fields with Potts pairwise
potentials, with a catch: the returned assignment is only guaranteed to be
optimal in a small perturbation of the original problem instance. In other
words, all local minima with respect to expansion moves are global minima to
slightly perturbed versions of the problem. On "real-world" instances, MAP
assignments of small perturbations of the problem should be very similar to the
MAP assignment(s) of the original problem instance. We design an algorithm that
can certify whether this is the case in practice. On several MAP inference
problem instances from computer vision, this algorithm certifies that MAP
solutions to all of these perturbations are very close to solutions of the
original instance. These results taken together give a cohesive explanation for
the good performance of "graph cuts" algorithms in practice. Every local
expansion minimum is a global minimum in a small perturbation of the problem,
and all of these global minima are close to the original solution.
| stat |
Sparse matrix linear models for structured high-throughput data | Recent technological advancements have led to the rapid generation of
high-throughput biological data, which can be used to address novel scientific
questions in broad areas of research. These data can be thought of as a large
matrix with covariates annotating both rows and columns of this matrix. Matrix
linear models provide a convenient way for modeling such data. In many
situations, sparse estimation of these models is desired. We present fast,
general methods for fitting sparse matrix linear models to structured
high-throughput data. We induce model sparsity using an L$_1$ penalty and
consider the case when the response matrix and the covariate matrices are
large. Due to data size, standard methods for estimation of these penalized
regression models fail if the problem is converted to the corresponding
univariate regression scenario. By leveraging matrix properties in the
structure of our model, we develop several fast estimation algorithms
(coordinate descent, FISTA, and ADMM) and discuss their trade-offs. We evaluate
our method's performance on simulated data, E. coli chemical genetic screening
data, and two Arabidopsis genetic datasets with multivariate responses. Our
algorithms have been implemented in the Julia programming language and are
available at https://github.com/senresearch/MatrixLMnet.jl.
| stat |
Bayesian deep learning with hierarchical prior: Predictions from limited
and noisy data | Datasets in engineering applications are often limited and contaminated,
mainly due to unavoidable measurement noise and signal distortion. Thus, using
conventional data-driven approaches to build a reliable discriminative model,
and further applying this identified surrogate to uncertainty analysis remains
to be very challenging. A deep learning approach is presented to provide
predictions based on limited and noisy data. To address noise perturbation, the
Bayesian learning method that naturally facilitates an automatic updating
mechanism is considered to quantify and propagate model uncertainties into
predictive quantities. Specifically, hierarchical Bayesian modeling (HBM) is
first adopted to describe model uncertainties, which allows the prior
assumption to be less subjective, while also makes the proposed surrogate more
robust. Next, the Bayesian inference is seamlessly integrated into the DL
framework, which in turn supports probabilistic programming by yielding a
probability distribution of the quantities of interest rather than their point
estimates. Variational inference (VI) is implemented for the posterior
distribution analysis where the intractable marginalization of the likelihood
function over parameter space is framed in an optimization format, and
stochastic gradient descent method is applied to solve this optimization
problem. Finally, Monte Carlo simulation is used to obtain an unbiased
estimator in the predictive phase of Bayesian inference, where the proposed
Bayesian deep learning (BDL) scheme is able to offer confidence bounds for the
output estimation by analyzing propagated uncertainties. The effectiveness of
Bayesian shrinkage is demonstrated in improving predictive performance using
contaminated data, and various examples are provided to illustrate concepts,
methodologies, and algorithms of this proposed BDL modeling technique.
| stat |
Propensity Score Weighting Analysis of Survival Outcomes Using
Pseudo-observations | Survival outcomes are common in comparative effectiveness studies and require
unique handling because they are usually incompletely observed due to
right-censoring. A ``once for all'' approach for causal inference with survival
outcomes constructs pseudo-observations and allows standard methods such as
propensity score weighting to proceed as if the outcomes are completely
observed. We propose a general class of model-free causal estimands with
survival outcomes on user-specified target populations. We develop
corresponding propensity score weighting estimators based on the
pseudo-observations and establish their asymptotic properties. In particular,
utilizing the functional delta-method and the von Mises expansion, we derive a
new closed-form variance of the weighting estimator that takes into account the
uncertainty due to both pseudo-observation calculation and propensity score
estimation. This allows valid and computationally efficient inference without
resampling. We also prove the optimal efficiency property of the overlap
weights within the class of balancing weights for survival outcomes. The
proposed methods are applicable to both binary and multiple treatments.
Extensive simulations are conducted to explore the operating characteristics of
the proposed method versus other commonly used alternatives. We apply the
proposed method to compare the causal effects of three popular treatment
approaches for prostate cancer patients.
| stat |
Bias Correction of Learned Generative Models using Likelihood-Free
Importance Weighting | A learned generative model often produces biased statistics relative to the
underlying data distribution. A standard technique to correct this bias is
importance sampling, where samples from the model are weighted by the
likelihood ratio under model and true distributions. When the likelihood ratio
is unknown, it can be estimated by training a probabilistic classifier to
distinguish samples from the two distributions. We employ this likelihood-free
importance weighting method to correct for the bias in generative models. We
find that this technique consistently improves standard goodness-of-fit metrics
for evaluating the sample quality of state-of-the-art deep generative models,
suggesting reduced bias. Finally, we demonstrate its utility on representative
applications in a) data augmentation for classification using generative
adversarial networks, and b) model-based policy evaluation using off-policy
data.
| stat |
vtreat: a data.frame Processor for Predictive Modeling | We look at common problems found in data that is used for predictive modeling
tasks, and describe how to address them with the vtreat R package. vtreat
prepares real-world data for predictive modeling in a reproducible and
statistically sound manner. We describe the theory of preparing variables so
that data has fewer exceptional cases, making it easier to safely use models in
production. Common problems dealt with include: infinite values, invalid
values, NA, too many categorical levels, rare categorical levels, and new
categorical levels (levels seen during application, but not during training).
Of special interest are techniques needed to avoid needlessly introducing
undesirable nested modeling bias (which is a risk when using a
data-preprocessor).
| stat |
Causaltoolbox---Estimator Stability for Heterogeneous Treatment Effects | Estimating heterogeneous treatment effects has become increasingly important
in many fields and life and death decisions are now based on these estimates:
for example, selecting a personalized course of medical treatment. Recently, a
variety of procedures relying on different assumptions have been suggested for
estimating heterogeneous treatment effects. Unfortunately, there are no
compelling approaches that allow identification of the procedure that has
assumptions that hew closest to the process generating the data set under study
and researchers often select one arbitrarily. This approach risks making
inferences that rely on incorrect assumptions and gives the experimenter too
much scope for $p$-hacking. A single estimator will also tend to overlook
patterns other estimators could have picked up. We believe that the conclusion
of many published papers might change had a different estimator been chosen and
we suggest that practitioners should evaluate many estimators and assess their
similarity when investigating heterogeneous treatment effects. We demonstrate
this by applying 28 different estimation procedures to an emulated
observational data set; this analysis shows that different estimation
procedures may give starkly different estimates. We also provide an extensible
\texttt{R} package which makes it straightforward for practitioners to follow
our recommendations.
| stat |
Smoothing and Interpolating Noisy GPS Data with Smoothing Splines | A comprehensive methodology is provided for smoothing noisy, irregularly
sampled data with non-Gaussian noise using smoothing splines. We demonstrate
how the spline order and tension parameter can be chosen a priori from physical
reasoning. We also show how to allow for non-Gaussian noise and outliers which
are typical in GPS signals. We demonstrate the effectiveness of our methods on
GPS trajectory data obtained from oceanographic floating instruments known as
drifters.
| stat |
Using Statistics to Automate Stochastic Optimization | Despite the development of numerous adaptive optimizers, tuning the learning
rate of stochastic gradient methods remains a major roadblock to obtaining good
practical performance in machine learning. Rather than changing the learning
rate at each iteration, we propose an approach that automates the most common
hand-tuning heuristic: use a constant learning rate until "progress stops,"
then drop. We design an explicit statistical test that determines when the
dynamics of stochastic gradient descent reach a stationary distribution. This
test can be performed easily during training, and when it fires, we decrease
the learning rate by a constant multiplicative factor. Our experiments on
several deep learning tasks demonstrate that this statistical adaptive
stochastic approximation (SASA) method can automatically find good learning
rate schedules and match the performance of hand-tuned methods using default
settings of its parameters. The statistical testing helps to control the
variance of this procedure and improves its robustness.
| stat |
A Single Scalable LSTM Model for Short-Term Forecasting of Disaggregated
Electricity Loads | Most electricity systems worldwide are deploying advanced metering
infrastructures to collect relevant operational data. In particular, smart
meters allow tracking electricity load consumption at a very disaggregated
level and at high frequency rates. This data opens the possibility of
developing new forecasting models with a potential positive impact in
electricity systems. We present a general methodology that is able to process
and forecast a large number of smart meter time series. Instead of using
traditional and univariate approaches, we develop a single but complex
recurrent neural-network model with long short-term memory that can capture
individual consumption patterns and also consumptions from different
households. The resulting model can accurately predict future loads
(short-term) of individual consumers, even if these were not included in the
original training set. This entails a great potential for large scale
applications as once the single network is trained, accurate individual
forecast for new consumers can be obtained at almost no computational cost. The
proposed model is tested under a large set of numerical experiments by using a
real-world dataset with thousands of disaggregated electricity consumption time
series. Furthermore, we explore how geo-demographic segmentation of consumers
may impact the forecasting accuracy of the model.
| stat |
A joint bayesian space-time model to integrate spatially misaligned air
pollution data in R-INLA | In air pollution studies, dispersion models provide estimates of
concentration at grid level covering the entire spatial domain, and are then
calibrated against measurements from monitoring stations. However, these
different data sources are misaligned in space and time. If misalignment is not
considered, it can bias the predictions. We aim at demonstrating how the
combination of multiple data sources, such as dispersion model outputs, ground
observations and covariates, leads to more accurate predictions of air
pollution at grid level. We consider nitrogen dioxide (NO2) concentration in
Greater London and surroundings for the years 2007-2011, and combine two
different dispersion models. Different sets of spatial and temporal effects are
included in order to obtain the best predictive capability. Our proposed model
is framed in between calibration and Bayesian melding techniques for data
fusion red. Unlike other examples, we jointly model the response (concentration
level at monitoring stations) and the dispersion model outputs on different
scales, accounting for the different sources of uncertainty. Our
spatio-temporal model allows us to reconstruct the latent fields of each model
component, and to predict daily pollution concentrations. We compare the
predictive capability of our proposed model with other established methods to
account for misalignment (e.g. bilinear interpolation), showing that in our
case study the joint model is a better alternative.
| stat |
Iterative Bayesian-based Localization Mechanism for Industry Verticals | We propose and evaluate an iterative localization mechanism employing
Bayesian inference to estimate the position of a target using received signal
strength measurements. The probability density functions of the target's
coordinates are estimated through a Bayesian network. Herein, we consider an
iterative procedure whereby our predictor (posterior distribution) is updated
in a sequential order whenever new measurements are made available. The
performance of the mechanism is assessed in terms of the respective root mean
square error and kernel density estimation of the target coordinates. Our
numerical results showed the proposed iterative mechanism achieves increasingly
better estimation of the target node position each updating round of the
Bayesian network with new input measurements.
| stat |
Nonparametric Variable Screening with Optimal Decision Stumps | Decision trees and their ensembles are endowed with a rich set of diagnostic
tools for ranking and screening variables in a predictive model. Despite the
widespread use of tree based variable importance measures, pinning down their
theoretical properties has been challenging and therefore largely unexplored.
To address this gap between theory and practice, we derive finite sample
performance guarantees for variable selection in nonparametric models using a
single-level CART decision tree (a decision stump). Under standard operating
assumptions in variable screening literature, we find that the marginal signal
strength of each variable and ambient dimensionality can be considerably weaker
and higher, respectively, than state-of-the-art nonparametric variable
selection methods. Furthermore, unlike previous marginal screening methods that
attempt to directly estimate each marginal projection via a truncated basis
expansion, the fitted model used here is a simple, parsimonious decision stump,
thereby eliminating the need for tuning the number of basis terms. Thus,
surprisingly, even though decision stumps are highly inaccurate for estimation
purposes, they can still be used to perform consistent model selection.
| stat |
Simple Bayesian testing of scientific expectations in linear regression
models | Scientific theories can often be formulated using equality and order
constraints on the relative effects in a linear regression model. For example,
it may be expected that the effect of the first predictor is larger than the
effect of the second predictor, and the second predictor is expected to be
larger than the third predictor. The goal is then to test such expectations
against competing scientific expectations or theories. In this paper a simple
default Bayes factor test is proposed for testing multiple hypotheses with
equality and order constraints on the effects of interest. The proposed testing
criterion can be computed without requiring external prior information about
the expected effects before observing the data. The method is implemented in
R-package called `{\tt lmhyp}' which is freely downloadable and ready to use.
The usability of the method and software is illustrated using empirical
applications from the social and behavioral sciences.
| stat |
BasisVAE: Translation-invariant feature-level clustering with
Variational Autoencoders | Variational Autoencoders (VAEs) provide a flexible and scalable framework for
non-linear dimensionality reduction. However, in application domains such as
genomics where data sets are typically tabular and high-dimensional, a
black-box approach to dimensionality reduction does not provide sufficient
insights. Common data analysis workflows additionally use clustering techniques
to identify groups of similar features. This usually leads to a two-stage
process, however, it would be desirable to construct a joint modelling
framework for simultaneous dimensionality reduction and clustering of features.
In this paper, we propose to achieve this through the BasisVAE: a combination
of the VAE and a probabilistic clustering prior, which lets us learn a one-hot
basis function representation as part of the decoder network. Furthermore, for
scenarios where not all features are aligned, we develop an extension to handle
translation-invariant basis functions. We show how a collapsed variational
inference scheme leads to scalable and efficient inference for BasisVAE,
demonstrated on various toy examples as well as on single-cell gene expression
data.
| stat |
Distributed Bayesian Varying Coefficient Modeling Using a Gaussian
Process Prior | Varying coefficient models (VCMs) are widely used for estimating nonlinear
regression functions in functional data models. Their Bayesian variants using
Gaussian process (GP) priors on the functional coefficients, however, have
received limited attention in massive data applications. This is primarily due
to the prohibitively slow posterior computations using Markov chain Monte Carlo
(MCMC) algorithms. We address this problem using a divide-and-conquer Bayesian
approach that operates in three steps. The first step creates a large number of
data subsets with much smaller sample sizes by sampling without replacement
from the full data. The second step formulates VCM as a linear mixed-effects
model and develops a data augmentation (DA)-type algorithm for obtaining MCMC
draws of the parameters and predictions on all the subsets in parallel. The
DA-type algorithm appropriately modifies the likelihood such that every subset
posterior distribution is an accurate approximation of the corresponding true
posterior distribution. The third step develops a combination algorithm for
aggregating MCMC-based estimates of the subset posterior distributions into a
single posterior distribution called the Aggregated Monte Carlo (AMC)
posterior. Theoretically, we derive minimax optimal posterior convergence rates
for the AMC posterior distributions of both the varying coefficients and the
mean regression function. We provide quantification on the orders of subset
sample sizes and the number of subsets according to the smoothness properties
of the multivariate GP. The empirical results show that the combination schemes
that satisfy our theoretical assumptions, including the one in the AMC
algorithm, have better nominal coverage, shorter credible intervals, smaller
mean square errors, and higher effective sample size than their main
competitors across diverse simulations and in a real data analysis.
| stat |
Sparsity-driven weighted ensemble classifier | In this study, a novel sparsity-driven weighted ensemble classifier (SDWEC)
that improves classification accuracy and minimizes the number of classifiers
is proposed. Using pre-trained classifiers, an ensemble in which base
classifiers votes according to assigned weights is formed. These assigned
weights directly affect classifier accuracy. In the proposed method, ensemble
weights finding problem is modeled as a cost function with the following terms:
(a) a data fidelity term aiming to decrease misclassification rate, (b) a
sparsity term aiming to decrease the number of classifiers, and (c) a
non-negativity constraint on the weights of the classifiers. As the proposed
cost function is non-convex thus hard to solve, convex relaxation techniques
and novel approximations are employed to obtain a numerically efficient
solution. Sparsity term of cost function allows trade-off between accuracy and
testing time when needed. The efficiency of SDWEC was tested on 11 datasets and
compared with the state-of-the art classifier ensemble methods. The results
show that SDWEC provides better or similar accuracy levels using fewer
classifiers and reduces testing time for ensemble.
| stat |
Online Change-Point Detection in High-Dimensional Covariance Structure
with Application to Dynamic Networks | In this paper, we develop an online change-point detection procedure in the
covariance structure of high-dimensional data. A new stopping rule is proposed
to terminate the process as early as possible when a change in covariance
structure occurs. The stopping rule allows temporal dependence and can be
applied to non-Gaussian data. An explicit expression for the average run length
(ARL) is derived, so that the level of threshold in the stopping rule can be
easily obtained with no need to run time-consuming Monte Carlo simulations. We
also establish an upper bound for the expected detection delay (EDD), the
expression of which demonstrates the impact of data dependence and magnitude of
change in the covariance structure. Simulation studies are provided to confirm
accuracy of the theoretical results. The practical usefulness of the proposed
procedure is illustrated by detecting the change of brain's covariance network
in a resting-state fMRI dataset.
| stat |
Causal effects based on distributional distances | In this paper we develop a framework for characterizing causal effects via
distributional distances. In particular we define a causal effect in terms of
the $L_1$ distance between different counterfactual outcome distributions,
rather than the typical mean difference in outcome values. Comparing entire
counterfactual outcome distributions can provide more nuanced and valuable
measures for exploring causal effects beyond the average treatment effect.
First, we propose a novel way to estimate counterfactual outcome densities,
which is of independent interest. Then we develop an efficient estimator of our
target causal effect. We go on to provide error bounds and asymptotic
properties of the proposed estimator, along with bootstrap-based confidence
intervals. Finally, we illustrate the methods via simulations and real data.
| stat |
Stochastic bandits with arm-dependent delays | Significant work has been recently dedicated to the stochastic delayed bandit
setting because of its relevance in applications. The applicability of existing
algorithms is however restricted by the fact that strong assumptions are often
made on the delay distributions, such as full observability, restrictive shape
constraints, or uniformity over arms. In this work, we weaken them
significantly and only assume that there is a bound on the tail of the delay.
In particular, we cover the important case where the delay distributions vary
across arms, and the case where the delays are heavy-tailed. Addressing these
difficulties, we propose a simple but efficient UCB-based algorithm called the
PatientBandits. We provide both problems-dependent and problems-independent
bounds on the regret as well as performance lower bounds.
| stat |
Domain Generalization by Marginal Transfer Learning | In the problem of domain generalization (DG), there are labeled training data
sets from several related prediction problems, and the goal is to make accurate
predictions on future unlabeled data sets that are not known to the learner.
This problem arises in several applications where data distributions fluctuate
because of environmental, technical, or other sources of variation. We
introduce a formal framework for DG, and argue that it can be viewed as a kind
of supervised learning problem by augmenting the original feature space with
the marginal distribution of feature vectors. While our framework has several
connections to conventional analysis of supervised learning algorithms, several
unique aspects of DG require new methods of analysis.
This work lays the learning theoretic foundations of domain generalization,
building on our earlier conference paper where the problem of DG was introduced
(Blanchard et al., 2011). We present two formal models of data generation,
corresponding notions of risk, and distribution-free generalization error
analysis. By focusing our attention on kernel methods, we also provide more
quantitative results and a universally consistent algorithm. An efficient
implementation is provided for this algorithm, which is experimentally compared
to a pooling strategy on one synthetic and three real-world data sets.
| stat |
Narrowest Significance Pursuit: inference for multiple change-points in
linear models | We propose Narrowest Significance Pursuit (NSP), a general and flexible
methodology for automatically detecting localised regions in data sequences
which each must contain a change-point, at a prescribed global significance
level. Here, change-points are understood as abrupt changes in the parameters
of an underlying linear model. NSP works by fitting the postulated linear model
over many regions of the data, using a certain multiresolution sup-norm loss,
and identifying the shortest interval on which the linearity is significantly
violated. The procedure then continues recursively to the left and to the right
until no further intervals of significance can be found. The use of the
multiresolution sup-norm loss is a key feature of NSP, as it enables the
transfer of significance considerations to the domain of the unobserved true
residuals, a substantial simplification. It also guarantees important
stochastic bounds which directly yield exact desired coverage probabilities,
regardless of the form or number of the regressors.
NSP works with a wide range of distributional assumptions on the errors,
including Gaussian with known or unknown variance, some light-tailed
distributions, and some heavy-tailed, possibly heterogeneous distributions via
self-normalisation. It also works in the presence of autoregression. The
mathematics of NSP is, by construction, uncomplicated, and its key
computational component uses simple linear programming. In contrast to the
widely studied "post-selection inference" approach, NSP enables the opposite
viewpoint and paves the way for the concept of "post-inference selection".
Pre-CRAN R code implementing NSP is available at https://github.com/pfryz/nsp.
| stat |
Power Calculations for Replication Studies | The reproducibility crisis has led to an increasing number of replication
studies being conducted. Sample sizes for replication studies are often
calculated using conditional power based on the effect estimate from the
original study. However, this approach is not well suited as it ignores the
uncertainty of the original result. Bayesian methods are used in clinical
trials to incorporate prior information into power calculations. We propose to
adapt this methodology to the replication framework and promote the use of
predictive instead of conditional power in the design of replication studies.
Moreover, we describe how extensions of the methodology to sequential clinical
trials can be tailored to replication studies. Conditional and predictive power
calculated at an interim analysis are compared and we argue that predictive
power is a useful tool to decide whether to stop a replication study
prematurely. A recent project on the replicability of social sciences is used
to illustrate the properties of the different methods.
| stat |
Estimation Methods for Item Factor Analysis: An Overview | Item factor analysis (IFA) refers to the factor models and statistical
inference procedures for analyzing multivariate categorical data. IFA
techniques are commonly used in social and behavioral sciences for analyzing
item-level response data. Such models summarize and interpret the dependence
structure among a set of categorical variables by a small number of latent
factors. In this chapter, we review the IFA modeling technique and commonly
used IFA models. Then we discuss estimation methods for IFA models and their
computation, with a focus on the situation where the sample size, the number of
items, and the number of factors are all large. Existing statistical softwares
for IFA are surveyed. This chapter is concluded with suggestions for practical
applications of IFA methods and discussions of future directions.
| stat |
Network Hawkes Process Models for Exploring Latent Hierarchy in Social
Animal Interactions | Group-based social dominance hierarchies are of essential interest in animal
behavior research. Studies often record aggressive interactions observed over
time, and models that can capture such dynamic hierarchy are therefore crucial.
Traditional ranking methods summarize interactions across time, using only
aggregate counts. Instead, we take advantage of the interaction timestamps,
proposing a series of network point process models with latent ranks. We
carefully design these models to incorporate important characteristics of
animal interaction data, including the winner effect, bursting and pair-flip
phenomena. Through iteratively constructing and evaluating these models we
arrive at the final cohort Markov-Modulated Hawkes process (C-MMHP), which best
characterizes all aforementioned patterns observed in interaction data. We
compare all models using simulated and real data. Using statistically developed
diagnostic perspectives, we demonstrate that the C-MMHP model outperforms other
methods, capturing relevant latent ranking structures that lead to meaningful
predictions for real data.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.