title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Learning summary features of time series for likelihood free inference | There has been an increasing interest from the scientific community in using
likelihood-free inference (LFI) to determine which parameters of a given
simulator model could best describe a set of experimental data. Despite
exciting recent results and a wide range of possible applications, an important
bottleneck of LFI when applied to time series data is the necessity of defining
a set of summary features, often hand-tailored based on domain knowledge. In
this work, we present a data-driven strategy for automatically learning summary
features from univariate time series and apply it to signals generated from
autoregressive-moving-average (ARMA) models and the Van der Pol Oscillator. Our
results indicate that learning summary features from data can compete and even
outperform LFI methods based on hand-crafted values such as autocorrelation
coefficients even in the linear case.
| stat |
The more you ask, the less you get: the negative impact of collaborative
overload on performance | This paper is about the possible negative impact of excessive collaboration
on the performance of top employees. With the rise of participatory culture and
developments in communications technology, management practices require greater
conceptual awareness about possible outcomes of increased organizational
interconnectivity. While there exists a sound theoretical basis for possible
burdens brought by collaborative overload, the literature never really manage
to measure and empirically test this phenomenon. We address this gap by
developing a methodological framework for the identification of organizational
actors at risk of operational capacity overload. Drawing on social network
analysis as the widely applied approach for the estimation of employees'
involvement in the information exchange networks, this paper describes
potential personal and organizational causes leading to the emergence of
collaborative overload. Relying on primary data gathered through a survey
conducted among employees in a large insurance company, we present a testable
model for overload detection. A second merit of the paper consists in finding a
novel identification strategy for empirical works on cross-sectional network
data, which often face the issue of endogeneity. This research suggests that
active collaborative activity does not cause a decrease throughout every aspect
of performance. We found that expertise sharing depends on a few key players
who take core knowledge assets upon themselves and thus run higher risks of
exposure to overload.
| stat |
Classification with Rejection Based on Cost-sensitive Classification | The goal of classification with rejection is to avoid risky misclassification
in error-critical applications such as medical diagnosis and product
inspection. In this paper, based on the relationship between classification
with rejection and cost-sensitive classification, we propose a novel method of
classification with rejection by learning an ensemble of cost-sensitive
classifiers, which satisfies all the following properties for the first time:
(i) it can avoid estimating class-posterior probabilities, resulting in
improved classification accuracy. (ii) it allows a flexible choice of losses
including non-convex ones, (iii) it does not require complicated modifications
when using different losses, (iv) it is applicable to both binary and
multiclass cases, and (v) it is theoretically justifiable for any
classification-calibrated loss. Experimental results demonstrate the usefulness
of our proposed approach in clean-labeled, noisy-labeled, and
positive-unlabeled classification.
| stat |
Increasing the efficiency of randomized trial estimates via linear
adjustment for a prognostic score | Estimating causal effects from randomized experiments is central to clinical
research. Reducing the statistical uncertainty in these analyses is an
important objective for statisticians. Registries, prior trials, and health
records constitute a growing compendium of historical data on patients under
standard-of-care conditions that may be exploitable to this end. However, most
methods for historical borrowing achieve reductions in variance by sacrificing
strict type-I error rate control. Here, we propose a use of historical data
that exploits linear covariate adjustment to improve the efficiency of trial
analyses without incurring bias. Specifically, we train a prognostic model on
the historical data, then estimate the treatment effect using a linear
regression while adjusting for the trial subjects' predicted outcomes (their
prognostic scores). We prove that, under certain conditions, this prognostic
covariate adjustment procedure attains the minimum variance possible among a
large class of estimators. When those conditions are not met, prognostic
covariate adjustment is still more efficient than raw covariate adjustment and
the gain in efficiency is proportional to a measure of the predictive accuracy
of the prognostic model. We demonstrate the approach using simulations and a
reanalysis of an Alzheimer's Disease clinical trial and observe meaningful
reductions in mean-squared error and the estimated variance. Lastly, we provide
a simplified formula for asymptotic variance that enables power and sample size
calculations that account for the gains from the prognostic model for clinical
trial design. Sample size reductions between 10% and 30% are attainable when
using prognostic models that explain a clinically realistic percentage of the
outcome variance.
| stat |
Posterior Contraction Rate of Sparse Latent Feature Models with
Application to Proteomics | The Indian buffet process (IBP) and phylogenetic Indian buffet process (pIBP)
can be used as prior models to infer latent features in a data set. The
theoretical properties of these models are under-explored, however, especially
in high dimensional settings. In this paper, we show that under mild sparsity
condition, the posterior distribution of the latent feature matrix, generated
via IBP or pIBP priors, converges to the true latent feature matrix
asymptotically. We derive the posterior convergence rate, referred to as the
contraction rate. We show that the convergence holds even when the
dimensionality of the latent feature matrix increases with the sample size,
therefore making the posterior inference valid in high dimensional setting. We
demonstrate the theoretical results using computer simulation, in which the
parallel-tempering Markov chain Monte Carlo method is applied to overcome
computational hurdles. The practical utility of the derived properties is
demonstrated by inferring the latent features in a reverse phase protein arrays
(RPPA) dataset under the IBP prior model. Software and dataset reported in the
manuscript are provided at http://www.compgenome.org/IBP.
| stat |
A Projection Based Conditional Dependence Measure with Applications to
High-dimensional Undirected Graphical Models | Measuring conditional dependence is an important topic in statistics with
broad applications including graphical models. Under a factor model setting, a
new conditional dependence measure based on projection is proposed. The
corresponding conditional independence test is developed with the asymptotic
null distribution unveiled where the number of factors could be
high-dimensional. It is also shown that the new test has control over the
asymptotic significance level and can be calculated efficiently. A generic
method for building dependency graphs without Gaussian assumption using the new
test is elaborated. Numerical results and real data analysis show the
superiority of the new method.
| stat |
Bayesian Variable Selection For Survival Data Using Inverse Moment
Priors | Efficient variable selection in high-dimensional cancer genomic studies is
critical for discovering genes associated with specific cancer types and for
predicting response to treatment. Censored survival data is prevalent in such
studies. In this article we introduce a Bayesian variable selection procedure
that uses a mixture prior composed of a point mass at zero and an inverse
moment prior in conjunction with the partial likelihood defined by the Cox
proportional hazard model. The procedure is implemented in the R package
BVSNLP, which supports parallel computing and uses a stochastic search method
to explore the model space. Bayesian model averaging is used for prediction.
The proposed algorithm provides better performance than other variable
selection procedures in simulation studies, and appears to provide more
consistent variable selection when applied to actual genomic datasets.
| stat |
Point process models for sequence detection in high-dimensional neural
spike trains | Sparse sequences of neural spikes are posited to underlie aspects of working
memory, motor production, and learning. Discovering these sequences in an
unsupervised manner is a longstanding problem in statistical neuroscience.
Promising recent work utilized a convolutive nonnegative matrix factorization
model to tackle this challenge. However, this model requires spike times to be
discretized, utilizes a sub-optimal least-squares criterion, and does not
provide uncertainty estimates for model predictions or estimated parameters. We
address each of these shortcomings by developing a point process model that
characterizes fine-scale sequences at the level of individual spikes and
represents sequence occurrences as a small number of marked events in
continuous time. This ultra-sparse representation of sequence events opens new
possibilities for spike train modeling. For example, we introduce learnable
time warping parameters to model sequences of varying duration, which have been
experimentally observed in neural circuits. We demonstrate these advantages on
experimental recordings from songbird higher vocal center and rodent
hippocampus.
| stat |
Memory- and Communication-Aware Model Compression for Distributed Deep
Learning Inference on IoT | Model compression has emerged as an important area of research for deploying
deep learning models on Internet-of-Things (IoT). However, for extremely
memory-constrained scenarios, even the compressed models cannot fit within the
memory of a single device and, as a result, must be distributed across multiple
devices. This leads to a distributed inference paradigm in which memory and
communication costs represent a major bottleneck. Yet, existing model
compression techniques are not communication-aware. Therefore, we propose
Network of Neural Networks (NoNN), a new distributed IoT learning paradigm that
compresses a large pretrained 'teacher' deep network into several disjoint and
highly-compressed 'student' modules, without loss of accuracy. Moreover, we
propose a network science-based knowledge partitioning algorithm for the
teacher model, and then train individual students on the resulting disjoint
partitions. Extensive experimentation on five image classification datasets,
for user-defined memory/performance budgets, show that NoNN achieves higher
accuracy than several baselines and similar accuracy as the teacher model,
while using minimal communication among students. Finally, as a case study, we
deploy the proposed model for CIFAR-10 dataset on edge devices and demonstrate
significant improvements in memory footprint (up to 24x), performance (up to
12x), and energy per node (up to 14x) compared to the large teacher model. We
further show that for distributed inference on multiple edge devices, our
proposed NoNN model results in up to 33x reduction in total latency w.r.t. a
state-of-the-art model compression baseline.
| stat |
Distance-Based Independence Screening for Canonical Analysis | This paper introduces a new method named Distance-based Independence
Screening for Canonical Analysis (DISCA) to reduce dimensions of two random
vectors with arbitrary dimensions. The objective of our method is to identify
the low dimensional linear projections of two random vectors, such that any
dimension reduction based on linear projection with lower dimensions will
surely affect some dependent structure -- the removed components are not
independent. The essence of DISCA is to use the distance correlation to
eliminate the "redundant" dimensions until infeasible. Unlike the existing
canonical analysis methods, DISCA does not require the dimensions of the
reduced subspaces of the two random vectors to be equal, nor does it require
certain distributional assumption on the random vectors. We show that under
mild conditions, our approach does undercover the lowest possible linear
dependency structures between two random vectors, and our conditions are weaker
than some sufficient linear subspace-based methods. Numerically, DISCA is to
solve a non-convex optimization problem. We formulate it as a
difference-of-convex (DC) optimization problem, and then further adopt the
alternating direction method of multipliers (ADMM) on the convex step of the DC
algorithms to parallelize/accelerate the computation. Some sufficient linear
subspace-based methods use potentially numerically-intensive bootstrap method
to determine the dimensions of the reduced subspaces in advance; our method
avoids this complexity. In simulations, we present cases that DISCA can solve
effectively, while other methods cannot. In both the simulation studies and
real data cases, when the other state-of-the-art dimension reduction methods
are applicable, we observe that DISCA performs either comparably or better than
most of them. Codes and an R package can be found in GitHub
https://github.com/ChuanpingYu/DISCA.
| stat |
Functional additive models for optimizing individualized treatment rules | A novel functional additive model is proposed which is uniquely modified and
constrained to model nonlinear interactions between a treatment indicator and a
potentially large number of functional and/or scalar pretreatment covariates.
The primary motivation for this approach is to optimize individualized
treatment rules based on data from a randomized clinical trial. We generalize
functional additive regression models by incorporating treatment-specific
components into additive effect components. A structural constraint is imposed
on the treatment-specific components in order to provide a class of additive
models with main effects and interaction effects that are orthogonal to each
other. If primary interest is in the interaction between treatment and the
covariates, as is generally the case when optimizing individualized treatment
rules, we can thereby circumvent the need to estimate the main effects of the
covariates, obviating the need to specify their form and thus avoiding the
issue of model misspecification. The methods are illustrated with data from a
depression clinical trial with electroencephalogram functional data as
patients' pretreatment covariates.
| stat |
Kernel density decomposition with an application to the social cost of
carbon | A kernel density is an aggregate of kernel functions, which are itself
densities and could be kernel densities. This is used to decompose a kernel
into its constituent parts. Pearson's test for equality of proportions is
applied to quantiles to test whether the component distributions differ from
one another. The proposed methods are illustrated with a meta-analysis of the
social cost of carbon. Different discount rates lead to significantly different
Pigou taxes, but not different growth rates. Estimates have not varied over
time. Different authors have contributed different estimates, but these
differences are insignificant. Kernel decomposition can be applied in many
other fields with discrete explanatory variables.
| stat |
Estimation and Inference on Treatment Effects Under Treatment-Based
Sampling | Causal inference in a program evaluation setting faces the problem of
external validity when the treatment effect in the target population is
different from the treatment effect identified from the population of which the
sample is representative. This paper focuses on a situation where such
discrepancy arises by a stratified sampling design based on the individual
treatment status and other characteristics. In such settings, the design
probability is known from the sampling design but the target population depends
on the underlying population share vector which is often unknown, and except
for special cases, the treatment effect parameters are not identified. First,
we propose a method of constructing confidence sets that are valid for a given
range of population shares. Second, when a benchmark population share vector
and a corresponding estimator of a treatment effect parameter are given, we
propose a method to discover the scope of external validity with familywise
error rate control. Third, we propose an optimal sampling design which
minimizes the semiparametric efficiency bound given a population share
associated with a target population. We provide Monte Carlo simulation results
and an empirical application to demonstrate the usefulness of our proposals.
| stat |
Sequential Computer Experimental Design for Estimating an Extreme
Probability or Quantile | A computer code can simulate a system's propagation of variation from random
inputs to output measures of quality. Our aim here is to estimate a critical
output tail probability or quantile without a large Monte Carlo experiment.
Instead, we build a statistical surrogate for the input-output relationship
with a modest number of evaluations and then sequentially add further runs,
guided by a criterion to improve the estimate. We compare two criteria in the
literature. Moreover, we investigate two practical questions: how to design the
initial code runs and how to model the input distribution. Hence, we close the
gap between the theory of sequential design and its application.
| stat |
Estimation of Structural Causal Model via Sparsely Mixing Independent
Component Analysis | We consider the problem of inferring the causal structure from observational
data, especially when the structure is sparse. This type of problem is usually
formulated as an inference of a directed acyclic graph (DAG) model. The linear
non-Gaussian acyclic model (LiNGAM) is one of the most successful DAG models,
and various estimation methods have been developed. However, existing methods
are not efficient for some reasons: (i) the sparse structure is not always
incorporated in causal order estimation, and (ii) the whole information of the
data is not used in parameter estimation. To address {these issues}, we propose
a new estimation method for a linear DAG model with non-Gaussian noises. The
proposed method is based on the log-likelihood of independent component
analysis (ICA) with two penalty terms related to the sparsity and the
consistency condition. The proposed method enables us to estimate the causal
order and the parameters simultaneously. For stable and efficient optimization,
we propose some devices, such as a modified natural gradient. Numerical
experiments show that the proposed method outperforms existing methods,
including LiNGAM and NOTEARS.
| stat |
Wisdom of crowds: much ado about nothing | The puzzling idea that the combination of independent estimates of the
magnitude of a quantity results in a very accurate prediction, which is
superior to any or, at least, to most of the individual estimates is known as
the wisdom of crowds. Here we use the Federal Reserve Bank of Philadelphia's
Survey of Professional Forecasters database to confront the statistical and
psychophysical explanations of this phenomenon. Overall we find that the data
do not support any of the proposed explanations of the wisdom of crowds. In
particular, we find a positive correlation between the variance (or diversity)
of the estimates and the crowd error in disagreement with some interpretations
of the diversity prediction theorem. In addition, contra the predictions of the
psychophysical augmented quincunx model, we find that the skew of the estimates
offers no information about the crowd error. More importantly, we find that the
crowd beats all individuals in less than 2% of the forecasts and beats most
individuals in less than 70% of the forecasts, which means that there is a
sporting chance that an individual selected at random will perform better than
the crowd. These results contrast starkly with the performance of non-natural
crowds composed of unbiased forecasters which beat most individuals in
practically all forecasts. The moderate statistical advantage of a real-world
crowd over its members does not justify the ado about its wisdom, which is most
likely a product of the selective attention fallacy.
| stat |
A method for deriving information from running R code | It is often useful to tap information from a running R script. Obvious use
cases include monitoring the consumption of resources (time, memory) and
logging. Perhaps less obvious cases include tracking changes in R objects
orcollecting output of unit tests. In this paper we demonstrate an approach
that abstracts collection and processing of such secondary information from the
running R script. Our approach is based on a combination of three elements. The
first element is to build a customized way to evaluate code. The second is
labeled \emph{local masking} and it involves temporarily masking auser-facing
function so an alternative version of it is called. The third element we label
\emph{local side effect}. This refers to the fact that the masking function
exports information to the secondary information flow without altering a global
state. The result is a method for building systems in pure R that lets users
create and control secondary flows of information with minimal impact on their
workflow, and no global side effects.
| stat |
A flat persistence diagram for improved visualization of persistent
homology | Visualization in the emerging field of topological data analysis has
progressed from persistence barcodes and persistence diagrams to display of
two-parameter persistent homology. Although persistence barcodes and diagrams
have permitted insight into the geometry underlying complex datasets,
visualization of even single-parameter persistent homology has significant room
for improvement. Here, we propose a modification to the conventional
persistence diagram - the flat persistence diagram - that more efficiently
displays information relevant to persistent homology and simultaneously
corrects for visual bias present in the former. Flat persistence diagrams
display equivalent information as their predecessor, while providing
researchers with an intuitive horizontal reference axis in contrast to the
usual diagonal reference line. Reducing visual bias through the use of
appropriate graphical displays not only provides more accurate, but also deeper
insights into the topology that underlies complex datasets. Introducing flat
persistence diagrams into widespread use would bring researchers one step
closer to practical application of topological data analysis.
| stat |
Bayesian workflow for disease transmission modeling in Stan | This tutorial shows how to build, fit, and criticize disease transmission
models in Stan, and should be useful to researchers interested in modeling the
SARS-CoV-2 pandemic and other infectious diseases in a Bayesian framework.
Bayesian modeling provides a principled way to quantify uncertainty and
incorporate both data and prior knowledge into the model estimates. Stan is an
expressive probabilistic programming language that abstracts the inference and
allows users to focus on the modeling. As a result, Stan code is readable and
easily extensible, which makes the modeler's work more transparent.
Furthermore, Stan's main inference engine, Hamiltonian Monte Carlo sampling, is
amiable to diagnostics, which means the user can verify whether the obtained
inference is reliable. In this tutorial, we demonstrate how to formulate, fit,
and diagnose a compartmental transmission model in Stan, first with a simple
Susceptible-Infected-Recovered (SIR) model, then with a more elaborate
transmission model used during the SARS-CoV-2 pandemic. We also cover advanced
topics which can further help practitioners fit sophisticated models; notably,
how to use simulations to probe the model and priors, and computational
techniques to scale-up models based on ordinary differential equations.
| stat |
Classification with the matrix-variate-$t$ distribution | Matrix-variate distributions can intuitively model the dependence structure
of matrix-valued observations that arise in applications with multivariate time
series, spatio-temporal or repeated measures. This paper develops an
Expectation-Maximization algorithm for discriminant analysis and classification
with matrix-variate $t$-distributions. The methodology shows promise on
simulated datasets or when applied to the forensic matching of fractured
surfaces or the classification of functional Magnetic Resonance, satellite or
hand gestures images.
| stat |
The diffusion-based extension of the Mat\'ern field to space-time | The Mat\' ern field is the most well known family of covariance functions
used for Gaussian processes in spatial models. We build upon the original
research of Whittle (1953, 1964) and develop the diffusion-based extension of
the Mat\' ern field to space-time (DEMF). We argue that this diffusion-based
extension is the natural extension of these processes, due to the strong
physical interpretation. The corresponding non-separable spatio-temporal
Gaussian process is a spatio-temporal analogue of the Mat\' ern field, with
range parameters in space and time, smoothness parameters in space and time,
and a separability parameter. We provide a sparse representation based on
finite element methods that is well suited for statistical inference.
| stat |
A Bayesian Nonparametric Approach for Evaluating the Causal Effect of
Treatment in Randomized Trials with Semi-Competing Risks | We develop a Bayesian nonparametric (BNP) approach to evaluate the causal
effect of treatment in a randomized trial where a nonterminal event may be
censored by a terminal event, but not vice versa (i.e., semi-competing risks).
Based on the idea of principal stratification, we define a novel estimand for
the causal effect of treatment on the nonterminal event. We introduce
identification assumptions, indexed by a sensitivity parameter, and show how to
draw inference using our BNP approach. We conduct simulation studies and
illustrate our methodology using data from a brain cancer trial.
| stat |
Measurement error models: from nonparametric methods to deep neural
networks | The success of deep learning has inspired recent interests in applying neural
networks in statistical inference. In this paper, we investigate the use of
deep neural networks for nonparametric regression with measurement errors. We
propose an efficient neural network design for estimating measurement error
models, in which we use a fully connected feed-forward neural network (FNN) to
approximate the regression function $f(x)$, a normalizing flow to approximate
the prior distribution of $X$, and an inference network to approximate the
posterior distribution of $X$. Our method utilizes recent advances in
variational inference for deep neural networks, such as the importance weight
autoencoder, doubly reparametrized gradient estimator, and non-linear
independent components estimation. We conduct an extensive numerical study to
compare the neural network approach with classical nonparametric methods and
observe that the neural network approach is more flexible in accommodating
different classes of regression functions and performs superior or comparable
to the best available method in nearly all settings.
| stat |
Impacts of Heat Decarbonisation on System Adequacy considering Increased
Meteorological Sensitivity | This paper explores the impacts of decarbonisation of heat on demand and
subsequently on the generation capacity required to secure against system
adequacy standards. Gas demand is explored as a proxy variable for modelling
the electrification of heating demand in existing housing stock, with a focus
on impacts on timescales of capacity markets (up to four years ahead). The work
considers the systemic changes that electrification of heating could introduce,
including biases that could be introduced if legacy modelling approaches
continue to prevail. Covariates from gas and electrical regression models are
combined to form a novel, time-collapsed system model, with demand-weather
sensitivities determined using lasso-regularized linear regression. It is
shown, using a GB case study with one million domestic heat pump installations
per year, that the sensitivity of electrical system demand to temperature (and
subsequently sensitivities to cold/warm winter seasons) could increase by 50%
following four years of heat demand electrification. A central estimate of 1.75
kW additional peak demand per heat pump is estimated, with variability across
three published heat demand profiles leading to a range of more than 14 GW in
the most extreme cases. It is shown that the legacy approach of scaling
historic demand, as compared to the explicit modelling of heat, could lead to
over-procurement of 0.79 GW due to bias in estimates of additional capacity to
secure. Failure to address this issue could lead to {\pounds}100m overspend on
capacity over ten years.
| stat |
Sharp Analysis of a Simple Model for Random Forests | Random forests have become an important tool for improving accuracy in
regression and classification problems since their inception by Leo Breiman in
2001. In this paper, we revisit a historically important random forest model
originally proposed by Breiman in 2004 and later studied by G\'erard Biau in
2012, where a feature is selected at random and the splits occurs at the
midpoint of the node along the chosen feature. If the regression function is
Lipschitz and depends only on a small subset of $ S $ out of $ d $ features, we
show that, given access to $ n $ observations and properly tuned split
probabilities, the mean-squared prediction error is $ O((n(\log
n)^{(S-1)/2})^{-\frac{1}{S\log2+1}}) $. This positively answers an outstanding
question of Biau about whether the rate of convergence for this random forest
model could be improved. Furthermore, by a refined analysis of the
approximation and estimation errors for linear models, we show that this rate
cannot be improved in general. Finally, we generalize our analysis and improve
extant prediction error bounds for another random forest model in which each
tree is constructed from subsampled data and the splits are performed at the
empirical median along a chosen feature.
| stat |
Linear Mixed Models for Comparing Dynamic Treatment Regimens on a
Longitudinal Outcome in Sequentially Randomized Trials | A dynamic treatment regimen (DTR) is a pre-specified sequence of decision
rules which maps baseline or time-varying measurements on an individual to a
recommended intervention or set of interventions. Sequential multiple
assignment randomized trials (SMARTs) represent an important data collection
tool for informing the construction of effective DTRs. A common primary aim in
a SMART is the marginal mean comparison between two or more of the DTRs
embedded in the trial. This manuscript develops a mixed effects modeling and
estimation approach for these primary aim comparisons based on a continuous,
longitudinal outcome. The method is illustrated using data from a SMART in
autism research.
| stat |
Towards Automatic Concept-based Explanations | Interpretability has become an important topic of research as more machine
learning (ML) models are deployed and widely used to make important decisions.
Most of the current explanation methods provide explanations through feature
importance scores, which identify features that are important for each
individual input. However, how to systematically summarize and interpret such
per sample feature importance scores itself is challenging. In this work, we
propose principles and desiderata for \emph{concept} based explanation, which
goes beyond per-sample features to identify higher-level human-understandable
concepts that apply across the entire dataset. We develop a new algorithm, ACE,
to automatically extract visual concepts. Our systematic experiments
demonstrate that \alg discovers concepts that are human-meaningful, coherent
and important for the neural network's predictions.
| stat |
Optimizing the tie-breaker regression discontinuity design | Motivated by customer loyalty plans and scholarship programs, we study
tie-breaker designs which are hybrids of randomized controlled trials (RCTs)
and regression discontinuity designs (RDDs). We quantify the statistical
efficiency of a tie-breaker design in which a proportion $\Delta$ of observed
subjects are in the RCT. In a two line regression, statistical efficiency
increases monotonically with $\Delta$, so efficiency is maximized by an RCT. We
point to additional advantages of tie-breakers versus RDD: for a nonparametric
regression the boundary bias is much less severe and for quadratic regression,
the variance is greatly reduced. For a two line model we can quantify the short
term value of the treatment allocation and this comparison favors smaller
$\Delta$ with the RDD being best. We solve for the optimal tradeoff between
these exploration and exploitation goals. The usual tie-breaker design applies
an RCT on the middle $\Delta$ subjects as ranked by the assignment variable. We
quantify the efficiency of other designs such as experimenting only in the
second decile from the top. We also show that in some general parametric models
a Monte Carlo evaluation can be replaced by matrix algebra.
| stat |
Conceptualising Natural and Quasi Experiments in Public Health | Background: Natural or quasi experiments are appealing for public health
research because they enable the evaluation of events or interventions that are
difficult or impossible to manipulate experimentally, such as many policy and
health system reforms. However, there remains ambiguity in the literature about
their definition and how they differ from randomised controlled experiments and
from other observational designs.
Methods: We conceptualise natural experiments in in the context of public
health evaluations, align the study design to the Target Trial Framework, and
provide recommendation for improvement of their design and reporting.
Results: Natural experiment studies combine features of experiments and
non-experiments. They differ from RCTs in that exposure allocation is not
controlled by researchers while they differ from other observational designs in
that they evaluate the impact of event or exposure changes. As a result they
are, in theory, less susceptible to bias than other observational study
designs. Importantly, the strength of causal inferences relies on the
plausibility that the exposure allocation can be considered "as-if randomised".
The target trial framework provides a systematic basis for assessing the
plausibility of such claims, and enables a structured method for assessing
other design elements.
Conclusions: Natural experiment studies should be considered a distinct study
design rather than a set of tools for analyses of non-randomised interventions.
Alignment of natural experiments to the Target Trial framework will clarify the
strength of evidence underpinning claims about the effectiveness of public
health interventions.
| stat |
Estimating the association between blood pressure variability and
cardiovascular disease: An application using the ARIC Study | The association between visit-to-visit systolic blood pressure variability
and cardiovascular events has recently received a lot of attention in the
cardiovascular literature. But blood pressure variability is usually estimated
on a person-by-person basis, and is therefore subject to considerable
measurement error. We demonstrate that hazard ratios estimated using this
approach are subject to bias due to regression dilution and we propose
alternative methods to reduce this bias: a two-stage method and a joint model.
For the two-stage method, in stage one repeated measurements are modelled using
a mixed effects model with a random component on the residual standard
deviation. The mixed effects model is used to estimate the blood pressure
standard deviation for each individual, which in stage two is used as a
covariate in a time-to-event model. For the joint model, the mixed effects
sub-model and time-to-event sub-model are fitted simultaneously using shared
random effects. We illustrate the methods using data from the Atherosclerosis
Risk in Communities (ARIC) study.
| stat |
Bayesian state space modelling for COVID-19: with Tennessee and New York
case studies | We develop a Bayesian inferential framework for the spread of COVID-19 using
mechanistic epidemiological models, such as SIR or SEIR, and allow the
effective contact rate to vary in time. A novel aspect of our approach is the
incorporation of a time-varying reporting rate accounting for the initial phase
of the pandemic before testing was widely available. By varying both the
reporting rate and the effective contact rate in time, our models can capture
changes in the data induced by external influences, such as public health
intervention measures, for example. We view COVID-19 incidence data as the
observed measurements of a hidden Markov model, with latent space represented
by the underlying epidemiological model, and employ a particle Markov chain
Monte Carlo (PMCMC) sampling scheme for Bayesian inference. Parameter inference
is performed via PMCMC on incidence data collated by the New York Times from
the states of New York and Tennessee from March 1, 2020 through August 30,
2020. Lastly, we perform Bayesian model selection on the different formulations
of the epidemiological models, make predictions from our fitted models, and
validate our predictions against the true incidence data for the week between
August 31, 2020 and September 7, 2020.
| stat |
Fast Mixing of Multi-Scale Langevin Dynamics under the Manifold
Hypothesis | Recently, the task of image generation has attracted much attention. In
particular, the recent empirical successes of the Markov Chain Monte Carlo
(MCMC) technique of Langevin Dynamics have prompted a number of theoretical
advances; despite this, several outstanding problems remain. First, the
Langevin Dynamics is run in very high dimension on a nonconvex landscape; in
the worst case, due to the NP-hardness of nonconvex optimization, it is thought
that Langevin Dynamics mixes only in time exponential in the dimension. In this
work, we demonstrate how the manifold hypothesis allows for the considerable
reduction of mixing time, from exponential in the ambient dimension to
depending only on the (much smaller) intrinsic dimension of the data. Second,
the high dimension of the sampling space significantly hurts the performance of
Langevin Dynamics; we leverage a multi-scale approach to help ameliorate this
issue and observe that this multi-resolution algorithm allows for a trade-off
between image quality and computational expense in generation.
| stat |
Parametric mode regression for bounded responses | We propose new parametric frameworks of regression analysis with the
conditional mode of a bounded response as the focal point of interest.
Covariate effects estimation and prediction based on the maximum likelihood
method under two new classes of regression models are demonstrated. We also
develop graphical and numerical diagnostic tools to detect various sources of
model misspecification. Predictions based on different central tendency
measures inferred using various regression models are compared using synthetic
data in simulations. Finally, we conduct regression analysis for data from the
Alzheimer's Disease Neuroimaging Initiative to demonstrate practical
implementation of the proposed methods. Supplementary materials that contain
technical details, and additional simulation and data analysis results are
available online.
| stat |
An Exact Solution Path Algorithm for SLOPE and Quasi-Spherical OSCAR | Sorted $L_1$ penalization estimator (SLOPE) is a regularization technique for
sorted absolute coefficients in high-dimensional regression. By arbitrarily
setting its regularization weights $\lambda$ under the monotonicity constraint,
SLOPE can have various feature selection and clustering properties. On weight
tuning, the selected features and their clusters are very sensitive to the
tuning parameters. Moreover, the exhaustive tracking of their changes is
difficult using grid search methods. This study presents a solution path
algorithm that provides the complete and exact path of solutions for SLOPE in
fine-tuning regularization weights. A simple optimality condition for SLOPE is
derived and used to specify the next splitting point of the solution path. This
study also proposes a new design of a regularization sequence $\lambda$ for
feature clustering, which is called the quasi-spherical and octagonal shrinkage
and clustering algorithm for regression (QS-OSCAR). QS-OSCAR is designed with a
contour surface of the regularization terms most similar to a sphere. Among
several regularization sequence designs, sparsity and clustering performance
are compared through simulation studies. The numerical observations show that
QS-OSCAR performs feature clustering more efficiently than other designs.
| stat |
Sample complexity and effective dimension for regression on manifolds | We consider the theory of regression on a manifold using reproducing kernel
Hilbert space methods. Manifold models arise in a wide variety of modern
machine learning problems, and our goal is to help understand the effectiveness
of various implicit and explicit dimensionality-reduction methods that exploit
manifold structure. Our first key contribution is to establish a novel
nonasymptotic version of the Weyl law from differential geometry. From this we
are able to show that certain spaces of smooth functions on a manifold are
effectively finite-dimensional, with a complexity that scales according to the
manifold dimension rather than any ambient data dimension. Finally, we show
that given (potentially noisy) function values taken uniformly at random over a
manifold, a kernel regression estimator (derived from the spectral
decomposition of the manifold) yields minimax-optimal error bounds that are
controlled by the effective dimension.
| stat |
From controlled to undisciplined data: estimating causal effects in the
era of data science using a potential outcome framework | This paper discusses the fundamental principles of causal inference - the
area of statistics that estimates the effect of specific occurrences,
treatments, interventions, and exposures on a given outcome from experimental
and observational data. We explain the key assumptions required to identify
causal effects, and highlight the challenges associated with the use of
observational data. We emphasize that experimental thinking is crucial in
causal inference. The quality of the data (not necessarily the quantity), the
study design, the degree to which the assumptions are met, and the rigor of the
statistical analysis allow us to credibly infer causal effects. Although we
advocate leveraging the use of big data and the application of machine learning
(ML) algorithms for estimating causal effects, they are not a substitute of
thoughtful study design. Concepts are illustrated via examples.
| stat |
Towards Better Generalization: BP-SVRG in Training Deep Neural Networks | Stochastic variance-reduced gradient (SVRG) is a classical optimization
method. Although it is theoretically proved to have better convergence
performance than stochastic gradient descent (SGD), the generalization
performance of SVRG remains open. In this paper we investigate the effects of
some training techniques, mini-batching and learning rate decay, on the
generalization performance of SVRG, and verify the generalization performance
of Batch-SVRG (B-SVRG). In terms of the relationship between optimization and
generalization, we believe that the average norm of gradients on each training
sample as well as the norm of average gradient indicate how flat the landscape
is and how well the model generalizes. Based on empirical observations of such
metrics, we perform a sign switch on B-SVRG and derive a practical algorithm,
BatchPlus-SVRG (BP-SVRG), which is numerically shown to enjoy better
generalization performance than B-SVRG, even SGD in some scenarios of deep
neural networks.
| stat |
GANchors: Realistic Image Perturbation Distributions for Anchors Using
Generative Models | We extend and improve the work of Model Agnostic Anchors for explanations on
image classification through the use of generative adversarial networks (GANs).
Using GANs, we generate samples from a more realistic perturbation
distribution, by optimizing under a lower dimensional latent space. This
increases the trust in an explanation, as results now come from images that are
more likely to be found in the original training set of a classifier, rather
than an overlay of random images. A large drawback to our method is the
computational complexity of sampling through optimization; to address this, we
implement more efficient algorithms, including a diverse encoder. Lastly, we
share results from the MNIST and CelebA datasets, and note that our
explanations can lead to smaller and higher precision anchors.
| stat |
The COVID-19 pandemic's impact on U.S. electricity demand and supply: an
early view from the data | After the onset of the recent COVID-19 pandemic, a number of studies reported
on possible changes in electricity consumption trends. The overall theme of
these reports was that ``electricity use has decreased during the pandemic, but
the power grid is still reliable''---mostly due to reduced economic activity.
In this paper we analyze electricity data upto end of May 2020, examining both
electricity demand and variables that can indicate stress on the power grid,
such as peak demand and demand ramp-rate. We limit this study to three states
in the USA: New York, California, and Florida. The results indicate that the
effect of the pandemic on electricity demand is not a simple reduction from
comparable time frames, and there are noticeable differences among regions. The
variables that can indicate stress on the grid also conveyed mixed messages:
some indicate an increase in stress, some indicate a decrease, and some do not
indicate any clear difference. A positive message is that some of the changes
that were observed around the time stay-at-home orders were issued appeared to
revert back by May 2020. A key challenge in ascribing any observed change to
the pandemic is correcting for weather. We provide a weather-correction method,
apply it to a small city-wide area, and discuss the implications of the
estimated changes in demand. The weather correction exercise underscored that
weather-correction is as challenging as it is important.
| stat |
Estimands and Inference in Cluster-Randomized Vaccine Trials | Cluster-randomized trials are often conducted to assess vaccine effects.
Defining estimands of interest before conducting a trial is integral to the
alignment between a study's objectives and the data to be collected and
analyzed. This paper considers estimands and estimators for overall, indirect,
and total vaccine effects in trials where clusters of individuals are
randomized to vaccine or control. The scenario is considered where individuals
self-select whether to participate in the trial and the outcome of interest is
measured on all individuals in each cluster. Unlike the overall, indirect, and
total effects, the direct effect of vaccination is shown in general not to be
estimable without further assumptions, such as no unmeasured confounding. An
illustrative example motivated by a cluster-randomized typhoid vaccine trial is
provided.
| stat |
A Framework for Interdomain and Multioutput Gaussian Processes | One obstacle to the use of Gaussian processes (GPs) in large-scale problems,
and as a component in deep learning system, is the need for bespoke derivations
and implementations for small variations in the model or inference. In order to
improve the utility of GPs we need a modular system that allows rapid
implementation and testing, as seen in the neural network community. We present
a mathematical and software framework for scalable approximate inference in
GPs, which combines interdomain approximations and multiple outputs. Our
framework, implemented in GPflow, provides a unified interface for many
existing multioutput models, as well as more recent convolutional structures.
This simplifies the creation of deep models with GPs, and we hope that this
work will encourage more interest in this approach.
| stat |
Best Subset Selection in Reduced Rank Regression | We design a new algorithm on the best subset selection model in reduced rank
regression.
| stat |
Generalized Liquid Association Analysis for Multimodal Data Integration | Multimodal data are now prevailing in scientific research. A central question
in multimodal integrative analysis is to understand how two data modalities
associate and interact with each other given another modality or demographic
variables. The problem can be formulated as studying the associations among
three sets of random variables, a question that has received relatively less
attention in the literature. In this article, we propose a novel generalized
liquid association analysis method, which offers a new and unique angle to this
important class of problems of studying three-way associations. We extend the
notion of liquid association of \citet{li2002LA} from the univariate setting to
the sparse, multivariate, and high-dimensional setting. We establish a
population dimension reduction model, transform the problem to sparse Tucker
decomposition of a three-way tensor, and develop a higher-order orthogonal
iteration algorithm for parameter estimation. We derive the non-asymptotic
error bound and asymptotic consistency of the proposed estimator, while
allowing the variable dimensions to be larger than and diverge with the sample
size. We demonstrate the efficacy of the method through both simulations and a
multimodal neuroimaging application for Alzheimer's disease research.
| stat |
Generalized Linear Models for Longitudinal Data with Biased Sampling
Designs: A Sequential Offsetted Regressions Approach | Biased sampling designs can be highly efficient when studying rare (binary)
or low variability (continuous) endpoints. We consider longitudinal data
settings in which the probability of being sampled depends on a repeatedly
measured response through an outcome-related, auxiliary variable. Such
auxiliary variable- or outcome-dependent sampling improves observed response
and possibly exposure variability over random sampling, {even though} the
auxiliary variable is not of scientific interest. {For analysis,} we propose a
generalized linear model based approach using a sequence of two offsetted
regressions. The first estimates the relationship of the auxiliary variable to
response and covariate data using an offsetted logistic regression model. The
offset hinges on the (assumed) known ratio of sampling probabilities for
different values of the auxiliary variable. Results from the auxiliary model
are used to estimate observation-specific probabilities of being sampled
conditional on the response and covariates, and these probabilities are then
used to account for bias in the second, target population model. We provide
asymptotic standard errors accounting for uncertainty in the estimation of the
auxiliary model, and perform simulation studies demonstrating substantial bias
reduction, correct coverage probability, and improved design efficiency over
simple random sampling designs. We illustrate the approaches with two examples.
| stat |
Learning Causal Semantic Representation for Out-of-Distribution
Prediction | Conventional supervised learning methods, especially deep ones, are found to
be sensitive to out-of-distribution (OOD) examples, largely because the learned
representation mixes the semantic factor with the variation factor due to their
domain-specific correlation, while only the semantic factor causes the output.
To address the problem, we propose a Causal Semantic Generative model (CSG)
based on a causal thought so that the two factors are modeled separately, and
develop methods to learn it on a single training domain and predict in a test
domain without (OOD generalization) or with unsupervised data (domain
adaptation). We prove that under proper conditions, CSG identifies the semantic
factor by learning from training data, and this semantic identification
guarantees the boundedness of OOD generalization error and the success of
adaptation. The methods and theory are built on the invariance principle of
causal generative mechanisms, which is fundamental and general. The methods are
based on variational Bayes, with a novel design for both efficient learning and
easy prediction. Empirical study demonstrates the improved test accuracy for
both OOD generalization and domain adaptation.
| stat |
Static and Dynamic Models for Multivariate Distribution Forecasts:
Proper Scoring Rule Tests of Factor-Quantile vs. Multivariate GARCH Models | A plethora of static and dynamic models exist to forecast Value-at-Risk and
other quantile-related metrics used in financial risk management. Industry
practice tends to favour simpler, static models such as historical simulation
or its variants whereas most academic research centres on dynamic models in the
GARCH family. While numerous studies examine the accuracy of multivariate
models for forecasting risk metrics, there is little research on accurately
predicting the entire multivariate distribution. Yet this is an essential
element of asset pricing or portfolio optimization problems having non-analytic
solutions. We approach this highly complex problem using a variety of proper
multivariate scoring rules to evaluate over 100,000 forecasts of
eight-dimensional multivariate distributions: of exchange rates, interest rates
and commodity futures. This way we test the performance of static models, viz.
empirical distribution functions and a new factor-quantile model, with commonly
used dynamic models in the asymmetric multivariate GARCH class.
| stat |
STAR: A general interactive framework for FDR control under structural
constraints | We propose a general framework based on selectively traversed accumulation
rules (STAR) for interactive multiple testing with generic structural
constraints on the rejection set. It combines accumulation tests from ordered
multiple testing with data-carving ideas from post-selection inference,
allowing for highly flexible adaptation to generic structural information. Our
procedure defines an interactive protocol for gradually pruning a candidate
rejection set, beginning with the set of all hypotheses and shrinking with each
step. By restricting the information at each step via a technique we call
masking, our protocol enables interaction while controlling the false discovery
rate (FDR) in finite samples for any data-adaptive update rule that the analyst
may choose. We suggest update rules for a variety of applications with complex
structural constraints, show that STAR performs well for problems ranging from
convex region detection to FDR control on directed acyclic graphs, and show how
to extend it to regression problems where knockoff statistics are available in
lieu of $p$-values.
| stat |
Real-time Prediction of COVID-19 related Mortality using Electronic
Health Records | Coronavirus Disease 2019 (COVID-19) is an emerging respiratory disease caused
by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) with rapid
human-to-human transmission and a high case fatality rate particularly in older
patients. Due to the exponential growth of infections, many healthcare systems
across the world are under pressure to care for increasing amounts of at-risk
patients. Given the high number of infected patients, identifying patients with
the highest mortality risk early is critical to enable effective intervention
and optimal prioritisation of care. Here, we present the COVID-19 Early Warning
System (CovEWS), a clinical risk scoring system for assessing COVID-19 related
mortality risk. CovEWS provides continuous real-time risk scores for individual
patients with clinically meaningful predictive performance up to 192 hours (8
days) in advance, and is automatically derived from patients' electronic health
records (EHRs) using machine learning. We trained and evaluated CovEWS using
de-identified data from a cohort of 66430 COVID-19 positive patients seen at
over 69 healthcare institutions in the United States (US), Australia, Malaysia
and India amounting to an aggregated total of over 2863 years of patient
observation time. On an external test cohort of 5005 patients, CovEWS predicts
COVID-19 related mortality from $78.8\%$ ($95\%$ confidence interval [CI]:
$76.0$, $84.7\%$) to $69.4\%$ ($95\%$ CI: $57.6, 75.2\%$) specificity at a
sensitivity greater than $95\%$ between respectively 1 and 192 hours prior to
observed mortality events - significantly outperforming existing generic and
COVID-19 specific clinical risk scores. CovEWS could enable clinicians to
intervene at an earlier stage, and may therefore help in preventing or
mitigating COVID-19 related mortality.
| stat |
Estimating Sleep & Work Hours from Alternative Data by Segmented
Functional Classification Analysis (SFCA) | Alternative data is increasingly adapted to predict human and economic
behaviour. This paper introduces a new type of alternative data by
re-conceptualising the internet as a data-driven insights platform at global
scale. Using data from a unique internet activity and location dataset drawn
from over 1.5 trillion observations of end-user internet connections, we
construct a functional dataset covering over 1,600 cities during a 7 year
period with temporal resolution of just 15min. To predict accurate temporal
patterns of sleep and work activity from this data-set, we develop a new
technique, Segmented Functional Classification Analysis (SFCA), and compare its
performance to a wide array of linear, functional, and classification methods.
To confirm the wider applicability of SFCA, in a second application we predict
sleep and work activity using SFCA from US city-wide electricity demand
functional data. Across both problems, SFCA is shown to out-perform current
methods.
| stat |
A Regularized Approach to Sparse Optimal Policy in Reinforcement
Learning | We propose and study a general framework for regularized Markov decision
processes (MDPs) where the goal is to find an optimal policy that maximizes the
expected discounted total reward plus a policy regularization term. The extant
entropy-regularized MDPs can be cast into our framework. Moreover, under our
framework, many regularization terms can bring multi-modality and sparsity,
which are potentially useful in reinforcement learning. In particular, we
present sufficient and necessary conditions that induce a sparse optimal
policy. We also conduct a full mathematical analysis of the proposed
regularized MDPs, including the optimality condition, performance error, and
sparseness control. We provide a generic method to devise regularization forms
and propose off-policy actor critic algorithms in complex environment settings.
We empirically analyze the numerical properties of optimal policies and compare
the performance of different sparse regularization forms in discrete and
continuous environments.
| stat |
Hamiltonian Monte-Carlo for Orthogonal Matrices | We consider the problem of sampling from posterior distributions for Bayesian
models where some parameters are restricted to be orthogonal matrices. Such
matrices are sometimes used in neural networks models for reasons of
regularization and stabilization of training procedures, and also can
parameterize matrices of bounded rank, positive-definite matrices and others.
In \citet{byrne2013geodesic} authors have already considered sampling from
distributions over manifolds using exact geodesic flows in a scheme similar to
Hamiltonian Monte Carlo (HMC). We propose new sampling scheme for a set of
orthogonal matrices that is based on the same approach, uses ideas of
Riemannian optimization and does not require exact computation of geodesic
flows. The method is theoretically justified by proof of symplecticity for the
proposed iteration. In experiments we show that the new scheme is comparable or
faster in time per iteration and more sample-efficient comparing to
conventional HMC with explicit orthogonal parameterization and Geodesic
Monte-Carlo. We also provide promising results of Bayesian ensembling for
orthogonal neural networks and low-rank matrix factorization.
| stat |
Model-based metrics: Sample-efficient estimates of predictive model
subpopulation performance | Machine learning models $-$ now commonly developed to screen, diagnose, or
predict health conditions $-$ are evaluated with a variety of performance
metrics. An important first step in assessing the practical utility of a model
is to evaluate its average performance over an entire population of interest.
In many settings, it is also critical that the model makes good predictions
within predefined subpopulations. For instance, showing that a model is fair or
equitable requires evaluating the model's performance in different demographic
subgroups. However, subpopulation performance metrics are typically computed
using only data from that subgroup, resulting in higher variance estimates for
smaller groups. We devise a procedure to measure subpopulation performance that
can be more sample-efficient than the typical subsample estimates. We propose
using an evaluation model $-$ a model that describes the conditional
distribution of the predictive model score $-$ to form model-based metric (MBM)
estimates. Our procedure incorporates model checking and validation, and we
propose a computationally efficient approximation of the traditional
nonparametric bootstrap to form confidence intervals. We evaluate MBMs on two
main tasks: a semi-synthetic setting where ground truth metrics are available
and a real-world hospital readmission prediction task. We find that MBMs
consistently produce more accurate and lower variance estimates of model
performance for small subpopulations.
| stat |
Rodent: Relevance determination in differential equations | We aim to identify the generating, ordinary differential equation (ODE) from
a set of trajectories of a partially observed system. Our approach does not
need prescribed basis functions to learn the ODE model, but only a rich set of
Neural Arithmetic Units. For maximal explainability of the learnt model, we
minimise the state size of the ODE as well as the number of non-zero parameters
that are needed to solve the problem. This sparsification is realized through a
combination of the Variational Auto-Encoder (VAE) and Automatic Relevance
Determination (ARD). We show that it is possible to learn not only one specific
model for a single process, but a manifold of models representing harmonic
signals as well as a manifold of Lotka-Volterra systems.
| stat |
A simulation study of disaggregation regression for spatial disease
mapping | Disaggregation regression has become an important tool in spatial disease
mapping for making fine-scale predictions of disease risk from aggregated
response data. By including high resolution covariate information and modelling
the data generating process on a fine scale, it is hoped that these models can
accurately learn the relationships between covariates and response at a fine
spatial scale. However, validating these high resolution predictions can be a
challenge, as often there is no data observed at this spatial scale. In this
study, disaggregation regression was performed on simulated data in various
settings and the resulting fine-scale predictions are compared to the simulated
ground truth. Performance was investigated with varying numbers of data points,
sizes of aggregated areas and levels of model misspecification. The
effectiveness of cross validation on the aggregate level as a measure of
fine-scale predictive performance was also investigated. Predictive performance
improved as the number of observations increased and as the size of the
aggregated areas decreased. When the model was well-specified, fine-scale
predictions were accurate even with small numbers of observations and large
aggregated areas. Under model misspecification predictive performance was
significantly worse for large aggregated areas but remained high when response
data was aggregated over smaller regions. Cross-validation correlation on the
aggregate level was a moderately good predictor of fine-scale predictive
performance. While the simulations are unlikely to capture the nuances of
real-life response data, this study gives insight into the effectiveness of
disaggregation regression in different contexts.
| stat |
Quantile Treatment Effects in Regression Kink Designs | The literature on regression kink designs develops identification results for
average effects of continuous treatments (Card, Lee, Pei, and Weber, 2015),
average effects of binary treatments (Dong, 2018), and quantile-wise effects of
continuous treatments (Chiang and Sasaki, 2019), but there has been no
identification result for quantile-wise effects of binary treatments to date.
In this paper, we fill this void in the literature by providing an
identification of quantile treatment effects in regression kink designs with
binary treatment variables. For completeness, we also develop large sample
theories for statistical inference and a practical guideline on estimation and
inference.
| stat |
Active Testing: Sample-Efficient Model Evaluation | We introduce active testing: a new framework for sample-efficient model
evaluation. While approaches like active learning reduce the number of labels
needed for model training, existing literature largely ignores the cost of
labeling test data, typically unrealistically assuming large test sets for
model evaluation. This creates a disconnect to real applications where test
labels are important and just as expensive, e.g. for optimizing
hyperparameters. Active testing addresses this by carefully selecting the test
points to label, ensuring model evaluation is sample-efficient. To this end, we
derive theoretically-grounded and intuitive acquisition strategies that are
specifically tailored to the goals of active testing, noting these are distinct
to those of active learning. Actively selecting labels introduces a bias; we
show how to remove that bias while reducing the variance of the estimator at
the same time. Active testing is easy to implement, effective, and can be
applied to any supervised machine learning method. We demonstrate this on
models including WideResNet and Gaussian processes on datasets including
CIFAR-100.
| stat |
A method to find an efficient and robust sampling strategy under model
uncertainty | We consider the problem of deciding on sampling strategy, in particular
sampling design. We propose a risk measure, whose minimizing value guides the
choice. The method makes use of a superpopulation model and takes into account
uncertainty about its parameters. The method is illustrated with a real
dataset, yielding satisfactory results. As a baseline, we use the strategy that
couples probability proportional-to-size sampling with the difference
estimator, as it is known to be optimal when the superpopulation model is fully
known. We show that, even under moderate misspecifications of the model, this
strategy is not robust and can be outperformed by some alternatives
| stat |
Fast Bayesian Record Linkage With Record-Specific Disagreement
Parameters | Researchers are often interested in linking individuals between two datasets
that lack a common unique identifier. Matching procedures often struggle to
match records with common names, birthplaces or other field values.
Computational feasibility is also a challenge, particularly when linking large
datasets. We develop a Bayesian method for automated probabilistic record
linkage and show it recovers more than 50% more true matches, holding accuracy
constant, than comparable methods in a matching of military recruitment data to
the 1900 US Census for which expert-labelled matches are available. Our
approach, which builds on a recent state-of-the-art Bayesian method, refines
the modelling of comparison data, allowing disagreement probability parameters
conditional on non-match status to be record-specific in the smaller of the two
datasets. This flexibility significantly improves matching when many records
share common field values. We show that our method is computationally feasible
in practice, despite the added complexity, with an R/C++ implementation that
achieves significant improvement in speed over comparable recent methods. We
also suggest a lightweight method for treatment of very common names and show
how to estimate true positive rate and positive predictive value when true
match status is unavailable.
| stat |
Prior sample size extensions for assessing prior impact and
prior--likelihood discordance | This paper outlines a framework for quantifying the prior's contribution to
posterior inference in the presence of prior-likelihood discordance, a broader
concept than the usual notion of prior-likelihood conflict. We achieve this
dual purpose by extending the classic notion of \textit{prior sample size},
$M$, in three directions: (I) estimating $M$ beyond conjugate families; (II)
formulating $M$ as a relative notion, i.e., as a function of the likelihood
sample size $k, M(k),$ which also leads naturally to a graphical diagnosis; and
(III) permitting negative $M$, as a measure of prior-likelihood conflict, i.e.,
harmful discordance. Our asymptotic regime permits the prior sample size to
grow with the likelihood data size, hence making asymptotic arguments
meaningful for investigating the impact of the prior relative to that of
likelihood. It leads to a simple asymptotic formula for quantifying the impact
of a proper prior that only involves computing a centrality and a spread
measure of the prior and the posterior. We use simulated and real data to
illustrate the potential of the proposed framework, including quantifying how
weak is a "weakly informative" prior adopted in a study of lupus nephritis.
Whereas we take a pragmatic perspective in assessing the impact of a prior on a
given inference problem under a specific evaluative metric, we also touch upon
conceptual and theoretical issues such as using improper priors and permitting
priors with asymptotically non-vanishing influence.
| stat |
Adaptive Importance Sampling for Efficient Stochastic Root Finding and
Quantile Estimation | In solving simulation-based stochastic root-finding or optimization problems
that involve rare events, such as in extreme quantile estimation, running crude
Monte Carlo can be prohibitively inefficient. To address this issue, importance
sampling can be employed to drive down the sampling error to a desirable level.
However, selecting a good importance sampler requires knowledge of the solution
to the problem at hand, which is the goal to begin with and thus forms a
circular challenge. We investigate the use of adaptive importance sampling to
untie this circularity. Our procedure sequentially updates the importance
sampler to reach the optimal sampler and the optimal solution simultaneously,
and can be embedded in both sample average approximation and stochastic
approximation-type algorithms. Our theoretical analysis establishes strong
consistency and asymptotic normality of the resulting estimators. We also
demonstrate, via a minimax perspective, the key role of using adaptivity in
controlling asymptotic errors. Finally, we illustrate the effectiveness of our
approach via numerical experiments.
| stat |
Using Embeddings to Correct for Unobserved Confounding in Networks | We consider causal inference in the presence of unobserved confounding. We
study the case where a proxy is available for the unobserved confounding in the
form of a network connecting the units. For example, the link structure of a
social network carries information about its members. We show how to
effectively use the proxy to do causal inference. The main idea is to reduce
the causal estimation problem to a semi-supervised prediction of both the
treatments and outcomes. Networks admit high-quality embedding models that can
be used for this semi-supervised prediction. We show that the method yields
valid inferences under suitable (weak) conditions on the quality of the
predictive model. We validate the method with experiments on a semi-synthetic
social network dataset. Code is available at
github.com/vveitch/causal-network-embeddings.
| stat |
Synthetic Difference in Differences | We present a new estimator for causal effects with panel data that builds on
insights behind the widely used difference in differences and synthetic control
methods. Relative to these methods, we find, both theoretically and
empirically, that the proposed "synthetic difference in differences" estimator
has desirable robustness properties, and that it performs well in settings
where the conventional estimators are commonly used in practice. We study the
asymptotic behavior of the estimator when the systematic part of the outcome
model includes latent unit factors interacted with latent time factors, and we
present conditions for consistency and asymptotic normality.
| stat |
Doubly Robust Semiparametric Inference Using Regularized Calibrated
Estimation with High-dimensional Data | Consider semiparametric estimation where a doubly robust estimating function
for a low-dimensional parameter is available, depending on two working models.
With high-dimensional data, we develop regularized calibrated estimation as a
general method for estimating the parameters in the two working models, such
that valid Wald confidence intervals can be obtained for the parameter of
interest under suitable sparsity conditions if either of the two working models
is correctly specified. We propose a computationally tractable two-step
algorithm and provide rigorous theoretical analysis which justifies
sufficiently fast rates of convergence for the regularized calibrated
estimators in spite of sequential construction and establishes a desired
asymptotic expansion for the doubly robust estimator. As concrete examples, we
discuss applications to partially linear, log-linear, and logistic models and
estimation of average treatment effects. Numerical studies in the former three
examples demonstrate superior performance of our method, compared with debiased
Lasso.
| stat |
Approximate Cross-Validation in High Dimensions with Guarantees | Leave-one-out cross-validation (LOOCV) can be particularly accurate among
cross-validation (CV) variants for machine learning assessment tasks -- e.g.,
assessing methods' error or variability. But it is expensive to re-fit a model
$N$ times for a dataset of size $N$. Previous work has shown that
approximations to LOOCV can be both fast and accurate -- when the unknown
parameter is of small, fixed dimension. But these approximations incur a
running time roughly cubic in dimension -- and we show that, besides
computational issues, their accuracy dramatically deteriorates in high
dimensions. Authors have suggested many potential and seemingly intuitive
solutions, but these methods have not yet been systematically evaluated or
compared. We find that all but one perform so poorly as to be unusable for
approximating LOOCV. Crucially, though, we are able to show, both empirically
and theoretically, that one approximation can perform well in high dimensions
-- in cases where the high-dimensional parameter exhibits sparsity. Under
interpretable assumptions, our theory demonstrates that the problem can be
reduced to working within an empirically recovered (small) support. This
procedure is straightforward to implement, and we prove that its running time
and error depend on the (small) support size even when the full parameter
dimension is large.
| stat |
A scalable hierarchical lasso for gene-environment interactions | We describe a regularized regression model for the selection of
gene-environment (GxE) interactions. The model focuses on a single
environmental exposure and induces a main-effect-before-interaction
hierarchical structure. We propose an efficient fitting algorithm and screening
rules that can discard large numbers of irrelevant predictors with high
accuracy. We present simulation results showing that the model outperforms
existing joint selection methods for (GxE) interactions in terms of selection
performance, scalability and speed, and provide a real data application. Our
implementation is available in the gesso R package.
| stat |
Visualisation of Brain Statistics with R-packages ggseg and ggseg3d | There is an increased emphasis on visualizing neuroimaging results in more
intuitive ways. Common statistical tools for dissemination, such as bar charts,
lack the spatial dimension that is inherent in neuroimaging data. Here we
present two packages for the statistical software R, ggseg and ggseg3d, that
integrate this spatial component. The ggseg and ggseg3d packages visualize
pre-defined brain segmentations as both 2D polygons and 3D meshes,
respectively. Both packages are integrated with other well-established
R-packages, allowing great flexibility. In this tutorial, we present the main
data and functions in the ggseg and ggseg3d packages for brain atlas
visualization. The main highlighted functions are able to display brain
segmentation plots in R. Further, the accompanying ggsegExtra-package includes
a wider collection of atlases, and is intended for community-based efforts to
develop more compatible atlases to ggseg and ggseg3d. Overall, the
ggseg-packages facilitate parcellation-based visualizations in R, improve and
ease the dissemination of the results, and increase the efficiency of the
workflows.
| stat |
Transporting treatment effects with incomplete attributes | The simultaneous availability of experimental and observational data to
estimate a treatment effect is both an opportunity and a statistical challenge:
Combining the information gathered from both data is a promising avenue to
build upon the internal validity of randomized controlled trials (RCTs) and a
greater external validity of observational data, but it raises methodological
issues, especially due to different sampling designs inducing distributional
shifts. We focus on the aim of transporting a causal effect estimated on an RCT
onto a target population described by a set of covariates. Available methods
such as inverse propensity weighting are not designed to handle missing values,
which are however common in both data. In addition to coupling the assumptions
for causal identifiability and for the missing values mechanism and to defining
appropriate strategies, one has to consider the specific structure of the data
with two sources and treatment and outcome only available in the RCT. We study
different approaches and their underlying assumptions on the data generating
processes and distribution of missing values and suggest several adapted
methods, in particular multiple imputation strategies. These methods are
assessed in an extensive simulation study and practical guidelines are provided
for different scenarios. This work is motivated by the analysis of a large
registry of over 20,000 major trauma patients and a multi-centered RCT studying
the effect of tranexamic acid administration on mortality. The analysis
illustrates how the missing values handling can impact the conclusion about the
effect transported from the RCT to the target population.
| stat |
Machine learning the real discriminant locus | Parameterized systems of polynomial equations arise in many applications in
science and engineering with the real solutions describing, for example,
equilibria of a dynamical system, linkages satisfying design constraints, and
scene reconstruction in computer vision. Since different parameter values can
have a different number of real solutions, the parameter space is decomposed
into regions whose boundary forms the real discriminant locus. This article
views locating the real discriminant locus as a supervised classification
problem in machine learning where the goal is to determine classification
boundaries over the parameter space, with the classes being the number of real
solutions. For multidimensional parameter spaces, this article presents a novel
sampling method which carefully samples the parameter space. At each sample
point, homotopy continuation is used to obtain the number of real solutions to
the corresponding polynomial system. Machine learning techniques including
nearest neighbor and deep learning are used to efficiently approximate the real
discriminant locus. One application of having learned the real discriminant
locus is to develop a real homotopy method that only tracks the real solution
paths unlike traditional methods which track all~complex~solution~paths.
Examples show that the proposed approach can efficiently approximate
complicated solution boundaries such as those arising from the equilibria of
the Kuramoto model.
| stat |
Asymmetric compressive learning guarantees with applications to
quantized sketches | The compressive learning framework reduces the computational cost of training
on large-scale datasets. In a sketching phase, the data is first compressed to
a lightweight sketch vector, obtained by mapping the data samples through a
well-chosen feature map, and averaging those contributions. In a learning
phase, the desired model parameters are then extracted from this sketch by
solving an optimization problem, which also involves a feature map. When the
feature map is identical during the sketching and learning phases, formal
statistical guarantees (excess risk bounds) have been proven.
However, the desirable properties of the feature map are different during
sketching and learning (e.g. quantized outputs, and differentiability,
respectively). We thus study the relaxation where this map is allowed to be
different for each phase. First, we prove that the existing guarantees carry
over to this asymmetric scheme, up to a controlled error term, provided some
Limited Projected Distortion (LPD) property holds. We then instantiate this
framework to the setting of quantized sketches, by proving that the LPD indeed
holds for binary sketch contributions. Finally, we further validate the
approach with numerical simulations, including a large-scale application in
audio event classification.
| stat |
Model Order Selection Based on Information Theoretic Criteria: Design of
the Penalty | Information theoretic criteria (ITC) have been widely adopted in engineering
and statistics for selecting, among an ordered set of candidate models, the one
that better fits the observed sample data. The selected model minimizes a
penalized likelihood metric, where the penalty is determined by the criterion
adopted. While rules for choosing a penalty that guarantees a consistent
estimate of the model order are known, theoretical tools for its design with
finite samples have never been provided in a general setting. In this paper, we
study model order selection for finite samples under a design perspective,
focusing on the generalized information criterion (GIC), which embraces the
most common ITC. The theory is general, and as case studies we consider: a) the
problem of estimating the number of signals embedded in additive white Gaussian
noise (AWGN) by using multiple sensors; b) model selection for the general
linear model (GLM), which includes e.g. the problem of estimating the number of
sinusoids in AWGN. The analysis reveals a trade-off between the probabilities
of overestimating and underestimating the order of the model. We then propose
to design the GIC penalty to minimize underestimation while keeping the
overestimation probability below a specified level. For the considered
problems, this method leads to analytical derivation of the optimal penalty for
a given sample size. A performance comparison between the penalty optimized GIC
and common AIC and BIC is provided, demonstrating the effectiveness of the
proposed design strategy.
| stat |
Generalized active learning and design of statistical experiments for
manifold-valued data | Characterizing the appearance of real-world surfaces is a fundamental problem
in multidimensional reflectometry, computer vision and computer graphics. For
many applications, appearance is sufficiently well characterized by the
bidirectional reflectance distribution function (BRDF). We treat BRDF
measurements as samples of points from high-dimensional non-linear non-convex
manifolds. BRDF manifolds form an infinite-dimensional space, but typically the
available measurements are very scarce for complicated problems such as BRDF
estimation. Therefore, an efficient learning strategy is crucial when
performing the measurements.
In this paper, we build the foundation of a mathematical framework that
allows to develop and apply new techniques within statistical design of
experiments and generalized proactive learning, in order to establish more
efficient sampling and measurement strategies for BRDF data manifolds.
| stat |
Akaike's Bayesian information criterion (ABIC) or not ABIC for
geophysical inversion | Akaike's Bayesian information criterion (ABIC) has been widely used in
geophysical inversion and beyond. However, little has been done to investigate
its statistical aspects. We present an alternative derivation of the marginal
distribution of measurements, whose maximization directly leads to the
invention of ABIC by Akaike. We show that ABIC is to statistically estimate the
variance of measurements and the prior variance by maximizing the marginal
distribution of measurements. The determination of the regularization parameter
on the basis of ABIC is actually equivalent to estimating the relative
weighting factor between the variance of measurements and the prior variance
for geophysical inverse problems. We show that if the noise level of
measurements is unknown, ABIC tends to produce a substantially biased estimate
of the variance of measurements. In particular, since the prior mean is
generally unknown but arbitrarily treated as zero in geophysical inversion,
ABIC does not produce a reasonable estimate for the prior variance either.
| stat |
Unlucky Number 13? Manipulating Evidence Subject to Snooping | Questionable research practices like HARKing or p-hacking have generated
considerable recent interest throughout and beyond the scientific community. We
subsume such practices involving secret data snooping that influences
subsequent statistical inference under the term MESSing (manipulating evidence
subject to snooping) and discuss, illustrate and quantify the possibly dramatic
effects of several forms of MESSing using an empirical and a simple theoretical
example. The empirical example uses numbers from the most popular German
lottery, which seem to suggest that 13 is an unlucky number.
| stat |
Representation formulas and pointwise properties for Barron functions | We study the natural function space for infinitely wide two-layer neural
networks and establish different representation formulae. In two cases, we
describe the space explicitly up to isomorphism. Using a convenient
representation, we study the pointwise properties of two-layer networks and
show that functions whose singular set is fractal or curved (for example
distance functions from smooth submanifolds) cannot be represented by
infinitely wide two-layer networks with finite path-norm.
| stat |
Proximal Learning for Individualized Treatment Regimes Under Unmeasured
Confounding | Data-driven individualized decision making has recently received increasing
research interests. Most existing methods rely on the assumption of no
unmeasured confounding, which unfortunately cannot be ensured in practice
especially in observational studies. Motivated by the recent proposed proximal
causal inference, we develop several proximal learning approaches to estimating
optimal individualized treatment regimes (ITRs) in the presence of unmeasured
confounding. In particular, we establish several identification results for
different classes of ITRs, exhibiting the trade-off between the risk of making
untestable assumptions and the value function improvement in decision making.
Based on these results, we propose several classification-based approaches to
finding a variety of restricted in-class optimal ITRs and develop their
theoretical properties. The appealing numerical performance of our proposed
methods is demonstrated via an extensive simulation study and one real data
application.
| stat |
Multiple imputation in data that grow over time: A comparison of three
strategies | Multiple imputation is a highly recommended technique to deal with missing
data, but the application to longitudinal datasets can be done in multiple
ways. When a new wave of longitudinal data arrives, we can treat the combined
data of multiple waves as a new missing data problem and overwrite existing
imputations with new values (re-imputation). Alternatively, we may keep the
existing imputations, and impute only the new data. We may do either a full
multiple imputation (nested) or a single imputation (appended) on the new data
per imputed set. This study compares these three strategies by means of
simulation. All techniques resulted in valid inference under a monotone
missingness pattern. A non-monotone missingness pattern led to biased and
non-confidence valid regression coefficients after nested and appended
imputation, depending on the correlation structure of the data. Correlations
within timepoints must be stronger than correlations between timepoints to
obtain valid inference. In an empirical example, the three strategies performed
similarly.We conclude that appended imputation is especially beneficial in
longitudinal datasets that suffer from dropout.
| stat |
A Note on a Simple and Practical Randomized Response Framework for
Eliciting Sensitive Dichotomous & Quantitative Information | Many issues of interest to social scientists and policymakers are of a
sensitive nature in the sense that they are intrusive, stigmatizing or
incriminating to the respondent. This results in refusals to cooperate or
evasive cooperation in studies using self-reports. In a seminal article Warner
proposed to curb this problem by generating an artificial variability in
responses to inoculate the individual meaning of answers to sensitive
questions. This procedure was further developed and extended, and came to be
known as the randomized response (RR) technique. Here, we propose a unified
treatment for eliciting sensitive binary as well as quantitative information
with RR based on a model where the inoculating elements are provided for by the
randomization device. The procedure is simple and we will argue that its
implementation in a computer-assisted setting may have superior practical
capabilities.
| stat |
Estimation of within-study covariances in multivariate meta-analysis | Multivariate meta-analysis can be adapted to a wide range of situations for
multiple outcomes and multiple treatment groups when combining studies
together. The within-study correlation between effect sizes is often assumed
known in multivariate meta-analysis while it is not always known practically.
In this paper, we propose a generic method to approximate the within-study
covariance for effect sizes in multivariate meta-analysis and apply this method
to the scenarios with multiple outcomes and one outcome with multiple treatment
groups respectively.
| stat |
Scalable Bayesian Multiple Changepoint Detection via Auxiliary
Uniformization | By attaching auxiliary event times to the chronologically ordered
observations, we formulate the Bayesian multiple changepoint problem of
discrete-time observations into that of continuous-time ones. A version of
forward-filtering backward-sampling (FFBS) algorithm is proposed for the
simulation of changepoints within a collapsed Gibbs sampling scheme. Ideally,
both the computational cost and memory cost of the FFBS algorithm can be
quadratically scaled down to the number of changepoints, instead of the number
of observations, which is otherwise prohibitive for a long sequence of
observations. The new formulation allows the number of changepoints accrue
unboundedly upon the arrivals of new data. Also, a time-varying changepoint
recurrence rate across different segments is assumed to characterize diverse
scales of run lengths of changepoints. We then suggest a continuous-time
Viterbi algorithm for obtaining the Maximum A Posteriori (MAP) estimates of
changepoints. We demonstrate the methods through simulation studies and real
data analysis.
| stat |
A spliced Gamma-Generalized Pareto model for short-term extreme wind
speed probabilistic forecasting | Renewable sources of energy such as wind power have become a sustainable
alternative to fossil fuel-based energy. However, the uncertainty and
fluctuation of the wind speed derived from its intermittent nature bring a
great threat to the wind power production stability, and to the wind turbines
themselves. Lately, much work has been done on developing models to forecast
average wind speed values, yet surprisingly little has focused on proposing
models to accurately forecast extreme wind speeds, which can damage the
turbines. In this work, we develop a flexible spliced Gamma-Generalized Pareto
model to forecast extreme and non-extreme wind speeds simultaneously. Our model
belongs to the class of latent Gaussian models, for which inference is
conveniently performed based on the integrated nested Laplace approximation
method. Considering a flexible additive regression structure, we propose two
models for the latent linear predictor to capture the spatio-temporal dynamics
of wind speeds. Our models are fast to fit and can describe both the bulk and
the tail of the wind speed distribution while producing short-term extreme and
non-extreme wind speed probabilistic forecasts.
| stat |
Unsupervised and Supervised Principal Component Analysis: Tutorial | This is a detailed tutorial paper which explains the Principal Component
Analysis (PCA), Supervised PCA (SPCA), kernel PCA, and kernel SPCA. We start
with projection, PCA with eigen-decomposition, PCA with one and multiple
projection directions, properties of the projection matrix, reconstruction
error minimization, and we connect to auto-encoder. Then, PCA with singular
value decomposition, dual PCA, and kernel PCA are covered. SPCA using both
scoring and Hilbert-Schmidt independence criterion are explained. Kernel SPCA
using both direct and dual approaches are then introduced. We cover all cases
of projection and reconstruction of training and out-of-sample data. Finally,
some simulations are provided on Frey and AT&T face datasets for verifying the
theory in practice.
| stat |
Sample size considerations for comparing dynamic treatment regimens in a
sequential multiple-assignment randomized trial with a continuous
longitudinal outcome | Clinicians and researchers alike are increasingly interested in how best to
personalize interventions. A dynamic treatment regimen (DTR) is a sequence of
pre-specified decision rules which can be used to guide the delivery of a
sequence of treatments or interventions that are tailored to the changing needs
of the individual. The sequential multiple-assignment randomized trial (SMART)
is a research tool which allows for the construction of effective DTRs. We
derive easy-to-use formulae for computing the total sample size for three
common two-stage SMART designs in which the primary aim is to compare mean
end-of-study outcomes for two embedded DTRs which recommend different
first-stage treatments. The formulae are derived in the context of a regression
model which leverages information from a longitudinal outcome collected over
the entire study. We show that the sample size formula for a SMART can be
written as the product of the sample size formula for a standard two-arm
randomized trial, a deflation factor that accounts for the increased
statistical efficiency resulting from a longitudinal analysis, and an inflation
factor that accounts for the design of a SMART. The SMART design inflation
factor is typically a function of the anticipated probability of response to
first-stage treatment. We review modeling and estimation for DTR effect
analyses using a longitudinal outcome from a SMART, as well as the estimation
of standard errors. We also present estimators for the covariance matrix for a
variety of common working correlation structures. Methods are motivated using
the ENGAGE study, a SMART aimed at developing a DTR for increasing motivation
to attend treatments among alcohol- and cocaine-dependent patients.
| stat |
Scalable Multiple Changepoint Detection for Functional Data Sequences | We propose the Multiple Changepoint Isolation (MCI) method for detecting
multiple changes in the mean and covariance of a functional process. We first
introduce a pair of projections to represent the high and low frequency
features of the data. We then apply total variation denoising and introduce a
new regionalization procedure to split the projections into multiple regions.
Denoising and regionalizing act to isolate each changepoint into its own
region, so that the classical univariate CUSUM statistic can be applied
region-wise to find all changepoints. Simulations show that our method
accurately detects the number and locations of changepoints under many
different scenarios. These include light and heavy tailed data, data with
symmetric and skewed distributions, sparsely and densely sampled changepoints,
and both mean and covariance changes. We show that our method outperforms a
recent multiple functional changepoint detector and several univariate
changepoint detectors applied to our proposed projections. We also show that
the MCI is more robust than existing approaches, and scales linearly with
sample size. Finally, we demonstrate our method on a large time series of water
vapor mixing ratio profiles from atmospheric emitted radiance interferometer
measurements.
| stat |
Spatial Spread Sampling Using Weakly Associated Vectors | Geographical data are generally autocorrelated. In this case, it is
preferable to select spread units. In this paper, we propose a new method for
selecting well-spread samples from a finite spatial population with equal or
unequal inclusion probabilities. The proposed method is based on the definition
of a spatial structure by using a stratification matrix. Our method exactly
satisfies given inclusion probabilities and provides samples that are very
well-spread. A set of simulations shows that our method outperforms other
existing methods such as the Generalized Random Tessellation Stratified (GRTS)
or the Local Pivotal Method (LPM). Analysis of the variance on a real dataset
shows that our method is more accurate than these two. Furthermore, a variance
estimator is proposed.
| stat |
Dynamic Window-level Granger Causality of Multi-channel Time Series | Granger causality method analyzes the time series causalities without
building a complex causality graph. However, the traditional Granger causality
method assumes that the causalities lie between time series channels and remain
constant, which cannot model the real-world time series data with dynamic
causalities along the time series channels. In this paper, we present the
dynamic window-level Granger causality method (DWGC) for multi-channel time
series data. We build the causality model on the window-level by doing the
F-test with the forecasting errors on the sliding windows. We propose the
causality indexing trick in our DWGC method to reweight the original time
series data. Essentially, the causality indexing is to decrease the
auto-correlation and increase the cross-correlation causal effects, which
improves the DWGC method. Theoretical analysis and experimental results on two
synthetic and one real-world datasets show that the improved DWGC method with
causality indexing better detects the window-level causalities.
| stat |
Selecting the Most Effective Nudge: Evidence from a Large-Scale
Experiment on Immunization | We evaluate a large-scale set of interventions to increase demand for
immunization in Haryana, India. The policies under consideration include the
two most frequently discussed tools--reminders and incentives--as well as an
intervention inspired by the networks literature. We cross-randomize whether
(a) individuals receive SMS reminders about upcoming vaccination drives; (b)
individuals receive incentives for vaccinating their children; (c) influential
individuals (information hubs, trusted individuals, or both) are asked to act
as "ambassadors" receiving regular reminders to spread the word about
immunization in their community. By taking into account different versions (or
"dosages") of each intervention, we obtain 75 unique policy combinations. We
develop a new statistical technique--a smart pooling and pruning procedure--for
finding a best policy from a large set, which also determines which policies
are effective and the effect of the best policy. We proceed in two steps.
First, we use a LASSO technique to collapse the data: we pool dosages of the
same treatment if the data cannot reject that they had the same impact, and
prune policies deemed ineffective. Second, using the remaining (pooled)
policies, we estimate the effect of the best policy, accounting for the
winner's curse. The key outcomes are (i) the number of measles immunizations
and (ii) the number of immunizations per dollar spent. The policy that has the
largest impact (information hubs, SMS reminders, incentives that increase with
each immunization) increases the number of immunizations by 44% relative to the
status quo. The most cost-effective policy (information hubs, SMS reminders, no
incentives) increases the number of immunizations per dollar by 9.1%.
| stat |
CP Degeneracy in Tensor Regression | Tensor linear regression is an important and useful tool for analyzing tensor
data. To deal with high dimensionality, CANDECOMP/PARAFAC (CP) low-rank
constraints are often imposed on the coefficient tensor parameter in the
(penalized) $M$-estimation. However, we show that the corresponding
optimization may not be attainable, and when this happens, the estimator is not
well-defined. This is closely related to a phenomenon, called CP degeneracy, in
low-rank tensor approximation problems. In this article, we provide useful
results of CP degeneracy in tensor regression problems. In addition, we provide
a general penalized strategy as a solution to overcome CP degeneracy. The
asymptotic properties of the resulting estimation are also studied. Numerical
experiments are conducted to illustrate our findings.
| stat |
Joint Curve Registration and Classification with Two-level Functional
Models | Many classification techniques when the data are curves or functions have
been recently proposed. However, the presence of misaligned problems in the
curves can influence the performance of most of them. In this paper, we propose
a model-based approach for simultaneous curve registration and classification.
The method is proposed to perform curve classification based on a functional
logistic regression model that relies on both scalar variables and functional
variables, and to align curves simultaneously via a data registration model.
EM-based algorithms are developed to perform maximum likelihood inference of
the proposed models. We establish the identifiability results for curve
registration model and investigate the asymptotic properties of the proposed
estimation procedures. Simulation studies are conducted to demonstrate the
finite sample performance of the proposed models. An application of the hyoid
bone movement data from stroke patients reveals the effectiveness of the new
models.
| stat |
Good Classifiers are Abundant in the Interpolating Regime | Within the machine learning community, the widely-used uniform convergence
framework has been used to answer the question of how complex,
over-parameterized models can generalize well to new data. This approach bounds
the test error of the worst-case model one could have fit to the data, but it
has fundamental limitations. Inspired by the statistical mechanics approach to
learning, we formally define and develop a methodology to compute precisely the
full distribution of test errors among interpolating classifiers from several
model classes. We apply our method to compute this distribution for several
real and synthetic datasets, with both linear and random feature classification
models. We find that test errors tend to concentrate around a small typical
value $\varepsilon^*$, which deviates substantially from the test error of the
worst-case interpolating model on the same datasets, indicating that "bad"
classifiers are extremely rare. We provide theoretical results in a simple
setting in which we characterize the full asymptotic distribution of test
errors, and we show that these indeed concentrate around a value
$\varepsilon^*$, which we also identify exactly. We then formalize a more
general conjecture supported by our empirical findings. Our results show that
the usual style of analysis in statistical learning theory may not be
fine-grained enough to capture the good generalization performance observed in
practice, and that approaches based on the statistical mechanics of learning
may offer a promising alternative.
| stat |
Learning Latent Space Energy-Based Prior Model | We propose to learn energy-based model (EBM) in the latent space of a
generator model, so that the EBM serves as a prior model that stands on the
top-down network of the generator model. Both the latent space EBM and the
top-down network can be learned jointly by maximum likelihood, which involves
short-run MCMC sampling from both the prior and posterior distributions of the
latent vector. Due to the low dimensionality of the latent space and the
expressiveness of the top-down network, a simple EBM in latent space can
capture regularities in the data effectively, and MCMC sampling in latent space
is efficient and mixes well. We show that the learned model exhibits strong
performances in terms of image and text generation and anomaly detection. The
one-page code can be found in supplementary materials.
| stat |
Improved Differentially Private Decentralized Source Separation for fMRI
Data | Blind source separation algorithms such as independent component analysis
(ICA) are widely used in the analysis of neuroimaging data. In order to
leverage larger sample sizes, different data holders/sites may wish to
collaboratively learn feature representations. However, such datasets are often
privacy-sensitive, precluding centralized analyses that pool the data at a
single site. In this work, we propose a differentially private algorithm for
performing ICA in a decentralized data setting. Conventional approaches to
decentralized differentially private algorithms may introduce too much noise
due to the typically small sample sizes at each site. We propose a novel
protocol that uses correlated noise to remedy this problem. We show that our
algorithm outperforms existing approaches on synthetic and real neuroimaging
datasets and demonstrate that it can sometimes reach the same level of utility
as the corresponding non-private algorithm. This indicates that it is possible
to have meaningful utility while preserving privacy.
| stat |
Quantile regression for compositional covariates | Quantile regression is a very important tool to explore the relationship
between the response variable and its covariates. Motivated by mean regression
with LASSO for compositional covariates proposed by Lin et al. (2014), we
consider quantile regression with no-penalty and penalty function. We develop
the computational algorithms based on linear programming. Numerical studies
indicate that our methods provides the better alternative than mean regression
under many settings, particularly for heavy-tailed or skewed distribution of
the error term. Finally, we study the fat data using the proposed method.
| stat |
Model enhancement and personalization using weakly supervised learning
for multi-modal mobile sensing | Always-on sensing of mobile device user's contextual information is critical
to many intelligent use cases nowadays such as healthcare, drive assistance,
voice UI. State-of-the-art approaches for predicting user context have proved
the value to leverage multiple sensing modalities for better accuracy. However,
those context inference algorithms that run on application processor nowadays
tend to drain heavy amount of power, making them not suitable for an always-on
implementation. We claim that not every sensing modality is suitable to be
activated all the time and it remains challenging to build an inference engine
using power friendly sensing modalities. Meanwhile, due to the diverse
population, we find it challenging to learn a context inference model that
generalizes well, with limited training data, especially when only using
always-on low power sensors. In this work, we propose an approach to leverage
the opportunistically-on counterparts in device to improve the always-on
prediction model, leading to a personalized solution. We model this problem
using a weakly supervised learning framework and provide both theoretical and
experimental results to validate our design. The proposed framework achieves
satisfying result in the IMU based activity recognition application we
considered.
| stat |
Restoration and extrapolation of structural transformation by dynamical
general equilibrium feedbacks | We model sectoral production by serially nesting (cascading) binary
compounding processes. The sequence of processes is discovered in a
self-similar hierarchical structure stylized in macroscopic input-output
transactions. The feedback system of unit cost functions, with recursively
estimated nest-wise CES parameters, is calibrated for sectoral productivities
to replicate two temporally distant cost share structures, observed in a set of
linked input--output tables. We model representative households by multifactor
CES, with parameters estimated by fixed effects regressions. By the integrated
dynamic general equilibrium model, we extrapolate potential structural
transformations, and measure the associated welfare changes, caused by
exogenous sectoral productivity shocks.
| stat |
Early Anomaly Detection in Time Series: A Hierarchical Approach for
Predicting Critical Health Episodes | The early detection of anomalous events in time series data is essential in
many domains of application. In this paper we deal with critical health events,
which represent a significant cause of mortality in intensive care units of
hospitals. The timely prediction of these events is crucial for mitigating
their consequences and improving healthcare. One of the most common approaches
to tackle early anomaly detection problems is standard classification methods.
In this paper we propose a novel method that uses a layered learning
architecture to address these tasks. One key contribution of our work is the
idea of pre-conditional events, which denote arbitrary but computable relaxed
versions of the event of interest. We leverage this idea to break the original
problem into two hierarchical layers, which we hypothesize are easier to solve.
The results suggest that the proposed approach leads to a better performance
relative to state of the art approaches for critical health episode prediction.
| stat |
A Swiss Army Infinitesimal Jackknife | The error or variability of machine learning algorithms is often assessed by
repeatedly re-fitting a model with different weighted versions of the observed
data. The ubiquitous tools of cross-validation (CV) and the bootstrap are
examples of this technique. These methods are powerful in large part due to
their model agnosticism but can be slow to run on modern, large data sets due
to the need to repeatedly re-fit the model. In this work, we use a linear
approximation to the dependence of the fitting procedure on the weights,
producing results that can be faster than repeated re-fitting by an order of
magnitude. This linear approximation is sometimes known as the "infinitesimal
jackknife" in the statistics literature, where it is mostly used as a
theoretical tool to prove asymptotic results. We provide explicit finite-sample
error bounds for the infinitesimal jackknife in terms of a small number of
simple, verifiable assumptions. Our results apply whether the weights and data
are stochastic or deterministic, and so can be used as a tool for proving the
accuracy of the infinitesimal jackknife on a wide variety of problems. As a
corollary, we state mild regularity conditions under which our approximation
consistently estimates true leave-$k$-out cross-validation for any fixed $k$.
These theoretical results, together with modern automatic differentiation
software, support the application of the infinitesimal jackknife to a wide
variety of practical problems in machine learning, providing a "Swiss Army
infinitesimal jackknife". We demonstrate the accuracy of our methods on a range
of simulated and real datasets.
| stat |
On the marginal likelihood and cross-validation | In Bayesian statistics, the marginal likelihood, also known as the evidence,
is used to evaluate model fit as it quantifies the joint probability of the
data under the prior. In contrast, non-Bayesian models are typically compared
using cross-validation on held-out data, either through $k$-fold partitioning
or leave-$p$-out subsampling. We show that the marginal likelihood is formally
equivalent to exhaustive leave-$p$-out cross-validation averaged over all
values of $p$ and all held-out test sets when using the log posterior
predictive probability as the scoring rule. Moreover, the log posterior
predictive is the only coherent scoring rule under data exchangeability. This
offers new insight into the marginal likelihood and cross-validation and
highlights the potential sensitivity of the marginal likelihood to the choice
of the prior. We suggest an alternative approach using cumulative
cross-validation following a preparatory training phase. Our work has
connections to prequential analysis and intrinsic Bayes factors but is
motivated through a different course.
| stat |
The statistical properties of RCTs and a proposal for shrinkage | We abstract the concept of a randomized controlled trial (RCT) as a triple
(beta,b,s), where beta is the primary efficacy parameter, b the estimate and s
the standard error (s>0). The parameter beta is either a difference of means, a
log odds ratio or a log hazard ratio. If we assume that b is unbiased and
normally distributed, then we can estimate the full joint distribution of
(beta,b,s) from a sample of pairs (b_i,s_i). We have collected 23,747 such
pairs from the Cochrane database to do so. Here, we report the estimated
distribution of the signal-to-noise ratio beta/s and the achieved power. We
estimate the median achieved power to be 0.13. We also consider the
exaggeration ratio which is the factor by which the magnitude of beta is
overestimated. We find that if the estimate is just significant at the 5%
level, we would expect it to overestimate the true effect by a factor of 1.7.
This exaggeration is sometimes referred to as the winner's curse and it is
undoubtedly to a considerable extent responsible for disappointing replication
results. For this reason, we believe it is important to shrink the unbiased
estimator, and we propose a method for doing so.
| stat |
Moving sum data segmentation for stochastics processes based on
invariance | The segmentation of data into stationary stretches also known as multiple
change point problem is important for many applications in time series analysis
as well as signal processing. Based on strong invariance principles, we analyse
data segmentation methodology using moving sum (MOSUM) statistics for a class
of regime-switching multivariate processes where each switch results in a
change in the drift. In particular, this framework includes the data
segmentation of multivariate partial sum, integrated diffusion and renewal
processes even if the distance between change points is sublinear. We study the
asymptotic behaviour of the corresponding change point estimators, show
consistency and derive the corresponding localisation rates which are minimax
optimal in a variety of situations including an unbounded number of changes in
Wiener processes with drift. Furthermore, we derive the limit distribution of
the change point estimators for local changes - a result that can in principle
be used to derive confidence intervals for the change points.
| stat |
Why Are the ARIMA and SARIMA not Sufficient | The autoregressive moving average (ARMA) model takes the significant position
in time series analysis for a wide-sense stationary time series. The difference
operator and seasonal difference operator, which are bases of ARIMA and SARIMA
(Seasonal ARIMA), respectively, were introduced to remove the trend and
seasonal component so that the original non-stationary time series could be
transformed into a wide-sense stationary one, which could then be handled by
Box-Jenkins methodology. However, such difference operators are more practical
experiences than exact theories by now. In this paper, we investigate the power
of the (resp. seasonal) difference operator from the perspective of spectral
analysis, linear system theory and digital filtering, and point out the
characteristics and limitations of (resp. seasonal) difference operator.
Besides, the general method that transforms a non-stationary (the
non-stationarity in the mean sense) stochastic process to be wide-sense
stationary will be presented.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.