title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Is it who you are or where you are? Accounting for compositional
differences in cross-site treatment variation | Multisite trials, in which treatment is randomized separately in multiple
sites, offer a unique opportunity to disentangle treatment effect variation due
to "compositional" differences in the distributions of unit-level features from
variation due to "contextual" differences in site-level features. In
particular, if we can re-weight (or "transport") each site to have a common
distribution of unit-level covariates, the remaining effect variation captures
contextual differences across sites. In this paper, we develop a framework for
transporting effects in multisite trials using approximate balancing weights,
where the weights are chosen to directly optimize unit-level covariate balance
between each site and the target distribution. We first develop our approach
for the general setting of transporting the effect of a single-site trial. We
then extend our method to multisite trials, assess its performance via
simulation, and use it to analyze a series of multisite trials of
welfare-to-work programs. Our method is available in the balancer R package.
| stat |
Analyzing Car Thefts and Recoveries with Connections to Modeling
Origin-Destination Point Patterns | For a given region, we have a dataset composed of car theft locations along
with a linked dataset of recovery locations which, due to partial recovery, is
a relatively small subset of the set of theft locations. For an investigator
seeking to understand the behavior of car thefts and recoveries in the region,
several questions are addressed. Viewing the set of theft locations as a point
pattern, can we propose useful models to explain the pattern? What types of
predictive models can be built to learn about recovery location given theft
location? Can the dependence between theft locations and recovery locations be
formalized? Can the flow between theft sites and recovery sites be captured?
Origin-destination modeling offers a natural framework for such problems.
However, here the data is not for areal units but rather is a pair of point
patterns, with the recovery point pattern only partially observed. We offer
modeling approaches for investigating the questions above and apply the
approaches to two datasets. One is small from the state of Neza in Mexico with
areal covariate information regarding population features and crime type. A
second, much larger one, is from Belo Horizonte in Brazil but lacks covariates.
| stat |
A Mixture of Linear-Linear Regression Models for Linear-Circular
Regression | We introduce a new approach to a linear-circular regression problem that
relates multiple linear predictors to a circular response. We follow a modeling
approach of a wrapped normal distribution that describes angular variables and
angular distributions and advances it for a linear-circular regression
analysis. Some previous works model a circular variable as projection of a
bivariate Gaussian random vector on the unit square, and the statistical
inference of the resulting model involves complicated sampling steps. The
proposed model treats circular responses as the result of the modulo operation
on unobserved linear responses. The resulting model is a mixture of multiple
linear-linear regression models. We present two EM algorithms for maximum
likelihood estimation of the mixture model, one for a parametric model and
another for a non-parametric model. The estimation algorithms provide a great
trade-off between computation and estimation accuracy, which was numerically
shown using five numerical examples. The proposed approach was applied to a
problem of estimating wind directions that typically exhibit complex patterns
with large variation and circularity.
| stat |
Rao-Blackwellization in the MCMC era | Rao-Blackwellization is a notion often occurring in the MCMC literature, with
possibly different meanings and connections with the original Rao--Blackwell
theorem (Rao, 1945 and Blackwell,1947), including a reduction of the variance
of the resulting Monte Carlo approximations. This survey reviews some of the
meanings of the term.
| stat |
Weighted Empirical Risk Minimization: Sample Selection Bias Correction
based on Importance Sampling | We consider statistical learning problems, when the distribution $P'$ of the
training observations $Z'_1,\; \ldots,\; Z'_n$ differs from the distribution
$P$ involved in the risk one seeks to minimize (referred to as the test
distribution) but is still defined on the same measurable space as $P$ and
dominates it. In the unrealistic case where the likelihood ratio
$\Phi(z)=dP/dP'(z)$ is known, one may straightforwardly extends the Empirical
Risk Minimization (ERM) approach to this specific transfer learning setup using
the same idea as that behind Importance Sampling, by minimizing a weighted
version of the empirical risk functional computed from the 'biased' training
data $Z'_i$ with weights $\Phi(Z'_i)$. Although the importance function
$\Phi(z)$ is generally unknown in practice, we show that, in various situations
frequently encountered in practice, it takes a simple form and can be directly
estimated from the $Z'_i$'s and some auxiliary information on the statistical
population $P$. By means of linearization techniques, we then prove that the
generalization capacity of the approach aforementioned is preserved when
plugging the resulting estimates of the $\Phi(Z'_i)$'s into the weighted
empirical risk. Beyond these theoretical guarantees, numerical results provide
strong empirical evidence of the relevance of the approach promoted in this
article.
| stat |
Causal Interaction Trees: Tree-Based Subgroup Identification for
Observational Data | We propose Causal Interaction Trees for identifying subgroups of participants
that have enhanced treatment effects using observational data. We extend the
Classification and Regression Tree algorithm by using splitting criteria that
focus on maximizing between-group treatment effect heterogeneity based on
subgroup-specific treatment effect estimators to dictate decision-making in the
algorithm. We derive properties of three subgroup-specific treatment effect
estimators that account for the observational nature of the data -- inverse
probability weighting, g-formula and doubly robust estimators. We study the
performance of the proposed algorithms using simulations and implement the
algorithms in an observational study that evaluates the effectiveness of right
heart catheterization on critically ill patients.
| stat |
Graphical Gaussian Process Models for Highly Multivariate Spatial Data | For multivariate spatial (Gaussian) process models, common cross-covariance
functions do not exploit graphical models to ensure process-level conditional
independence among the variables. This is undesirable, especially for highly
multivariate settings, where popular cross-covariance functions such as the
multivariate Mat\'ern suffer from a "curse of dimensionality" as the number of
parameters and floating point operations scale up in quadratic and cubic order,
respectively, in the number of variables. We propose a class of multivariate
"graphical Gaussian Processes" using a general construction called "stitching"
that crafts cross-covariance functions from graphs and ensure process-level
conditional independence among variables. For the Mat\'ern family of functions,
stitching yields a multivariate GP whose univariate components are exactly
Mat\'ern GPs, and conforms to process-level conditional independence as
specified by the graphical model. For highly multivariate settings and
decomposable graphical models, stitching offers massive computational gains and
parameter dimension reduction. We demonstrate the utility of the graphical
Mat\'ern GP to jointly model highly multivariate spatial data using simulation
examples and an application to air-pollution modelling.
| stat |
Towards Practical Lipschitz Bandits | Stochastic Lipschitz bandit algorithms balance exploration and exploitation,
and have been used for a variety of important task domains. In this paper, we
present a framework for Lipschitz bandit methods that adaptively learns
partitions of context- and arm-space. Due to this flexibility, the algorithm is
able to efficiently optimize rewards and minimize regret, by focusing on the
portions of the space that are most relevant. In our analysis, we link
tree-based methods to Gaussian processes. In light of our analysis, we design a
novel hierarchical Bayesian model for Lipschitz bandit problems. Our
experiments show that our algorithms can achieve state-of-the-art performance
in challenging real-world tasks such as neural network hyperparameter tuning.
| stat |
Estimation and Optimization of Composite Outcomes | There is tremendous interest in precision medicine as a means to improve
patient outcomes by tailoring treatment to individual characteristics. An
individualized treatment rule formalizes precision medicine as a map from
patient information to a recommended treatment. A treatment rule is defined to
be optimal if it maximizes the mean of a scalar outcome in a population of
interest, e.g., symptom reduction. However, clinical and intervention
scientists often must balance multiple and possibly competing outcomes, e.g.,
symptom reduction and the risk of an adverse event. One approach to precision
medicine in this setting is to elicit a composite outcome which balances all
competing outcomes; unfortunately, eliciting a composite outcome directly from
patients is difficult without a high-quality instrument, and an expert-derived
composite outcome may not account for heterogeneity in patient preferences. We
propose a new paradigm for the study of precision medicine using observational
data that relies solely on the assumption that clinicians are approximately
(i.e., imperfectly) making decisions to maximize individual patient utility.
Estimated composite outcomes are subsequently used to construct an estimator of
an individualized treatment rule which maximizes the mean of patient-specific
composite outcomes. The estimated composite outcomes and estimated optimal
individualized treatment rule provide new insights into patient preference
heterogeneity, clinician behavior, and the value of precision medicine in a
given domain. We derive inference procedures for the proposed estimators under
mild conditions and demonstrate their finite sample performance through a suite
of simulation experiments and an illustrative application to data from a study
of bipolar depression.
| stat |
BDgraph: An R Package for Bayesian Structure Learning in Graphical
Models | Graphical models provide powerful tools to uncover complicated patterns in
multivariate data and are commonly used in Bayesian statistics and machine
learning. In this paper, we introduce the R package BDgraph which performs
Bayesian structure learning for general undirected graphical models
(decomposable and non-decomposable) with continuous, discrete, and mixed
variables. The package efficiently implements recent improvements in the
Bayesian literature, including that of Mohammadi and Wit (2015) and Dobra and
Mohammadi (2018). To speed up computations, the computationally intensive tasks
have been implemented in C++ and interfaced with R, and the package has
parallel computing capabilities. In addition, the package contains several
functions for simulation and visualization, as well as several multivariate
datasets taken from the literature and used to describe the package
capabilities. The paper includes a brief overview of the statistical methods
which have been implemented in the package. The main part of the paper explains
how to use the package. Furthermore, we illustrate the package's functionality
in both real and artificial examples.
| stat |
Comparison of Machine Learning Models in Food Authentication Studies | The underlying objective of food authentication studies is to determine
whether unknown food samples have been correctly labelled. In this paper we
study three near infrared (NIR) spectroscopic datasets from food samples of
different types: meat samples (labelled by species), olive oil samples
(labelled by their geographic origin) and honey samples (labelled as pure or
adulterated by different adulterants). We apply and compare a large number of
classification, dimension reduction and variable selection approaches to these
datasets. NIR data pose specific challenges to classification and variable
selection: the datasets are high - dimensional where the number of cases ($n$)
$<<$ number of features ($p$) and the recorded features are highly serially
correlated. In this paper we carry out comparative analysis of different
approaches and find that partial least squares, a classic tool employed for
these types of data, outperforms all the other approaches considered.
| stat |
How Much Can We See? A Note on Quantifying Explainability of Machine
Learning Models | One of the most popular approaches to understanding feature effects of modern
black box machine learning models are partial dependence plots (PDP). These
plots are easy to understand but only able to visualize low order dependencies.
The paper is about the question 'How much can we see?': A framework is
developed to quantify the explainability of arbitrary machine learning models,
i.e. up to what degree the visualization as given by a PDP is able to explain
the predictions of the model. The result allows for a judgement whether an
attempt to explain a black box model is sufficient or not.
| stat |
Large-dimensional Factor Analysis without Moment Constraints | Large-dimensional factor model has drawn much attention in the big-data era,
in order to reduce the dimensionality and extract underlying features using a
few latent common factors. Conventional methods for estimating the factor model
typically requires finite fourth moment of the data, which ignores the effect
of heavy-tailedness and thus may result in unrobust or even inconsistent
estimation of the factor space and common components. In this paper, we propose
to recover the factor space by performing principal component analysis to the
spatial Kendall's tau matrix instead of the sample covariance matrix. In a
second step, we estimate the factor scores by the ordinary least square (OLS)
regression. Theoretically, we show that under the elliptical distribution
framework the factor loadings and scores as well as the common components can
be estimated consistently without any moment constraint. The convergence rates
of the estimated factor loadings, scores and common components are provided.
The finite sample performance of the proposed procedure is assessed through
thorough simulations. An analysis of a financial data set of asset returns
shows the superiority of the proposed method over the classical PCA method.
| stat |
Non-stationary Gaussian models with physical barriers | The classical tools in spatial statistics are stationary models, like the
Mat\'ern field. However, in some applications there are boundaries, holes, or
physical barriers in the study area, e.g. a coastline, and stationary models
will inappropriately smooth over these features, requiring the use of a
non-stationary model.
We propose a new model, the Barrier model, which is different from the
established methods as it is not based on the shortest distance around the
physical barrier, nor on boundary conditions. The Barrier model is based on
viewing the Mat\'ern correlation, not as a correlation function on the shortest
distance between two points, but as a collection of paths through a
Simultaneous Autoregressive (SAR) model. We then manipulate these local
dependencies to cut off paths that are crossing the physical barriers. To make
the new SAR well behaved, we formulate it as a stochastic partial differential
equation (SPDE) that can be discretised to represent the Gaussian field, with a
sparse precision matrix that is automatically positive definite.
The main advantage with the Barrier model is that the computational cost is
the same as for the stationary model. The model is easy to use, and can deal
with both sparse data and very complex barriers, as shown in an application in
the Finnish Archipelago Sea. Additionally, the Barrier model is better at
reconstructing the modified Horseshoe test function than the standard models
used in R-INLA.
| stat |
Quantitative Robustness of Localized Support Vector Machines | The huge amount of available data nowadays is a challenge for kernel-based
machine learning algorithms like SVMs with respect to runtime and storage
capacities. Local approaches might help to relieve these issues and to improve
statistical accuracy. It has already been shown that these local approaches are
consistent and robust in a basic sense. This article refines the analysis of
robustness properties towards the so-called influence function which expresses
the differentiability of the learning method: We show that there is a
differentiable dependency of our locally learned predictor on the underlying
distribution. The assumptions of the proven theorems can be verified without
knowing anything about this distribution. This makes the results interesting
also from an applied point of view.
| stat |
Learning Landmark-Based Ensembles with Random Fourier Features and
Gradient Boosting | We propose a Gradient Boosting algorithm for learning an ensemble of kernel
functions adapted to the task at hand. Unlike state-of-the-art Multiple Kernel
Learning techniques that make use of a pre-computed dictionary of kernel
functions to select from, at each iteration we fit a kernel by approximating it
as a weighted sum of Random Fourier Features (RFF) and by optimizing their
barycenter. This allows us to obtain a more versatile method, easier to setup
and likely to have better performance. Our study builds on a recent result
showing one can learn a kernel from RFF by computing the minimum of a
PAC-Bayesian bound on the kernel alignment generalization loss, which is
obtained efficiently from a closed-form solution. We conduct an experimental
analysis to highlight the advantages of our method w.r.t. both Boosting-based
and kernel-learning state-of-the-art methods.
| stat |
Subsampling Winner Algorithm for Feature Selection in Large Regression
Data | Feature selection from a large number of covariates (aka features) in a
regression analysis remains a challenge in data science, especially in terms of
its potential of scaling to ever-enlarging data and finding a group of
scientifically meaningful features. For example, to develop new, responsive
drug targets for ovarian cancer, the actual false discovery rate (FDR) of a
practical feature selection procedure must also match the target FDR. The
popular approach to feature selection, when true features are sparse, is to use
a penalized likelihood or a shrinkage estimation, such as a LASSO, SCAD,
Elastic Net, or MCP procedure (call them benchmark procedures). We present a
different approach using a new subsampling method, called the Subsampling
Winner algorithm (SWA). The central idea of SWA is analogous to that used for
the selection of US national merit scholars. SWA uses a "base procedure" to
analyze each of the subsamples, computes the scores of all features according
to the performance of each feature from all subsample analyses, obtains the
"semifinalist" based on the resulting scores, and then determines the
"finalists," i.e., the most important features. Due to its subsampling nature,
SWA can scale to data of any dimension in principle. The SWA also has the
best-controlled actual FDR in comparison with the benchmark procedures and the
randomForest, while having a competitive true-feature discovery rate. We also
suggest practical add-on strategies to SWA with or without a penalized
benchmark procedure to further assure the chance of "true" discovery. Our
application of SWA to the ovarian serous cystadenocarcinoma specimens from the
Broad Institute revealed functionally important genes and pathways, which we
verified by additional genomics tools. This second-stage investigation is
essential in the current discussion of the proper use of P-values.
| stat |
Data-driven discovery of coordinates and governing equations | The discovery of governing equations from scientific data has the potential
to transform data-rich fields that lack well-characterized quantitative
descriptions. Advances in sparse regression are currently enabling the
tractable identification of both the structure and parameters of a nonlinear
dynamical system from data. The resulting models have the fewest terms
necessary to describe the dynamics, balancing model complexity with descriptive
ability, and thus promoting interpretability and generalizability. This
provides an algorithmic approach to Occam's razor for model discovery. However,
this approach fundamentally relies on an effective coordinate system in which
the dynamics have a simple representation. In this work, we design a custom
autoencoder to discover a coordinate transformation into a reduced space where
the dynamics may be sparsely represented. Thus, we simultaneously learn the
governing equations and the associated coordinate system. We demonstrate this
approach on several example high-dimensional dynamical systems with
low-dimensional behavior. The resulting modeling framework combines the
strengths of deep neural networks for flexible representation and sparse
identification of nonlinear dynamics (SINDy) for parsimonious models. It is the
first method of its kind to place the discovery of coordinates and models on an
equal footing.
| stat |
Evaluating stochastic seeding strategies in networks | When trying to maximize the adoption of a behavior in a population connected
by a social network, it is common to strategize about where in the network to
seed the behavior, often with an element of randomness. Selecting seeds
uniformly at random is a basic but compelling strategy in that it distributes
seeds broadly throughout the network. A more sophisticated stochastic strategy,
one-hop targeting, is to select random network neighbors of random individuals;
this exploits a version of the friendship paradox, whereby the friend of a
random individual is expected to have more friends than a random individual,
with the hope that seeding a behavior at more connected individuals leads to
more adoption. Many seeding strategies have been proposed, but empirical
evaluations have demanded large field experiments designed specifically for
this purpose and have yielded relatively imprecise comparisons of strategies.
Here we show how stochastic seeding strategies can be evaluated more
efficiently in such experiments, how they can be evaluated "off-policy" using
existing data arising from experiments designed for other purposes, and how to
design more efficient experiments. In particular, we consider contrasts between
stochastic seeding strategies and analyze nonparametric estimators adapted from
policy evaluation and importance sampling. We use simulations on real networks
to show that the proposed estimators and designs can increase precision while
yielding valid inference. We then apply our proposed estimators to two field
experiments, one that assigned households to an intensive marketing
intervention and one that assigned students to an anti-bullying intervention.
| stat |
A Measurement of In-Betweenness and Inference Based on Shape Theories | We propose a statistical framework to investigate whether a given
subpopulation lies between two other subpopulations in a multivariate feature
space. This methodology is motivated by a biological question from a
collaborator: Is a newly discovered cell type between two known types in
several given features? We propose two in-betweenness indices (IBI) to quantify
the in-betweenness exhibited by a random triangle formed by the summary
statistics of the three subpopulations. Statistical inference methods are
provided for triangle shape and IBI metrics. The application of our methods is
demonstrated in three examples: the classic Iris data set, a study of risk of
relapse across three breast cancer subtypes, and the motivating neuronal cell
data with measured electrophysiological features.
| stat |
What Kinds of Functions do Deep Neural Networks Learn? Insights from
Variational Spline Theory | We develop a variational framework to understand the properties of functions
learned by deep neural networks with ReLU activation functions fit to data. We
propose a new function space, which is reminiscent of classical bounded
variation spaces, that captures the compositional structure associated with
deep neural networks. We derive a representer theorem showing that deep ReLU
networks are solutions to regularized data fitting problems in this function
space. The function space consists of compositions of functions from the
(non-reflexive) Banach spaces of second-order bounded variation in the Radon
domain. These are Banach spaces with sparsity-promoting norms, giving insight
into the role of sparsity in deep neural networks. The neural network solutions
have skip connections and rank bounded weight matrices, providing new
theoretical support for these common architectural choices. The variational
problem we study can be recast as a finite-dimensional neural network training
problem with regularization schemes related to the notions of weight decay and
path-norm regularization. Finally, our analysis builds on techniques from
variational spline theory, providing new connections between deep neural
networks and splines.
| stat |
On the Regularity of Attention | Attention is a powerful component of modern neural networks across a wide
variety of domains. In this paper, we seek to quantify the regularity (i.e. the
amount of smoothness) of the attention operation. To accomplish this goal, we
propose a new mathematical framework that uses measure theory and integral
operators to model attention. We show that this framework is consistent with
the usual definition, and that it captures the essential properties of
attention. Then we use this framework to prove that, on compact domains, the
attention operation is Lipschitz continuous and provide an estimate of its
Lipschitz constant. Additionally, by focusing on a specific type of attention,
we extend these Lipschitz continuity results to non-compact domains. We also
discuss the effects regularity can have on NLP models, and applications to
invertible and infinitely-deep networks.
| stat |
Multivariate Cluster Weighted Models Using Skewed Distributions | Much work has been done in the area of the cluster weighted model (CWM),
which extends the finite mixture of regression model to include modelling of
the covariates. Although many types of distributions have been considered for
both the response and covariates, to our knowledge skewed distributions have
not yet been considered in this paradigm. Herein, a family of 24 novel CWMs are
considered which allows both the covariates and response variables to be
modelled using one of four skewed distributions, or the normal distribution.
Parameter estimation is performed using the expectation-maximization algorithm
and both simulated and real data are used for illustration.
| stat |
Particle Smoothing Variational Objectives | A body of recent work has focused on constructing a variational family of
filtered distributions using Sequential Monte Carlo (SMC). Inspired by this
work, we introduce Particle Smoothing Variational Objectives (SVO), a novel
backward simulation technique and smoothed approximate posterior defined
through a subsampling process. SVO augments support of the proposal and boosts
particle diversity. Recent literature argues that increasing the number of
samples K to obtain tighter variational bounds may hurt the proposal learning,
due to a signal-to-noise ratio (SNR) of gradient estimators decreasing at the
rate $\mathcal{O}(1/\sqrt{K})$. As a second contribution, we develop
theoretical and empirical analysis of the SNR in filtering SMC, which motivates
our choice of biased gradient estimators. We prove that introducing bias by
dropping Categorical terms from the gradient estimate or using Gumbel-Softmax
mitigates the adverse effect on the SNR. We apply SVO to three nonlinear latent
dynamics tasks and provide statistics to rigorously quantify the predictions of
filtered and smoothed objectives. SVO consistently outperforms filtered
objectives when given fewer Monte Carlo samples on three nonlinear systems of
increasing complexity.
| stat |
Infrastructure Recovery Curve Estimation Using Gaussian Process
Regression on Expert Elicited Data | Infrastructure recovery time estimation is critical to disaster management
and planning. Inspired by recent resilience planning initiatives, we consider a
situation where experts are asked to estimate the time for different
infrastructure systems to recover to certain functionality levels after a
scenario hazard event. We propose a methodological framework to use
expert-elicited data to estimate the expected recovery time curve of a
particular infrastructure system. This framework uses the Gaussian process
regression (GPR) to capture the experts' estimation-uncertainty and satisfy
known physical constraints of recovery processes. The framework is designed to
find a balance between the data collection cost of expert elicitation and the
prediction accuracy of GPR. We evaluate the framework on realistically
simulated expert-elicited data concerning the two case study events, the 1995
Great Hanshin-Awaji Earthquake and the 2011 Great East Japan Earthquake.
| stat |
Regression Analysis of Unmeasured Confounding | When studying the causal effect of $x$ on $y$, researchers may conduct
regression and report a confidence interval for the slope coefficient
$\beta_{x}$. This common confidence interval provides an assessment of
uncertainty from sampling error, but it does not assess uncertainty from
confounding. An intervention on $x$ may produce a response in $y$ that is
unexpected, and our misinterpretation of the slope happens when there are
confounding factors $w$. When $w$ are measured we may conduct multiple
regression, but when $w$ are unmeasured it is common practice to include a
precautionary statement when reporting the confidence interval, warning against
unwarranted causal interpretation. If the goal is robust causal interpretation
then we can do something more informative. Uncertainty in the specification of
three confounding parameters can be propagated through an equation to produce a
confounding interval. Here we develop supporting mathematical theory and
describe an example application. Our proposed methodology applies well to
studies of a continuous response or rare outcome. It is a general method for
quantifying error from model uncertainty. Whereas confidence intervals are used
to assess uncertainty from unmeasured individuals, confounding intervals can be
used to assess uncertainty from unmeasured attributes.
| stat |
Nonparametric Estimation of Low Rank Matrix Valued Function | Let $A:[0,1]\rightarrow\mathbb{H}_m$ (the space of Hermitian matrices) be a
matrix valued function which is low rank with entries in H\"{o}lder class
$\Sigma(\beta,L)$. The goal of this paper is to study statistical estimation of
$A$ based on the regression model $\mathbb{E}(Y_j|\tau_j,X_j) = \langle
A(\tau_j), X_j \rangle,$ where $\tau_j$ are i.i.d. uniformly distributed in
$[0,1]$, $X_j$ are i.i.d. matrix completion sampling matrices, $Y_j$ are
independent bounded responses. We propose an innovative nuclear norm penalized
local polynomial estimator and establish an upper bound on its point-wise risk
measured by Frobenius norm. Then we extend this estimator globally and prove an
upper bound on its integrated risk measured by $L_2$-norm. We also propose
another new estimator based on bias-reducing kernels to study the case when $A$
is not necessarily low rank and establish an upper bound on its risk measured
by $L_{\infty}$-norm. We show that the obtained rates are all optimal up to
some logarithmic factor in minimax sense. Finally, we propose an adaptive
estimation procedure based on Lepskii's method and model selection with data
splitting which is computationally efficient and can be easily implemented and
parallelized.
| stat |
Nonparametric sequential change-point detection for multivariate time
series based on empirical distribution functions | The aim of sequential change-point detection is to issue an alarm when it is
thought that certain probabilistic properties of the monitored observations
have changed. This work is concerned with nonparametric, closed-end testing
procedures based on differences of empirical distribution functions that are
designed to be particularly sensitive to changes in the comtemporary
distribution of multivariate time series. The proposed detectors are
adaptations of statistics used in a posteriori (offline) change-point testing
and involve a weighting allowing to give more importance to recent
observations. The resulting sequential change-point detection procedures are
carried out by comparing the detectors to threshold functions estimated through
resampling such that the probability of false alarm remains approximately
constant over the monitoring period. A generic result on the asymptotic
validity of such a way of estimating a threshold function is stated. As a
corollary, the asymptotic validity of the studied sequential tests based on
empirical distribution functions is proven when these are carried out using a
dependent multiplier bootstrap for multivariate time series. Large-scale Monte
Carlo experiments demonstrate the good finite-sample properties of the
resulting procedures. The application of the derived sequential tests is
illustrated on financial data.
| stat |
Functional probabilistic programming for scalable Bayesian modelling | Bayesian inference involves the specification of a statistical model by a
statistician or practitioner, with careful thought about what each parameter
represents. This results in particularly interpretable models which can be used
to explain relationships present in the observed data. Bayesian models are
useful when an experiment has only a small number of observations and in
applications where transparency of data driven decisions is important.
Traditionally, parameter inference in Bayesian statistics has involved
constructing bespoke MCMC (Markov chain Monte Carlo) schemes for each newly
proposed statistical model. This results in plausible models not being
considered since efficient inference schemes are challenging to develop or
implement. Probabilistic programming aims to reduce the barrier to performing
Bayesian inference by developing a domain specific language (DSL) for model
specification which is decoupled from the parameter inference algorithms. This
paper introduces functional programming principles which can be used to develop
an embedded probabilistic programming language. Model inference can be carried
out using any generic inference algorithm. In this paper Hamiltonian Monte
Carlo (HMC) is used, an efficient MCMC method requiring the gradient of the
un-normalised log-posterior, calculated using automatic differentiation. The
concepts are illustrated using the Scala programming language.
| stat |
Grouping effects of sparse CCA models in variable selection | The sparse canonical correlation analysis (SCCA) is a bi-multivariate
association model that finds sparse linear combinations of two sets of
variables that are maximally correlated with each other. In addition to the
standard SCCA model, a simplified SCCA criterion which maixmizes the
cross-covariance between a pair of canonical variables instead of their
cross-correlation, is widely used in the literature due to its computational
simplicity. However, the behaviors/properties of the solutions of these two
models remain unknown in theory. In this paper, we analyze the grouping effect
of the standard and simplified SCCA models in variable selection. In
high-dimensional settings, the variables often form groups with high
within-group correlation and low between-group correlation. Our theoretical
analysis shows that for grouped variable selection, the simplified SCCA jointly
selects or deselects a group of variables together, while the standard SCCA
randomly selects a few dominant variables from each relevant group of
correlated variables. Empirical results on synthetic data and real imaging
genetics data verify the finding of our theoretical analysis.
| stat |
Using Experimental Data to Evaluate Methods for Observational Causal
Inference | Methods that infer causal dependence from observational data are central to
many areas of science, including medicine, economics, and the social sciences.
A variety of theoretical properties of these methods have been proven, but
empirical evaluation remains a challenge, largely due to the lack of
observational data sets for which treatment effect is known. We propose and
analyze observational sampling from randomized controlled trials (OSRCT), a
method for evaluating causal inference methods using data from randomized
controlled trials (RCTs). This method can be used to create constructed
observational data sets with corresponding unbiased estimates of treatment
effect, substantially increasing the number of data sets available for
evaluating causal inference methods. We show that, in expectation, OSRCT
creates data sets that are equivalent to those produced by randomly sampling
from empirical data sets in which all potential outcomes are available. We
analyze several properties of OSRCT theoretically and empirically, and we
demonstrate its use by comparing the performance of four causal inference
methods using data from eleven RCTs.
| stat |
The revisited knockoffs method for variable selection in L1-penalised
regressions | We consider the problem of variable selection in regression models. In
particular, we are interested in selecting explanatory covariates linked with
the response variable and we want to determine which covariates are relevant,
that is which covariates are involved in the model. In this framework, we deal
with L1-penalised regression models. To handle the choice of the penalty
parameter to perform variable selection, we develop a new method based on the
knockoffs idea. This revisited knockoffs method is general, suitable for a wide
range of regressions with various types of response variables. Besides, it also
works when the number of observations is smaller than the number of covariates
and gives an order of importance of the covariates. Finally, we provide many
experimental results to corroborate our method and compare it with other
variable selection methods.
| stat |
Emulating computationally expensive dynamical simulators using Gaussian
processes | A Gaussian process (GP)-based methodology is proposed to emulate
computationally expensive dynamical computer models or simulators. The method
relies on emulating the short-time numerical flow map of the model. The flow
map returns the solution of a dynamic system at an arbitrary time for a given
initial condition. The prediction of the flow map is performed via a GP whose
kernel is estimated using random Fourier features. This gives a distribution
over the flow map such that each realisation serves as an approximation to the
flow map. A realisation is then employed in an iterative manner to perform
one-step ahead predictions and forecast the whole time series. Repeating this
procedure with multiple draws from the emulated flow map provides a probability
distribution over the time series. The mean and variance of that distribution
are used as the model output prediction and a measure of the associated
uncertainty, respectively. The proposed method is used to emulate several
dynamic non-linear simulators including the well-known Lorenz attractor and van
der Pol oscillator. The results show that our approach has a high prediction
performance in emulating such systems with an accurate representation of the
prediction uncertainty.
| stat |
Bayesian Nonparametrics for Directional Statistics | We introduce a density basis of the trigonometric polynomials that is
suitable to mixture modelling. Statistical and geometric properties are
derived, suggesting it as a circular analogue to the Bernstein polynomial
densities. Nonparametric priors are constructed using this basis and a
simulation study shows that the use of the resulting Bayes estimator may
provide gains over comparable circular density estimators previously suggested
in the literature. From a theoretical point of view, we propose a general prior
specification framework for density estimation on compact metric space using
sieve priors. This is tailored to density bases such as the one considered
herein and may also be used to exploit their particular shape-preserving
properties. Furthermore, strong posterior consistency is shown to hold under
notably weak regularity assumptions and adaptative convergence rates are
obtained in terms of the approximation properties of positive linear operators
generating our models.
| stat |
On MCMC for variationally sparse Gaussian processes: A pseudo-marginal
approach | Gaussian processes (GPs) are frequently used in machine learning and
statistics to construct powerful models. However, when employing GPs in
practice, important considerations must be made, regarding the high
computational burden, approximation of the posterior, choice of the covariance
function and inference of its hyperparmeters. To address these issues, Hensman
et al. (2015) combine variationally sparse GPs with Markov chain Monte Carlo
(MCMC) to derive a scalable, flexible and general framework for GP models.
Nevertheless, the resulting approach requires intractable likelihood
evaluations for many observation models. To bypass this problem, we propose a
pseudo-marginal (PM) scheme that offers asymptotically exact inference as well
as computational gains through doubly stochastic estimators for the intractable
likelihood and large datasets. In complex models, the advantages of the PM
scheme are particularly evident, and we demonstrate this on a two-level GP
regression model with a nonparametric covariance function to capture
non-stationarity.
| stat |
A Stochastic Covariance Shrinkage Approach in Ensemble Transform Kalman
Filtering | The Ensemble Kalman Filters (EnKF) employ a Monte-Carlo approach to represent
covariance information, and are affected by sampling errors in operational
settings where the number of model realizations is much smaller than the model
state dimension. To alleviate the effects of these errors EnKF relies on
model-specific heuristics such as covariance localization, which takes
advantage of the spatial locality of correlations among the model variables.
This work proposes an approach to alleviate sampling errors that utilizes a
locally averaged-in-time dynamics of the model, described in terms of a
climatological covariance of the dynamical system. We use this covariance as
the target matrix in covariance shrinkage methods, and develop a stochastic
covariance shrinkage approach where synthetic ensemble members are drawn to
enrich both the ensemble subspace and the ensemble transformation.
| stat |
On Training and Evaluation of Neural Network Approaches for Model
Predictive Control | The contribution of this paper is a framework for training and evaluation of
Model Predictive Control (MPC) implemented using constrained neural networks.
Recent studies have proposed to use neural networks with differentiable convex
optimization layers to implement model predictive controllers. The motivation
is to replace real-time optimization in safety critical feedback control
systems with learnt mappings in the form of neural networks with optimization
layers. Such mappings take as the input the state vector and predict the
control law as the output. The learning takes place using training data
generated from off-line MPC simulations. However, a general framework for
characterization of learning approaches in terms of both model validation and
efficient training data generation is lacking in literature. In this paper, we
take the first steps towards developing such a coherent framework. We discuss
how the learning problem has similarities with system identification, in
particular input design, model structure selection and model validation. We
consider the study of neural network architectures in PyTorch with the explicit
MPC constraints implemented as a differentiable optimization layer using CVXPY.
We propose an efficient approach of generating MPC input samples subject to the
MPC model constraints using a hit-and-run sampler. The corresponding true
outputs are generated by solving the MPC offline using OSOP. We propose
different metrics to validate the resulting approaches. Our study further aims
to explore the advantages of incorporating domain knowledge into the network
structure from a training and evaluation perspective. Different model
structures are numerically tested using the proposed framework in order to
obtain more insights in the properties of constrained neural networks based
MPC.
| stat |
Measurement bias: a structural perspective | The causal structure for measurement bias (MB) remains controversial. Aided
by the Directed Acyclic Graph (DAG), this paper proposes a new structure for
measuring one singleton variable whose MB arises in the selection of an
imperfect I/O device-like measurement system. For effect estimation, however,
an extra source of MB arises from any redundant association between a measured
exposure and a measured outcome. The misclassification will be bidirectionally
differential for a common outcome, unidirectionally differential for a causal
relation, and non-differential for a common cause between the measured exposure
and the measured outcome or a null effect. The measured exposure can actually
affect the measured outcome, or vice versa. Reverse causality is a concept
defined at the level of measurement. Our new DAGs have clarified the structures
and mechanisms of MB.
| stat |
Stein Variational Gradient Descent: A General Purpose Bayesian Inference
Algorithm | We propose a general purpose variational inference algorithm that forms a
natural counterpart of gradient descent for optimization. Our method
iteratively transports a set of particles to match the target distribution, by
applying a form of functional gradient descent that minimizes the KL
divergence. Empirical studies are performed on various real world models and
datasets, on which our method is competitive with existing state-of-the-art
methods. The derivation of our method is based on a new theoretical result that
connects the derivative of KL divergence under smooth transforms with Stein's
identity and a recently proposed kernelized Stein discrepancy, which is of
independent interest.
| stat |
Multi-Objective Counterfactual Explanations | Counterfactual explanations are one of the most popular methods to make
predictions of black box machine learning models interpretable by providing
explanations in the form of `what-if scenarios'. Most current approaches
optimize a collapsed, weighted sum of multiple objectives, which are naturally
difficult to balance a-priori. We propose the Multi-Objective Counterfactuals
(MOC) method, which translates the counterfactual search into a multi-objective
optimization problem. Our approach not only returns a diverse set of
counterfactuals with different trade-offs between the proposed objectives, but
also maintains diversity in feature space. This enables a more detailed
post-hoc analysis to facilitate better understanding and also more options for
actionable user responses to change the predicted outcome. Our approach is also
model-agnostic and works for numerical and categorical input features. We show
the usefulness of MOC in concrete cases and compare our approach with
state-of-the-art methods for counterfactual explanations.
| stat |
Estimating state occupation and transition probabilities in non-Markov
multi-state models subject to both random left-truncation and right-censoring | The Aalen-Johansen estimator generalizes the Kaplan-Meier estimator for
independently left-truncated and right-censored survival data to estimating the
transition probability matrix of a time-inhomogeneous Markov model with finite
state space. Such multi-state models have a wide range of applications for
modelling complex courses of a disease over the course of time, but the Markov
assumption may often be in doubt. If censoring is entirely unrelated to the
multi-state data, it has been noted that the Aalen-Johansen estimator,
standardized by the initial empirical distribution of the multi-state model,
still consistently estimates the state occupation probabilities. Recently, this
result has been extended to transition probabilities using landmarking, which
is, inter alia, useful for dynamic prediction. We complement these results in
three ways. Firstly, delayed study entry is a common phenomenon in
observational studies, and we extend the earlier results to multi-state data
also subject to left-truncation. Secondly, we present a rigorous proof of
consistency of the Aalen-Johansen estimator for state occupation probabilities,
on which also correctness of the landmarking approach hinges, correcting,
simplifying and extending the earlier result. Thirdly, our rigorous proof
motivates wild bootstrap resampling. Our developments for left-truncation are
motivated by a prospective observational study on the occurrence and the impact
of a multi-resistant infectious organism in patients undergoing surgery. Both
the real data example and simulation studies are presented. Studying wild
bootstrap is motivated by the fact that, unlike drawing with replacement from
the data, it is desirable to have a technique that works both with non-Markov
models subject to random left-truncation and right-censoring and with Markov
models where left-truncation and right-censoring need not be entirely random.
| stat |
Online Stochastic Gradient Descent with Arbitrary Initialization Solves
Non-smooth, Non-convex Phase Retrieval | In recent literature, a general two step procedure has been formulated for
solving the problem of phase retrieval. First, a spectral technique is used to
obtain a constant-error initial estimate, following which, the estimate is
refined to arbitrary precision by first-order optimization of a non-convex loss
function. Numerical experiments, however, seem to suggest that simply running
the iterative schemes from a random initialization may also lead to
convergence, albeit at the cost of slightly higher sample complexity. In this
paper, we prove that, in fact, constant step size online stochastic gradient
descent (SGD) converges from arbitrary initializations for the non-smooth,
non-convex amplitude squared loss objective. In this setting, online SGD is
also equivalent to the randomized Kaczmarz algorithm from numerical analysis.
Our analysis can easily be generalized to other single index models. It also
makes use of new ideas from stochastic process theory, including the notion of
a summary state space, which we believe will be of use for the broader field of
non-convex optimization.
| stat |
Multi-Goal Prior Selection: A Way to Reconcile Bayesian and Classical
Approaches for Random Effects Models | The two-level normal hierarchical model has played an important role in
statistical theory and applications. In this paper, we first introduce a
general adjusted maximum likelihood method for estimating the unknown variance
component of the model and the associated empirical best linear unbiased
predictor of the random effects. We then discuss a new idea for selecting prior
for the hyperparameters. The prior, called a multi-goal prior, produces
Bayesian solutions for hyperparmeters and random effects that match (in the
higher order asymptotic sense) the corresponding classical solution in linear
mixed model with respect to several properties. Moreover, we establish for the
first time an analytical equivalence of the posterior variances under the
proposed multi-goal prior and the corresponding parametric bootstrap
second-order mean squared error estimates in the context of a random effects
model.
| stat |
Covariate Balancing by Uniform Transformer | In observational studies, it is important to balance covariates in different
treatment groups in order to estimate treatment effects. One of the most
commonly used methods for such purpose is the weighting method. The performance
quality of this method usually depends on either the correct model
specification for the propensity score or strong regularity conditions for the
underlying model, which might not hold in practice. In this paper, we introduce
a new robust and computationally efficient framework of weighting methods for
covariate balancing, which allows us to conduct model-free inferences for the
sake of robustness and integrate an extra `unlabeled' data set if available.
Unlike existing methods, the new framework reduces the weights construction
problem to a classical density estimation problem by applying a data-driven
transformation to the observed covariates. We characterize the theoretical
properties of the new estimators of average treatment effect under a
nonparametric setting and show that they are able to work robustly under low
regularity conditions. The new framework is also applied to several numerical
examples using both simulated and real datasets to demonstrate its practical
merits.
| stat |
Classification with Valid and Adaptive Coverage | Conformal inference, cross-validation+, and the jackknife+ are hold-out
methods that can be combined with virtually any machine learning algorithm to
construct prediction sets with guaranteed marginal coverage. In this paper, we
develop specialized versions of these techniques for categorical and unordered
response labels that, in addition to providing marginal coverage, are also
fully adaptive to complex data distributions, in the sense that they perform
favorably in terms of approximate conditional coverage compared to alternative
methods. The heart of our contribution is a novel conformity score, which we
explicitly demonstrate to be powerful and intuitive for classification
problems, but whose underlying principle is potentially far more general.
Experiments on synthetic and real data demonstrate the practical value of our
theoretical guarantees, as well as the statistical advantages of the proposed
methods over the existing alternatives.
| stat |
Selective Inference in Propensity Score Analysis | Selective inference (post-selection inference) is a methodology that has
attracted much attention in recent years in the fields of statistics and
machine learning. Naive inference based on data that are also used for model
selection tends to show an overestimation, and so the selective inference
conditions the event that the model was selected. In this paper, we develop
selective inference in propensity score analysis with a semiparametric
approach, which has become a standard tool in causal inference. Specifically,
for the most basic causal inference model in which the causal effect can be
written as a linear sum of confounding variables, we conduct Lasso-type
variable selection by adding an $\ell_1$ penalty term to the loss function that
gives a semiparametric estimator. Confidence intervals are then given for the
coefficients of the selected confounding variables, conditional on the event of
variable selection, with asymptotic guarantees. An important property of this
method is that it does not require modeling of nonparametric regression
functions for the outcome variables, as is usually the case with semiparametric
propensity score analysis.
| stat |
Constrained inference through posterior projections | Bayesian approaches are appealing for constrained inference problems in
allowing a probabilistic characterization of uncertainty, while providing a
computational machinery for incorporating complex constraints in hierarchical
models. However, the usual Bayesian strategy of placing a prior on the
constrained space and conducting posterior computation with Markov chain Monte
Carlo algorithms is often intractable. An alternative is to conduct inference
for a less constrained posterior and project samples to the constrained space
through a minimal distance mapping. We formalize and provide a unifying
framework for such posterior projections. For theoretical tractability, we
initially focus on constrained parameter spaces corresponding to closed and
convex subsets of the original space. We then consider non-convex Stiefel
manifolds. We provide a general formulation of the projected posterior and show
that it can be viewed as an update of a data-dependent prior with the
likelihood for particular classes of priors and likelihood functions. We also
show that asymptotic properties of the unconstrained posterior are transferred
to the projected posterior. Posterior projections are illustrated through
multiple examples, both in simulation studies and real data applications.
| stat |
The Soft Multivariate Truncated Normal Distribution with Applications to
Bayesian Constrained Estimation | We propose a new distribution, called the soft tMVN distribution, which
provides a smooth approximation to the truncated multivariate normal (tMVN)
distribution with linear constraints. An efficient blocked Gibbs sampler is
developed to sample from the soft tMVN distribution in high dimensions. We
provide theoretical support to the approximation capability of the soft tMVN
and provide further empirical evidence thereof. The soft tMVN distribution can
be used to approximate simulations from a multivariate truncated normal
distribution with linear constraints, or itself as a prior in shape-constrained
problems.
| stat |
MM for Penalized Estimation | Penalized estimation can conduct variable selection and parameter estimation
simultaneously. The general framework is to minimize a loss function subject to
a penalty designed to generate sparse variable selection. The
majorization-minimization (MM) algorithm is a computational scheme for
stability and simplicity, and the MM algorithm has been widely applied in
penalized estimation. Much of the previous work have focused on convex loss
functions such as generalized linear models. When data are contaminated with
outliers, robust loss functions can generate more reliable estimates. Recent
literature has witnessed a growing impact of nonconvex loss-based methods,
which can generate robust estimation for data contaminated with outliers. This
article investigates MM algorithm for penalized estimation, provide innovative
optimality conditions and establish convergence theory with both convex and
nonconvex loss functions. With respect to applications, we focus on several
nonconvex loss functions, which were formerly studied in machine learning for
regression and classification problems. Performance of the proposed algorithms
are evaluated on simulated and real data including healthcare costs and cancer
clinical status. Efficient implementations of the algorithms are available in
the R package mpath in CRAN.
| stat |
Nonparametric Analysis of Delayed Treatment Effects using
Single-Crossing Constraints | Clinical trials involving novel immuno-oncology (IO) therapies frequently
exhibit survival profiles which violate the proportional hazards assumption due
to a delay in treatment effect, and in such settings, the survival curves in
the two treatment arms may have a crossing before the two curves eventually
separate. To flexibly model such scenarios, we describe a nonparametric
approach for estimating the treatment arm-specific survival functions which
constrains these two survival functions to cross at most once without making
any additional assumptions about how the survival curves are related. A main
advantage of our approach is that it provides an estimate of a crossing time if
such a crossing exists, and moreover, our method generates interpretable
measures of treatment benefit including crossing-conditional survival
probabilities and crossing-conditional estimates of restricted residual mean
life. We demonstrate the use and effectiveness of our approach with a large
simulation study and an analysis of reconstructed outcomes from a recent
combination-therapy trial.
| stat |
Few-shot Learning for Time-series Forecasting | Time-series forecasting is important for many applications. Forecasting
models are usually trained using time-series data in a specific target task.
However, sufficient data in the target task might be unavailable, which leads
to performance degradation. In this paper, we propose a few-shot learning
method that forecasts a future value of a time-series in a target task given a
few time-series in the target task. Our model is trained using time-series data
in multiple training tasks that are different from target tasks. Our model uses
a few time-series to build a forecasting function based on a recurrent neural
network with an attention mechanism. With the attention mechanism, we can
retrieve useful patterns in a small number of time-series for the current
situation. Our model is trained by minimizing an expected test error of
forecasting next timestep values. We demonstrate the effectiveness of the
proposed method using 90 time-series datasets.
| stat |
Adversarial Robust Low Rank Matrix Estimation: Compressed Sensing and
Matrix Completion | We consider robust low rank matrix estimation when outputs are contaminated
by adversary. Our method covers matrix compressed sensing (including lasso as a
partial problem) and matrix completion. We attain fast convergence rates by
using convex estimators.
| stat |
Stochastic Gradient Trees | We present an algorithm for learning decision trees using stochastic gradient
information as the source of supervision. In contrast to previous approaches to
gradient-based tree learning, our method operates in the incremental learning
setting rather than the batch learning setting, and does not make use of soft
splits or require the construction of a new tree for every update. We
demonstrate how one can apply these decision trees to different problems by
changing only the loss function, using classification, regression, and
multi-instance learning as example applications. In the experimental
evaluation, our method performs similarly to standard incremental
classification trees, outperforms state of the art incremental regression
trees, and achieves comparable performance with batch multi-instance learning
methods.
| stat |
Neural Approximate Sufficient Statistics for Implicit Models | We consider the fundamental problem of how to automatically construct summary
statistics for implicit generative models where the evaluation of the
likelihood function is intractable, but sampling data from the model is
possible. The idea is to frame the task of constructing sufficient statistics
as learning mutual information maximizing representations of the data with the
help of deep neural networks. The infomax learning procedure does not need to
estimate any density or density ratio. We apply our approach to both
traditional approximate Bayesian computation and recent neural likelihood
methods, boosting their performance on a range of tasks.
| stat |
Simultaneous and Temporal Autoregressive Network Models | While logistic regression models are easily accessible to researchers, when
applied to network data there are unrealistic assumptions made about the
dependence structure of the data. For temporal networks measured in discrete
time, recent work has made good advances \citep{almquist2014logistic}, but
there is still the assumption that the dyads are conditionally independent
given the edge histories. This assumption can be quite strong and is sometimes
difficult to justify. If time steps are rather large, one would typically
expect not only the existence of temporal dependencies among the dyads across
observed time points but also the existence of simultaneous dependencies
affecting how the dyads of the network co-evolve. We propose a general
observation driven model for dynamic networks which overcomes this problem by
modeling both the mean and the covariance structures as functions of the edge
histories using a flexible autoregressive approach. This approach can be shown
to fit into a generalized linear mixed model framework. We propose a
visualization method which provides evidence concerning the existence of
simultaneous dependence. We describe a simulation study to determine the
method's performance in the presence and absence of simultaneous dependence,
and we analyze both a proximity network from conference attendees and a world
trade network. We also use this last data set to illustrate how simultaneous
dependencies become more prominent as the time intervals become coarser.
| stat |
Consistency of Anchor-based Spectral Clustering | Anchor-based techniques reduce the computational complexity of spectral
clustering algorithms. Although empirical tests have shown promising results,
there is currently a lack of theoretical support for the anchoring approach. We
define a specific anchor-based algorithm and show that it is amenable to
rigorous analysis, as well as being effective in practice. We establish the
theoretical consistency of the method in an asymptotic setting where data is
sampled from an underlying continuous probability distribution. In particular,
we provide sharp asymptotic conditions for the algorithm parameters which
ensure that the anchor-based method can recover with high probability disjoint
clusters that are mutually separated by a positive distance. We illustrate the
performance of the algorithm on synthetic data and explain how the theoretical
convergence analysis can be used to inform the practical choice of parameter
scalings. We also test the accuracy and efficiency of the algorithm on two
large scale real data sets. We find that the algorithm offers clear advantages
over standard spectral clustering. We also find that it is competitive with the
state-of-the-art LSC method of Chen and Cai (Twenty-Fifth AAAI Conference on
Artificial Intelligence, 2011), while having the added benefit of a consistency
guarantee.
| stat |
Semiparametric Sensitivity Analysis: Unmeasured Confounding In
Observational Studies | Establishing cause-effect relationships from observational data often relies
on untestable assumptions. It is crucial to know whether, and to what extent,
the conclusions drawn from non-experimental studies are robust to potential
unmeasured confounding. In this paper, we focus on the average causal effect
(ACE) as our target of inference. We build on the work of Franks et al.
(2019)and Robins (2000) by specifying non-identified sensitivity parameters
that govern a contrast between the conditional (on measured covariates)
distributions of the outcome under treatment (control) between treated and
untreated individuals. We use semiparametric theory to derive the
non-parametric efficient influence function of the ACE, for fixed sensitivity
parameters. We use this influence function to construct a one-step
bias-corrected estimator of the ACE. Our estimator depends on semiparametric
models for the distribution of the observed data; importantly, these models do
not impose any restrictions on the values of sensitivity analysis parameters.
We establish sufficient conditions ensuring that our estimator has root-n
asymptotics. We use our methodology to evaluate the causal effect of smoking
during pregnancy on birth weight. We also evaluate the performance of
estimation procedure in a simulation study.
| stat |
Change-point detection in dynamic networks via graphon estimation | We propose a general approach for change-point detection in dynamic networks.
The proposed method is model-free and covers a wide range of dynamic networks.
The key idea behind our approach is to effectively utilize the network
structure in designing change-point detection algorithms. This is done via an
initial step of graphon estimation, where we propose a modified neighborhood
smoothing~(MNBS) algorithm for estimating the link probability matrices of a
dynamic network. Based on the initial graphon estimation, we then develop a
screening and thresholding algorithm for multiple change-point detection in
dynamic networks. The convergence rate and consistency for the change-point
detection procedure are derived as well as those for MNBS. When the number of
nodes is large~(e.g., exceeds the number of temporal points), our approach
yields a faster convergence rate in detecting change-points comparing with an
algorithm that simply employs averaged information of the dynamic network
across time. Numerical experiments demonstrate robust performance of the
proposed algorithm for change-point detection under various types of dynamic
networks, and superior performance over existing methods is observed. A real
data example is provided to illustrate the effectiveness and practical impact
of the procedure.
| stat |
Provable More Data Hurt in High Dimensional Least Squares Estimator | This paper investigates the finite-sample prediction risk of the
high-dimensional least squares estimator. We derive the central limit theorem
for the prediction risk when both the sample size and the number of features
tend to infinity. Furthermore, the finite-sample distribution and the
confidence interval of the prediction risk are provided. Our theoretical
results demonstrate the sample-wise nonmonotonicity of the prediction risk and
confirm "more data hurt" phenomenon.
| stat |
Multivariate time series models for mixed data | We introduce a general approach for modeling the dynamic of multivariate time
series when the data are of mixed type (binary/count/continuous). Our method is
quite flexible and conditionally on past values, each coordinate at time $t$
can have a distribution compatible with a standard univariate time series model
such as GARCH, ARMA, INGARCH or logistic models whereas past values of the
other coordinates play the role of exogenous covariates in the dynamic. The
simultaneous dependence in the multivariate time series can be modeled with a
copula. Additional exogenous covariates are also allowed in the dynamic. We
first study usual stability properties of these models and then show that
autoregressive parameters can be consistently estimated equation-by-equation
using a pseudo-maximum likelihood method, leading to a fast implementation even
when the number of time series is large. Moreover, we prove consistency results
when a parametric copula model is fitted to the time series and in the case of
Gaussian copulas, we show that the likelihood estimator of the correlation
matrix is strongly consistent. We carefully check all our assumptions for two
prototypical examples: a GARCH/INGARCH model and logistic/log-linear INGARCH
model. Our results are illustrated with numerical experiments as well as two
real data sets.
| stat |
Machine-Learning Tests for Effects on Multiple Outcomes | In this paper we present tools for applied researchers that re-purpose
off-the-shelf methods from the computer-science field of machine learning to
create a "discovery engine" for data from randomized controlled trials (RCTs).
The applied problem we seek to solve is that economists invest vast resources
into carrying out RCTs, including the collection of a rich set of candidate
outcome measures. But given concerns about inference in the presence of
multiple testing, economists usually wind up exploring just a small subset of
the hypotheses that the available data could be used to test. This prevents us
from extracting as much information as possible from each RCT, which in turn
impairs our ability to develop new theories or strengthen the design of policy
interventions. Our proposed solution combines the basic intuition of reverse
regression, where the dependent variable of interest now becomes treatment
assignment itself, with methods from machine learning that use the data
themselves to flexibly identify whether there is any function of the outcomes
that predicts (or has signal about) treatment group status. This leads to
correctly-sized tests with appropriate $p$-values, which also have the
important virtue of being easy to implement in practice. One open challenge
that remains with our work is how to meaningfully interpret the signal that
these methods find.
| stat |
The Raise Regression: Justification, properties and application | Multicollinearity produces an inflation in the variance of the Ordinary Least
Squares estimators due to the correlation between two or more independent
variables (including the constant term). A widely applied solution is to
estimate with penalized estimators (such as the ridge estimator, the Liu
estimator, etc.) which exchange the mean square error by the bias. Although the
variance diminishes with these procedures, all seems to indicate that the
inference is lost and also the goodness of fit. Alternatively, the raise
regression (\cite{Garcia2011} and \cite{Salmeron2017}) allows the mitigation of
the problems generated by multicollinearity but without losing the inference
and keeping the coefficient of determination. This paper completely formalizes
the raise estimator summarizing all the previous contributions: its mean square
error, the variance inflation factor, the condition number, the adequate
selection of the variable to be raised, the successive raising and the relation
between the raise and the ridge estimator. As a novelty, it is also presented
the estimation method, the relation between the raise and the residualization,
it is analyzed the norm of the estimator and the behaviour of the individual
and joint significance test and the behaviour of the mean square error and the
coefficient of variation. The usefulness of the raise regression as alternative
to mitigate the multicollinearity is illustrated with two empirical
applications.
| stat |
The role of swabs in modeling the COVID-19 outbreak in the most affected
regions of Italy | The daily fluctuations in the released number of Covid-19 cases played a big
role both at the beginning and in the most critical weeks of the outbreak, when
local authorities in Italy had to decide whether to impose a lockdown and at
which level. Public opinion was focused on this information as well, to
understand how quickly the epidemic was spreading. When an increase/decrease
was communicated, especially a large one, it was not easy to understand if it
was due to a change in the epidemic evolution or if it was a fluctuation due to
other reasons, such as an increase in the number of swabs or a delay in the
swab processing. In this work, we propose a nonlinear asymmetric diffusion
model, which includes information on the daily number of swabs, to describe
daily fluctuations in the number of confirmed cases in addition to the main
trend of the outbreak evolution. The class of diffusion models has been
selected to develop our proposal, as it also allows estimation of the total
number of confirmed cases at the end of the outbreak. The proposed model is
compared to six existing models, among which the logistic and the SIRD models
are used as benchmarks, in the five most affected Italian regions.
| stat |
Reluctant Interaction Modeling | Including pairwise interactions between the predictors of a regression model
can produce better predicting models. However, to fit such interaction models
on typical data sets in biology and other fields can often require solving
enormous variable selection problems with billions of interactions. The scale
of such problems demands methods that are computationally cheap (both in time
and memory) yet still have sound statistical properties. Motivated by these
large-scale problem sizes, we adopt a very simple guiding principle: One should
prefer a main effect over an interaction if all else is equal. This
"reluctance" to interactions, while reminiscent of the hierarchy principle for
interactions, is much less restrictive. We design a computationally efficient
method built upon this principle and provide theoretical results indicating
favorable statistical properties. Empirical results show dramatic computational
improvement without sacrificing statistical properties. For example, the
proposed method can solve a problem with 10 billion interactions with 5-fold
cross-validation in under 7 hours on a single CPU.
| stat |
A Bayesian Nonparametric model for textural pattern heterogeneity | Cancer radiomics is an emerging discipline promising to elucidate lesion
phenotypes and tumor heterogeneity through patterns of enhancement, texture,
morphology, and shape. The prevailing technique for image texture analysis
relies on the construction and synthesis of Gray-Level Co-occurrence Matrices
(GLCM). Practice currently reduces the structured count data of a GLCM to
reductive and redundant summary statistics for which analysis requires variable
selection and multiple comparisons for each application, thus limiting
reproducibility. In this article, we develop a Bayesian multivariate
probabilistic framework for the analysis and unsupervised clustering of a
sample of GLCM objects. By appropriately accounting for skewness and
zero-inflation of the observed counts and simultaneously adjusting for existing
spatial autocorrelation at nearby cells, the methodology facilitates estimation
of texture pattern distributions within the GLCM lattice itself. The techniques
are applied to cluster images of adrenal lesions obtained from CT scans with
and without administration of contrast. We further assess whether the resultant
subtypes are clinically oriented by investigating their correspondence with
pathological diagnoses. Additionally, we compare performance to a class of
machine-learning approaches currently used in cancer radiomics with simulation
studies.
| stat |
A general modelling framework for open wildlife populations based on the
Polya Tree prior | Wildlife monitoring for open populations can be performed using a number of
different survey methods. Each survey method gives rise to a type of data and,
in the last five decades, a large number of associated statistical models have
been developed for analysing these data. Although these models have been
parameterised and fitted using different approaches, they have all been
designed to model the pattern with which individuals enter and exit the
population and to estimate the population size. However, existing approaches
rely on a predefined model structure and complexity, either by assuming that
parameters are specific to sampling occasions, or by employing parametric
curves. Instead, we propose a novel Bayesian nonparametric framework for
modelling entry and exit patterns based on the Polya Tree (PT) prior for
densities. Our Bayesian non-parametric approach avoids overfitting when
inferring entry and exit patterns while simultaneously allowing more
flexibility than is possible using parametric curves. We apply our new
framework to capture-recapture, count and ring-recovery data and we introduce
the replicated PT prior for defining classes of models for these data.
Additionally, we define the Hierarchical Logistic PT prior for jointly
modelling related data and we consider the Optional PT prior for modelling long
time series of data. We demonstrate our new approach using five different case
studies on birds, amphibians and insects.
| stat |
bamlss: A Lego Toolbox for Flexible Bayesian Regression (and Beyond) | Over the last decades, the challenges in applied regression and in predictive
modeling have been changing considerably: (1) More flexible model
specifications are needed as big(ger) data become available, facilitated by
more powerful computing infrastructure. (2) Full probabilistic modeling rather
than predicting just means or expectations is crucial in many applications. (3)
Interest in Bayesian inference has been increasing both as an appealing
framework for regularizing or penalizing model estimation as well as a natural
alternative to classical frequentist inference. However, while there has been a
lot of research in all three areas, also leading to associated software
packages, a modular software implementation that allows to easily combine all
three aspects has not yet been available. For filling this gap, the R package
bamlss is introduced for Bayesian additive models for location, scale, and
shape (and beyond). At the core of the package are algorithms for
highly-efficient Bayesian estimation and inference that can be applied to
generalized additive models (GAMs) or generalized additive models for location,
scale, and shape (GAMLSS), also known as distributional regression. However,
its building blocks are designed as "Lego bricks" encompassing various
distributions (exponential family, Cox, joint models, ...), regression terms
(linear, splines, random effects, tensor products, spatial fields, ...), and
estimators (MCMC, backfitting, gradient boosting, lasso, ...). It is
demonstrated how these can be easily recombined to make classical models more
flexible or create new custom models for specific modeling challenges.
| stat |
Semiparametric Methods for Exposure Misclassification in Propensity
Score-Based Time-to-Event Data Analysis | In epidemiology, identifying the effect of exposure variables in relation to
a time-to-event outcome is a classical research area of practical importance.
Incorporating propensity score in the Cox regression model, as a measure to
control for confounding, has certain advantages when outcome is rare. However,
in situations involving exposure measured with moderate to substantial error,
identifying the exposure effect using propensity score in Cox models remains a
challenging yet unresolved problem. In this paper, we propose an estimating
equation method to correct for the exposure misclassification-caused bias in
the estimation of exposure-outcome associations. We also discuss the asymptotic
properties and derive the asymptotic variances of the proposed estimators. We
conduct a simulation study to evaluate the performance of the proposed
estimators in various settings. As an illustration, we apply our method to
correct for the misclassification-caused bias in estimating the association of
PM2.5 level with lung cancer mortality using a nationwide prospective cohort,
the Nurses' Health Study (NHS). The proposed methodology can be applied using
our user-friendly R function published online.
| stat |
Revealing interpretable object representations from human behavior | To study how mental object representations are related to behavior, we
estimated sparse, non-negative representations of objects using human
behavioral judgments on images representative of 1,854 object categories. These
representations predicted a latent similarity structure between objects, which
captured most of the explainable variance in human behavioral judgments.
Individual dimensions in the low-dimensional embedding were found to be highly
reproducible and interpretable as conveying degrees of taxonomic membership,
functionality, and perceptual attributes. We further demonstrated the
predictive power of the embeddings for explaining other forms of human
behavior, including categorization, typicality judgments, and feature ratings,
suggesting that the dimensions reflect human conceptual representations of
objects beyond the specific task.
| stat |
Sparse dimension reduction based on energy and ball statistics | As its name suggests, sufficient dimension reduction (SDR) targets to
estimate a subspace from data that contains all information sufficient to
explain a dependent variable. Ample approaches exist to SDR, some of the most
recent of which rely on minimal to no model assumptions. These are defined
according to an optimization criterion that maximizes a nonparametric measure
of association. The original estimators are nonsparse, which means that all
variables contribute to the model. However, in many practical applications, an
SDR technique may be called for that is sparse and as such, intrinsically
performs sufficient variable selection (SVS). This paper examines how such a
sparse SDR estimator can be constructed. Three variants are investigated,
depending on different measures of association: distance covariance, martingale
difference divergence and ball covariance. A simulation study shows that each
of these estimators can achieve correct variable selection in highly nonlinear
contexts, yet are sensitive to outliers and computationally intensive. The
study sheds light on the subtle differences between the methods. Two examples
illustrate how these new estimators can be applied in practice, with a slight
preference for the option based on martingale difference divergence in the
bioinformatics example.
| stat |
The Extended Dawid-Skene Model: Fusing Information from Multiple Data
Schemas | While label fusion from multiple noisy annotations is a well understood
concept in data wrangling (tackled for example by the Dawid-Skene (DS) model),
we consider the extended problem of carrying out learning when the labels
themselves are not consistently annotated with the same schema. We show that
even if annotators use disparate, albeit related, label-sets, we can still draw
inferences for the underlying full label-set. We propose the Inter-Schema
AdapteR (ISAR) to translate the fully-specified label-set to the one used by
each annotator, enabling learning under such heterogeneous schemas, without the
need to re-annotate the data. We apply our method to a mouse behavioural
dataset, achieving significant gains (compared with DS) in out-of-sample
log-likelihood (-3.40 to -2.39) and F1-score (0.785 to 0.864).
| stat |
D-optimal Design for Network A/B Testing | A/B testing refers to the statistical procedure of conducting an experiment
to compare two treatments, A and B, applied to different testing subjects. It
is widely used by technology companies such as Facebook, LinkedIn, and Netflix,
to compare different algorithms, web-designs, and other online products and
services. The subjects participating these online A/B testing experiments are
users who are connected in different scales of social networks. Two connected
subjects are similar in terms of their social behaviors, education and
financial background, and other demographic aspects. Hence, it is only natural
to assume that their reactions to the online products and services are related
to their network adjacency. In this paper, we propose to use the conditional
auto-regressive model to present the network structure and include the network
effects in the estimation and inference of the treatment effect. A D-optimal
design criterion is developed based on the proposed model. Mixed integer
programming formulations are developed to obtain the D-optimal designs. The
effectiveness of the proposed method is shown through numerical results with
synthetic networks and real social networks.
| stat |
Spatial Matrix Completion for Spatially-Misaligned and High-Dimensional
Air Pollution Data | In health-pollution cohort studies, accurate predictions of pollutant
concentrations at new locations are needed, since the locations of fixed
monitoring sites and study participants are often spatially misaligned. For
multi-pollution data, principal component analysis (PCA) is often incorporated
to obtain low-rank (LR) structure of the data prior to spatial prediction.
Recently developed predictive PCA modifies the traditional algorithm to improve
the overall predictive performance by leveraging both LR and spatial structures
within the data. However, predictive PCA requires complete data or an initial
imputation step. Nonparametric imputation techniques without accounting for
spatial information may distort the underlying structure of the data, and thus
further reduce the predictive performance. We propose a convex optimization
problem inspired by the LR matrix completion framework and develop a proximal
algorithm to solve it. Missing data are imputed and handled concurrently within
the algorithm, which eliminates the necessity of a separate imputation step. We
show that our algorithm has low computational burden and leads to reliable
predictive performance as the severity of missing data increases.
| stat |
Deep Variable-Block Chain with Adaptive Variable Selection | The architectures of deep neural networks (DNN) rely heavily on the
underlying grid structure of variables, for instance, the lattice of pixels in
an image. For general high dimensional data with variables not associated with
a grid, the multi-layer perceptron and deep brief network are often used.
However, it is frequently observed that those networks do not perform
competitively and they are not helpful for identifying important variables. In
this paper, we propose a framework that imposes on blocks of variables a chain
structure obtained by step-wise greedy search so that the DNN architecture can
leverage the constructed grid. We call this new neural network Deep
Variable-Block Chain (DVC). Because the variable blocks are used for
classification in a sequential manner, we further develop the capacity of
selecting variables adaptively according to a number of regions trained by a
decision tree. Our experiments show that DVC outperforms other generic DNNs and
other strong classifiers. Moreover, DVC can achieve high accuracy at much
reduced dimensionality and sometimes reveals drastically different sets of
relevant variables for different regions.
| stat |
Statistical Integration of Heterogeneous Data with PO2PLS | The availability of multi-omics data has revolutionized the life sciences by
creating avenues for integrated system-level approaches. Data integration links
the information across datasets to better understand the underlying biological
processes. However, high-dimensionality, correlations and heterogeneity pose
statistical and computational challenges. We propose a general framework,
probabilistic two-way partial least squares (PO2PLS), which addresses these
challenges. PO2PLS models the relationship between two datasets using joint and
data-specific latent variables. For maximum likelihood estimation of the
parameters, we implement a fast EM algorithm and show that the estimator is
asymptotically normally distributed. A global test for testing the relationship
between two datasets is proposed, and its asymptotic distribution is derived.
Notably, several existing omics integration methods are special cases of
PO2PLS. Via extensive simulations, we show that PO2PLS performs better than
alternatives in feature selection and prediction performance. In addition, the
asymptotic distribution appears to hold when the sample size is sufficiently
large. We illustrate PO2PLS with two examples from commonly used study designs:
a large population cohort and a small case-control study. Besides recovering
known relationships, PO2PLS also identified novel findings. The methods are
implemented in our R-package PO2PLS. Supplementary materials for this article
are available online.
| stat |
Gaussian orthogonal latent factor processes for large incomplete
matrices of correlated data | We introduce the Gaussian orthogonal latent factor processes for modeling and
predicting large correlated data. To handle the computational challenge, we
first decompose the likelihood function of the Gaussian random field with
multi-dimensional input domain into a product of densities at the orthogonal
components with lower dimensional inputs. The continuous-time Kalman filter is
implemented to efficiently compute the likelihood function without making
approximation. We also show that the posterior distribution of the factor
processes are independent, as a consequence of prior independence of factor
processes and orthogonal factor loading matrix. For studies with a large sample
size, we propose a flexible way to model the mean in the model and derive the
closed-form marginal posterior distribution. Both simulated and real data
applications confirm the outstanding performance of this method.
| stat |
Multidimensional molecular changes-environment interaction analysis for
disease outcomes | For the outcomes and phenotypes of complex diseases, multiple types of
molecular (genetic, genomic, epigenetic, etc.) changes, environmental risk
factors, and their interactions have been found to have important
contributions. In each of the existing studies, only the interactions between
one type of molecular changes and environmental risk factors have been
analyzed. In recent biomedical studies, multidimensional profiling, under which
data on multiple types of molecular changes is collected on the same subjects,
is becoming popular. A myriad of recent studies have shown that collectively
analyzing multiple types of molecular changes is not only biologically sensible
but also leads to improved estimation and prediction. In this study, we conduct
M-E interaction analysis, with M standing for multidimensional molecular
changes and E standing for environmental risk factors, which can accommodate
multiple types of molecular measurements and sufficiently account for their
overlapping information (attributable to regulations) as well as independent
information. The proposed approach is based on the penalization technique, has
a solid statistical ground, and can be effectively realized. Extensive
simulation shows that it outperforms multiple closely relevant alternatives. In
the analysis of TCGA (The Cancer Genome Atlas) data on lung adenocarcinoma and
cutaneous melanoma, sensible findings with superior stability and prediction
are made.
| stat |
Leveraging Administrative Data for Bias Audits: Assessing Disparate
Coverage with Mobility Data for COVID-19 Policy | Anonymized smartphone-based mobility data has been widely adopted in devising
and evaluating COVID-19 response strategies such as the targeting of public
health resources. Yet little attention has been paid to measurement validity
and demographic bias, due in part to the lack of documentation about which
users are represented as well as the challenge of obtaining ground truth data
on unique visits and demographics. We illustrate how linking large-scale
administrative data can enable auditing mobility data for bias in the absence
of demographic information and ground truth labels. More precisely, we show
that linking voter roll data -- containing individual-level voter turnout for
specific voting locations along with race and age -- can facilitate the
construction of rigorous bias and reliability tests. These tests illuminate a
sampling bias that is particularly noteworthy in the pandemic context: older
and non-white voters are less likely to be captured by mobility data. We show
that allocating public health resources based on such mobility data could
disproportionately harm high-risk elderly and minority groups.
| stat |
Memorizing without overfitting: Bias, variance, and interpolation in
over-parameterized models | The bias-variance trade-off is a central concept in supervised learning. In
classical statistics, increasing the complexity of a model (e.g., number of
parameters) reduces bias but also increases variance. Until recently, it was
commonly believed that optimal performance is achieved at intermediate model
complexities which strike a balance between bias and variance. Modern Deep
Learning methods flout this dogma, achieving state-of-the-art performance using
"over-parameterized models" where the number of fit parameters is large enough
to perfectly fit the training data. As a result, understanding bias and
variance in over-parameterized models has emerged as a fundamental problem in
machine learning. Here, we use methods from statistical physics to derive
analytic expressions for bias and variance in three minimal models for
over-parameterization (linear regression and two-layer neural networks with
linear and nonlinear activation functions), allowing us to disentangle
properties stemming from the model architecture and random sampling of data.
All three models exhibit a phase transition to an interpolation regime where
the training error is zero, with linear neural-networks possessing an
additional phase transition between regimes with zero and nonzero bias. The
test error diverges at the interpolation transition for all three models.
However, beyond the transition, it decreases again for the neural network
models due to a decrease in both bias and variance with model complexity. We
also show that over-parameterized models can overfit even in the absence of
noise. We synthesize these results to construct a holistic understanding of
generalization error and the bias-variance trade-off in over-parameterized
models.
| stat |
MIP-BOOST: Efficient and Effective $L_0$ Feature Selection for Linear
Regression | Recent advances in mathematical programming have made Mixed Integer
Optimization a competitive alternative to popular regularization methods for
selecting features in regression problems. The approach exhibits unquestionable
foundational appeal and versatility, but also poses important challenges. Here
we propose MIP-BOOST, a revision of standard Mixed Integer Programming feature
selection that reduces the computational burden of tuning the critical sparsity
bound parameter and improves performance in the presence of feature
collinearity and of signals that vary in nature and strength. The final outcome
is a more efficient and effective $L_0$ Feature Selection method for
applications of realistic size and complexity, grounded on rigorous
cross-validation tuning and exact optimization of the associated Mixed Integer
Program. Computational viability and improved performance in realistic
scenarios is achieved through three independent but synergistic proposals.
| stat |
Bayesian Modeling for Exposure Response Curve via Gaussian Processes:
Causal Effects of Exposure to Air Pollution on Health Outcomes | Motivated by environmental health research on air pollution, we address the
challenge of uncertainty quantification for causal exposure-response functions.
A causal exposure-response function (CERF) describes the relationship between a
continuously varying exposure (or treatment) and its causal effect on a
outcome. We propose a new Bayesian approach that relies on a Gaussian process
(GP) model to estimate the CERF. We parametrize the covariance (kernel)
function of the GP to mimic matching via a Generalized Propensity Score (GPS).
The tuning parameters of the matching function are chosen to optimize covariate
balance. Our approach achieves automatic uncertainty evaluation of the CERF
with high computational efficiency, enables change point detection through
inference on higher order derivatives of the CERF, and yields the desired
separation of design and analysis phases for causal estimation. We provide
theoretical results showing the correspondence between our Bayesian GP
framework and traditional approaches in causal inference for estimating causal
effects of a continuous exposure. We apply the methods to 520,711 ZIP code
level observations to estimate the causal effect of long-term exposures to PM
2.5 on all cause mortality among Medicare enrollees in the United States.
| stat |
Deep Distribution Regression | Due to their flexibility and predictive performance, machine-learning based
regression methods have become an important tool for predictive modeling and
forecasting. However, most methods focus on estimating the conditional mean or
specific quantiles of the target quantity and do not provide the full
conditional distribution, which contains uncertainty information that might be
crucial for decision making. In this article, we provide a general solution by
transforming a conditional distribution estimation problem into a constrained
multi-class classification problem, in which tools such as deep neural
networks. We propose a novel joint binary cross-entropy loss function to
accomplish this goal. We demonstrate its performance in various simulation
studies comparing to state-of-the-art competing methods. Additionally, our
method shows improved accuracy in a probabilistic solar energy forecasting
problem.
| stat |
Analyzing initial stage of COVID-19 transmission through Bayesian
time-varying model | Recent outbreak of the novel coronavirus COVID-19 has affected all of our
lives in one way or the other. While medical researchers are working hard to
find a cure and doctors/nurses to attend the affected individuals, measures
such as `lockdown', `stay-at-home', `social distancing' are being implemented
in different parts of the world to curb its further spread. To model the
non-stationary spread, we propose a novel time-varying semiparametric AR$(p)$
model for the count valued time-series of newly affected cases, collected every
day and also extend it to propose a novel time-varying INGARCH model. Our
proposed structures of the models are amenable to Hamiltonian Monte Carlo (HMC)
sampling for efficient computation. We substantiate our methods by simulations
that show superiority compared to some of the close existing methods. Finally
we analyze the daily time series data of newly confirmed cases to study its
spread through different government interventions.
| stat |
Sampling with censored data: a practical guide | In this review, we present a simple guide for researchers to obtain
pseudo-random samples with censored data. We focus our attention on the most
common types of censored data, such as type I, type II, and random censoring.
We discussed the necessary steps to sample pseudo-random values from long-term
survival models where an additional cure fraction is informed. For illustrative
purposes, these techniques are applied in the Weibull distribution. The
algorithms and codes in R are presented, enabling the reproducibility of our
study.
| stat |
Pseudo Bayesian Estimation of One-way ANOVA Model in Complex Surveys | We devise survey-weighted pseudo posterior distribution estimators under
2-stage informative sampling of both primary clusters and secondary nested
units for a one-way ANOVA population generating model as a simple canonical
case where population model random effects are defined to be coincident with
the primary clusters. We consider estimation on an observed informative sample
under both an augmented pseudo likelihood that co-samples random effects, as
well as an integrated likelihood that marginalizes out the random effects from
the survey-weighted augmented pseudo likelihood. This paper includes a
theoretical exposition that enumerates easily verified conditions for which
estimation under the augmented pseudo posterior is guaranteed to be consistent
at the true generating parameters. We reveal in simulation that both approaches
produce asymptotically unbiased estimation of the generating hyperparameters
for the random effects when a key condition on the sum of within cluster
weighted residuals is met. We present a comparison with frequentist EM and a
methods that requires pairwise sampling weights.
| stat |
Radial Bayesian Neural Networks: Beyond Discrete Support In Large-Scale
Bayesian Deep Learning | We propose Radial Bayesian Neural Networks (BNNs): a variational approximate
posterior for BNNs which scales well to large models while maintaining a
distribution over weight-space with full support. Other scalable Bayesian deep
learning methods, like MC dropout or deep ensembles, have discrete support-they
assign zero probability to almost all of the weight-space. Unlike these
discrete support methods, Radial BNNs' full support makes them suitable for use
as a prior for sequential inference. In addition, they solve the conceptual
challenges with the a priori implausibility of weight distributions with
discrete support. The Radial BNN is motivated by avoiding a sampling problem in
'mean-field' variational inference (MFVI) caused by the so-called 'soap-bubble'
pathology of multivariate Gaussians. We show that, unlike MFVI, Radial BNNs are
robust to hyperparameters and can be efficiently applied to a challenging
real-world medical application without needing ad-hoc tweaks and intensive
tuning. In fact, in this setting Radial BNNs out-perform discrete-support
methods like MC dropout. Lastly, by using Radial BNNs as a theoretically
principled, robust alternative to MFVI we make significant strides in a
Bayesian continual learning evaluation.
| stat |
Burglary in London: Insights from Statistical Heterogeneous Spatial
Point Processes | To obtain operational insights regarding the crime of burglary in London we
consider the estimation of effects of covariates on the intensity of spatial
point patterns. By taking into account localised properties of criminal
behaviour, we propose a spatial extension to model-based clustering methods
from the mixture modelling literature. The proposed Bayesian model is a finite
mixture of Poisson generalised linear models such that each location is
probabilistically assigned to one of the clusters. Each cluster is
characterised by the regression coefficients which we subsequently use to
interpret the localised effects of the covariates. Using a blocking structure
of the study region, our approach allows specifying spatial dependence between
nearby locations. We estimate the proposed model using Markov Chain Monte Carlo
methods and provide a Python implementation.
| stat |
Cautious Reinforcement Learning via Distributional Risk in the Dual
Domain | We study the estimation of risk-sensitive policies in reinforcement learning
problems defined by a Markov Decision Process (MDPs) whose state and action
spaces are countably finite. Prior efforts are predominately afflicted by
computational challenges associated with the fact that risk-sensitive MDPs are
time-inconsistent. To ameliorate this issue, we propose a new definition of
risk, which we call caution, as a penalty function added to the dual objective
of the linear programming (LP) formulation of reinforcement learning. The
caution measures the distributional risk of a policy, which is a function of
the policy's long-term state occupancy distribution. To solve this problem in
an online model-free manner, we propose a stochastic variant of primal-dual
method that uses Kullback-Lieber (KL) divergence as its proximal term. We
establish that the number of iterations/samples required to attain
approximately optimal solutions of this scheme matches tight dependencies on
the cardinality of the state and action spaces, but differs in its dependence
on the infinity norm of the gradient of the risk measure. Experiments
demonstrate the merits of this approach for improving the reliability of reward
accumulation without additional computational burdens.
| stat |
Bayesian Inference over the Stiefel Manifold via the Givens
Representation | We introduce an approach based on the Givens representation for posterior
inference in statistical models with orthogonal matrix parameters, such as
factor models and probabilistic principal component analysis (PPCA). We show
how the Givens representation can be used to develop practical methods for
transforming densities over the Stiefel manifold into densities over subsets of
Euclidean space. We show how to deal with issues arising from the topology of
the Stiefel manifold and how to inexpensively compute the change-of-measure
terms. We introduce an auxiliary parameter approach that limits the impact of
topological issues. We provide both analysis of our methods and numerical
examples demonstrating the effectiveness of the approach. We also discuss how
our Givens representation can be used to define general classes of
distributions over the space of orthogonal matrices. We then give
demonstrations on several examples showing how the Givens approach performs in
practice in comparison with other methods.
| stat |
Covariate Adaptive Family-wise Error Rate Control for Genome-Wide
Association Studies | The family-wise error rate (FWER) has been widely used in genome-wide
association studies. With the increasing availability of functional genomics
data, it is possible to increase the detection power by leveraging these
genomic functional annotations. Previous efforts to accommodate covariates in
multiple testing focus on the false discovery rate control while
covariate-adaptive FWER-controlling procedures remain under-developed. Here we
propose a novel covariate-adaptive FWER-controlling procedure that incorporates
external covariates which are potentially informative of either the statistical
power or the prior null probability. An efficient algorithm is developed to
implement the proposed method. We prove its asymptotic validity and obtain the
rate of convergence through a perturbation-type argument. Our numerical studies
show that the new procedure is more powerful than competing methods and
maintains robustness across different settings. We apply the proposed approach
to the UK Biobank data and analyze 27 traits with 9 million single-nucleotide
polymorphisms tested for associations. Seventy-five genomic annotations are
used as covariates. Our approach detects more genome-wide significant loci than
other methods in 21 out of the 27 traits.
| stat |
Depth for curve data and applications | John W. Tukey (1975) defined statistical data depth as a function that
determines centrality of an arbitrary point with respect to a data cloud or to
a probability measure. During the last decades, this seminal idea of data depth
evolved into a powerful tool proving to be useful in various fields of science.
Recently, extending the notion of data depth to the functional setting
attracted a lot of attention among theoretical and applied statisticians. We go
further and suggest a notion of data depth suitable for data represented as
curves, or trajectories, which is independent of the parametrization. We show
that our curve depth satisfies theoretical requirements of general depth
functions that are meaningful for trajectories. We apply our methodology to
diffusion tensor brain images and also to pattern recognition of hand written
digits and letters. Supplementary Materials are available online.
| stat |
Differentially Private SGD with Non-Smooth Loss | In this paper, we are concerned with differentially private SGD algorithms in
the setting of stochastic convex optimization (SCO). Most of existing work
requires the loss to be Lipschitz continuous and strongly smooth, and the model
parameter to be uniformly bounded. However, these assumptions are restrictive
as many popular losses violate these conditions including the hinge loss for
SVM, the absolute loss in robust regression, and even the least square loss in
an unbounded domain. We significantly relax these restrictive assumptions and
establish privacy and generalization (utility) guarantees for private SGD
algorithms using output and gradient perturbations associated with non-smooth
convex losses. Specifically, the loss function is relaxed to have
$\alpha$-H\"{o}lder continuous gradient (referred to as $\alpha$-H\"{o}lder
smoothness) which instantiates the Lipschitz continuity ($\alpha=0$) and strong
smoothness ($\alpha=1$). We prove that noisy SGD with $\alpha$-H\"older smooth
losses using gradient perturbation can guarantee
$(\epsilon,\delta)$-differential privacy (DP) and attain optimal excess
population risk
$O\Big(\frac{\sqrt{d\log(1/\delta)}}{n\epsilon}+\frac{1}{\sqrt{n}}\Big)$, up to
logarithmic terms, with gradient complexity (i.e. the total number of
iterations) $T =O( n^{2-\alpha\over 1+\alpha}+ n).$ This shows an important
trade-off between $\alpha$-H\"older smoothness of the loss and the
computational complexity $T$ for private SGD with statistically optimal
performance. In particular, our results indicate that $\alpha$-H\"older
smoothness with $\alpha\ge {1/2}$ is sufficient to guarantee
$(\epsilon,\delta)$-DP of noisy SGD algorithms while achieving optimal excess
risk with linear gradient complexity $T = O(n).$
| stat |
Efficient Experimental Design for Regularized Linear Models | Regularized linear models, such as Lasso, have attracted great attention in
statistical learning and data science. However, there is sporadic work on
constructing efficient data collection for regularized linear models. In this
work, we propose an experimental design approach, using nearly orthogonal Latin
hypercube designs, to enhance the variable selection accuracy of the
regularized linear models. Systematic methods for constructing such designs are
presented. The effectiveness of the proposed method is illustrated with several
examples.
| stat |
On Mendelian Randomization Mixed-Scale Treatment Effect Robust
Identification (MR MiSTERI) and Estimation for Causal Inference | Standard Mendelian randomization analysis can produce biased results if the
genetic variant defining the instrumental variable (IV) is confounded and/or
has a horizontal pleiotropic effect on the outcome of interest not mediated by
the treatment. We provide novel identification conditions for the causal effect
of a treatment in presence of unmeasured confounding, by leveraging an invalid
IV for which both the IV independence and exclusion restriction assumptions may
be violated. The proposed Mendelian Randomization Mixed-Scale Treatment Effect
Robust Identification (MR MiSTERI) approach relies on (i) an assumption that
the treatment effect does not vary with the invalid IV on the additive scale;
and (ii) that the selection bias due to confounding does not vary with the
invalid IV on the odds ratio scale; and (iii) that the residual variance for
the outcome is heteroscedastic and thus varies with the invalid IV. We formally
establish that their conjunction can identify a causal effect even with an
invalid IV subject to pleiotropy. MiSTERI is shown to be particularly
advantageous in presence of pervasive heterogeneity of pleiotropic effects on
additive scale, a setting in which two recently proposed robust estimation
methods MR GxE and MR GENIUS can be severely biased. In order to incorporate
multiple, possibly correlated and weak IVs, a common challenge in MR studies,
we develop a MAny Weak Invalid Instruments (MR MaWII MiSTERI) approach for
strengthened identification and improved accuracy MaWII MiSTERI is shown to be
robust to horizontal pleiotropy, violation of IV independence assumption and
weak IV bias. Both simulation studies and real data analysis results
demonstrate the robustness of the proposed MR MiSTERI methods.
| stat |
Calibration Scoring Rules for Practical Prediction Training | In situations where forecasters are scored on the quality of their
probabilistic predictions, it is standard to use `proper' scoring rules to
perform such scoring. These rules are desirable because they give forecasters
no incentive to lie about their probabilistic beliefs. However, in the real
world context of creating a training program designed to help people improve
calibration through prediction practice, there are a variety of desirable
traits for scoring rules that go beyond properness. These potentially may have
a substantial impact on the user experience, usability of the program, or
efficiency of learning. The space of proper scoring rules is too broad, in the
sense that most proper scoring rules lack these other desirable properties. On
the other hand, the space of proper scoring rules is potentially also too
narrow, in the sense that we may sometimes choose to give up properness when it
conflicts with other properties that are even more desirable from the point of
view of usability and effective training. We introduce a class of scoring rules
that we call `Practical' scoring rules, designed to be intuitive to users in
the context of `right' vs. `wrong' probabilistic predictions. We also introduce
two specific scoring rules for prediction intervals, the `Distance' and `Order
of magnitude' rules. These rules are designed to satisfy a variety of
properties that, based on user testing, we believe are desirable for applied
calibration training.
| stat |
A Unified Framework for Time-to-Event Dose-Finding Designs | In dose-finding trials, due to staggered enrollment, it might be desirable to
make dose assignment decisions in real-time in the presence of pending toxicity
outcomes, for example, when patient accrual is fast or the dose-limiting
toxicity is late-onset. Patients' time-to-event information may be utilized to
facilitate such decisions. We propose a unified statistical framework for
time-to-event modeling in dose-finding trials, which leads to two classes of
time-to-event designs: TITE deigns and POD designs. TITE designs are based on
inference on toxicity probabilities, while POD designs are based on inference
on dose-finding decisions. These two classes of designs contain existing
designs as special cases and also give rise to new designs. We discuss and
study the theoretical properties of these designs, including large-sample
convergence properties, coherence principles, and the underlying decision
rules. To facilitate the use of time-to-event designs in practice, we introduce
efficient computational algorithms and review common practical considerations,
such as safety rules and suspension rules. Finally, the operating
characteristics of several designs are evaluated and compared through computer
simulations.
| stat |
Bayesian posterior repartitioning for nested sampling | Priors in Bayesian analyses often encode informative domain knowledge that
can be useful in making the inference process more efficient. Occasionally,
however, priors may be unrepresentative of the parameter values for a given
dataset, which can result in inefficient parameter space exploration, or even
incorrect inferences, particularly for nested sampling (NS) algorithms. Simply
broadening the prior in such cases may be inappropriate or impossible in some
applications. Hence our previous solution to this problem, known as posterior
repartitioning (PR), redefines the prior and likelihood while keeping their
product fixed, so that the posterior inferences and evidence estimates remain
unchanged, but the efficiency of the NS process is significantly increased. In
its most practical form, PR raises the prior to some power $\beta$, which is
introduced as an auxiliary variable that must be determined on a case-by-case
basis, usually by lowering $\beta$ from unity according to some pre-defined
`annealing schedule' until the resulting inferences converge to a consistent
solution. Here we present a very simple yet powerful alternative Bayesian
approach, in which $\beta$ is instead treated as a hyperparameter that is
inferred from the data alongside the original parameters of the problem, and
then marginalised over to obtain the final inference. We show through numerical
examples that this Bayesian PR (BPR) method provides a very robust,
self-adapting and computationally efficient `hands-off' solution to the problem
of unrepresentative priors in Bayesian inference using NS. Moreover, unlike the
original PR method, we show that even for representative priors BPR has a
negligible computational overhead relative to standard nesting sampling, which
suggests that it should be used as the default in all NS analyses.
| stat |
Mapping unobserved item-respondent interactions: A latent space item
response model with interaction map | Classic item response models assume that all items with the same difficulty
have the same response probability among all respondents with the same ability.
These assumptions, however, may very well be violated in practice, and it is
not straightforward to assess whether these assumptions are violated, because
neither the abilities of respondents nor the difficulties of items are
observed. An example is an educational assessment where unobserved
heterogeneity is present, arising from unobserved variables such as cultural
background and upbringing of students, the quality of mentorship and other
forms of emotional and professional support received by students, and other
unobserved variables that may affect response probabilities. To address such
violations of assumptions, we introduce a novel latent space model which
assumes that both items and respondents are embedded in an unobserved metric
space, with the probability of a correct response decreasing as a function of
the distance between the respondent's and the item's position in the latent
space. The resulting latent space approach provides an interaction map that
represents interactions of respondents and items, and helps derive insightful
diagnostic information on items as well as respondents. In practice, such
interaction maps enable teachers to detect students from underrepresented
groups who need more support than other students. We provide empirical evidence
to demonstrate the usefulness of the proposed latent space approach, along with
simulation results.
| stat |
Greedy inference with structure-exploiting lazy maps | We propose a framework for solving high-dimensional Bayesian inference
problems using \emph{structure-exploiting} low-dimensional transport maps or
flows. These maps are confined to a low-dimensional subspace (hence, lazy), and
the subspace is identified by minimizing an upper bound on the
Kullback--Leibler divergence (hence, structured). Our framework provides a
principled way of identifying and exploiting low-dimensional structure in an
inference problem. It focuses the expressiveness of a transport map along the
directions of most significant discrepancy from the posterior, and can be used
to build deep compositions of lazy maps, where low-dimensional projections of
the parameters are iteratively transformed to match the posterior. We prove
weak convergence of the generated sequence of distributions to the posterior,
and we demonstrate the benefits of the framework on challenging inference
problems in machine learning and differential equations, using inverse
autoregressive flows and polynomial maps as examples of the underlying density
estimators.
| stat |
Deep Learning for Ranking Response Surfaces with Applications to Optimal
Stopping Problems | In this paper, we propose deep learning algorithms for ranking response
surfaces, with applications to optimal stopping problems in financial
mathematics. The problem of ranking response surfaces is motivated by
estimating optimal feedback policy maps in stochastic control problems, aiming
to efficiently find the index associated to the minimal response across the
entire continuous input space $\mathcal{X} \subseteq \mathbb{R}^d$. By
considering points in $\mathcal{X}$ as pixels and indices of the minimal
surfaces as labels, we recast the problem as an image segmentation problem,
which assigns a label to every pixel in an image such that pixels with the same
label share certain characteristics. This provides an alternative method for
efficiently solving the problem instead of using sequential design in our
previous work [R. Hu and M. Ludkovski, SIAM/ASA Journal on Uncertainty
Quantification, 5 (2017), 212--239].
Deep learning algorithms are scalable, parallel and model-free, i.e., no
parametric assumptions needed on the response surfaces. Considering ranking
response surfaces as image segmentation allows one to use a broad class of deep
neural networks, e.g., UNet, SegNet, DeconvNet, which have been widely applied
and numerically proved to possess high accuracy in the field. We also
systematically study the dependence of deep learning algorithms on the input
data generated on uniform grids or by sequential design sampling, and observe
that the performance of deep learning is {\it not} sensitive to the noise and
locations (close to/away from boundaries) of training data. We present a few
examples including synthetic ones and the Bermudan option pricing problem to
show the efficiency and accuracy of this method.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.