title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Sample and Computationally Efficient Simulation Metamodeling in High
Dimensions | Stochastic kriging has been widely employed for simulation metamodeling to
predict the response surface of a complex simulation model. However, its use is
limited to cases where the design space is low-dimensional, because the number
of design points required for stochastic kriging to produce accurate
prediction, in general, grows exponentially in the dimension of the design
space. The large sample size results in both a prohibitive sample cost for
running the simulation model and a severe computational challenge due to the
need of inverting large covariance matrices. Based on tensor Markov kernels and
sparse grid experimental designs, we develop a novel methodology that
dramatically alleviates the curse of dimensionality. We show that the sample
complexity of the proposed methodology grows very mildly in the dimension, even
under model misspecification. We also develop fast algorithms that compute
stochastic kriging in its exact form without any approximation schemes. We
demonstrate via extensive numerical experiments that our methodology can handle
problems with a design space of hundreds of dimensions, improving both
prediction accuracy and computational efficiency by orders of magnitude
relative to typical alternative methods in practice.
| stat |
Visualization of Covariance Structures for Multivariate Spatio-Temporal
Random Fields | The prevalence of multivariate space-time data collected from monitoring
networks and satellites or generated from numerical models has brought much
attention to multivariate spatio-temporal statistical models, where the
covariance function plays a key role in modeling, inference, and prediction.
For multivariate space-time data, understanding the spatio-temporal
variability, within and across variables, is essential in employing a realistic
covariance model. Meanwhile, the complexity of generic covariances often makes
model fitting very challenging, and simplified covariance structures, including
symmetry and separability, can reduce the model complexity and facilitate the
inference procedure. However, a careful examination of these properties is
needed in real applications. In the work presented here, we formally define
these properties for multivariate spatio-temporal random fields and use
functional data analysis techniques to visualize them, hence providing
intuitive interpretations. We then propose a rigorous rank-based testing
procedure to conclude whether the simplified properties of covariance are
suitable for the underlying multivariate space-time data. The good performance
of our method is illustrated through synthetic data, for which we know the true
structure. We also investigate the covariance of bivariate wind speed, a key
variable in renewable energy, over a coastal and an inland area in Saudi
Arabia.
| stat |
PoD-BIN: A Probability of Decision Bayesian Interval Design for
Time-to-Event Dose-Finding Trials with Multiple Toxicity Grades | We consider a Bayesian framework based on "probability of decision" for
dose-finding trial designs. The proposed PoD-BIN design evaluates the posterior
predictive probabilities of up-and-down decisions. In PoD-BIN, multiple grades
of toxicity, categorized as the mild toxicity (MT) and dose-limiting toxicity
(DLT), are modeled simultaneously, and the primary outcome of interests is
time-to-toxicity for both MT and DLT. This allows the possibility of enrolling
new patients when previously enrolled patients are still being followed for
toxicity, thus potentially shortening trial length. The Bayesian decision rules
in PoD-BIN utilize the probability of decisions to balance the need to speed up
the trial and the risk of exposing patients to overly toxic doses. We
demonstrate via numerical examples the resulting balance of speed and safety of
PoD-BIN and compare to existing designs.
| stat |
Randomization as Regularization: A Degrees of Freedom Explanation for
Random Forest Success | Random forests remain among the most popular off-the-shelf supervised machine
learning tools with a well-established track record of predictive accuracy in
both regression and classification settings. Despite their empirical success as
well as a bevy of recent work investigating their statistical properties, a
full and satisfying explanation for their success has yet to be put forth. Here
we aim to take a step forward in this direction by demonstrating that the
additional randomness injected into individual trees serves as a form of
implicit regularization, making random forests an ideal model in low
signal-to-noise ratio (SNR) settings. Specifically, from a model-complexity
perspective, we show that the mtry parameter in random forests serves much the
same purpose as the shrinkage penalty in explicitly regularized regression
procedures like lasso and ridge regression. To highlight this point, we design
a randomized linear-model-based forward selection procedure intended as an
analogue to tree-based random forests and demonstrate its surprisingly strong
empirical performance. Numerous demonstrations on both real and synthetic data
are provided.
| stat |
Optimal Statistical Rates for Decentralised Non-Parametric Regression
with Linear Speed-Up | We analyse the learning performance of Distributed Gradient Descent in the
context of multi-agent decentralised non-parametric regression with the square
loss function when i.i.d. samples are assigned to agents. We show that if
agents hold sufficiently many samples with respect to the network size, then
Distributed Gradient Descent achieves optimal statistical rates with a number
of iterations that scales, up to a threshold, with the inverse of the spectral
gap of the gossip matrix divided by the number of samples owned by each agent
raised to a problem-dependent power. The presence of the threshold comes from
statistics. It encodes the existence of a "big data" regime where the number of
required iterations does not depend on the network topology. In this regime,
Distributed Gradient Descent achieves optimal statistical rates with the same
order of iterations as gradient descent run with all the samples in the
network. Provided the communication delay is sufficiently small, the
distributed protocol yields a linear speed-up in runtime compared to the
single-machine protocol. This is in contrast to decentralised optimisation
algorithms that do not exploit statistics and only yield a linear speed-up in
graphs where the spectral gap is bounded away from zero. Our results exploit
the statistical concentration of quantities held by agents and shed new light
on the interplay between statistics and communication in decentralised methods.
Bounds are given in the standard non-parametric setting with source/capacity
assumptions.
| stat |
A statistical analysis of time trends in atmospheric ethane | Ethane is the most abundant non-methane hydrocarbon in the Earth's atmosphere
and an important precursor of tropospheric ozone through various chemical
pathways. Ethane is also an indirect greenhouse gas (global warming potential),
influencing the atmospheric lifetime of methane through the consumption of the
hydroxyl radical (OH). Understanding the development of trends and identifying
trend reversals in atmospheric ethane is therefore crucial. Our dataset
consists of four series of daily ethane columns obtained from ground-based FTIR
measurements. As many other decadal time series, our data are characterized by
autocorrelation, heteroskedasticity, and seasonal effects. Additionally,
missing observations due to instrument failure or unfavorable measurement
conditions are common in such series. The goal of this paper is therefore to
analyze trends in atmospheric ethane with statistical tools that correctly
address these data features. We present selected methods designed for the
analysis of time trends and trend reversals. We consider bootstrap inference on
broken linear trends and smoothly varying nonlinear trends. In particular, for
the broken trend model, we propose a bootstrap method for inference on the
break location and the corresponding changes in slope. For the smooth trend
model we construct simultaneous confidence bands around the nonparametrically
estimated trend. Our autoregressive wild bootstrap approach, combined with a
seasonal filter, is able to handle all issues mentioned above.
| stat |
Deep Divergence-Based Approach to Clustering | A promising direction in deep learning research consists in learning
representations and simultaneously discovering cluster structure in unlabeled
data by optimizing a discriminative loss function. As opposed to supervised
deep learning, this line of research is in its infancy, and how to design and
optimize suitable loss functions to train deep neural networks for clustering
is still an open question. Our contribution to this emerging field is a new
deep clustering network that leverages the discriminative power of
information-theoretic divergence measures, which have been shown to be
effective in traditional clustering. We propose a novel loss function that
incorporates geometric regularization constraints, thus avoiding degenerate
structures of the resulting clustering partition. Experiments on synthetic
benchmarks and real datasets show that the proposed network achieves
competitive performance with respect to other state-of-the-art methods, scales
well to large datasets, and does not require pre-training steps.
| stat |
On a Bayesian Approach to Malware Detection and Classification through
$n$-gram Profiles | Detecting and correctly classifying malicious executables has become one of
the major concerns in cyber security, especially because traditional detection
systems have become less effective with the increasing number and danger of
threats found nowadays. One way to differentiate benign from malicious
executables is to leverage on their hexadecimal representation by creating a
set of binary features that completely characterise each executable. In this
paper we present a novel supervised learning Bayesian nonparametric approach
for binary matrices, that provides an effective probabilistic approach for
malware detection. Moreover, and due to the model's flexible assumptions, we
are able to use it in a multi-class framework where the interest relies in
classifying malware into known families. Finally, a generalisation of the model
which provides a deeper understanding of the behaviour across groups for each
feature is also developed.
| stat |
A variable selection approach for highly correlated predictors in
high-dimensional genomic data | In genomic studies, identifying biomarkers associated with a variable of
interest is a major concern in biomedical research. Regularized approaches are
classically used to perform variable selection in high-dimensional linear
models. However, these methods can fail in highly correlated settings. We
propose a novel variable selection approach called WLasso, taking these
correlations into account. It consists in rewriting the initial
high-dimensional linear model to remove the correlation between the biomarkers
(predictors) and in applying the generalized Lasso criterion. The performance
of WLasso is assessed using synthetic data in several scenarios and compared
with recent alternative approaches. The results show that when the biomarkers
are highly correlated, WLasso outperforms the other approaches in sparse
high-dimensional frameworks. The method is also successfully illustrated on
publicly available gene expression data in breast cancer. Our method is
implemented in the WLasso R package which is available from the Comprehensive R
Archive Network.
| stat |
Dynamic Linear Discriminant Analysis in High Dimensional Space | High-dimensional data that evolve dynamically feature predominantly in the
modern data era. As a partial response to this, recent years have seen
increasing emphasis to address the dimensionality challenge. However, the
non-static nature of these datasets is largely ignored. This paper addresses
both challenges by proposing a novel yet simple dynamic linear programming
discriminant (DLPD) rule for binary classification. Different from the usual
static linear discriminant analysis, the new method is able to capture the
changing distributions of the underlying populations by modeling their means
and covariances as smooth functions of covariates of interest. Under an
approximate sparse condition, we show that the conditional misclassification
rate of the DLPD rule converges to the Bayes risk in probability uniformly over
the range of the variables used for modeling the dynamics, when the
dimensionality is allowed to grow exponentially with the sample size. The
minimax lower bound of the estimation of the Bayes risk is also established,
implying that the misclassification rate of our proposed rule is minimax-rate
optimal. The promising performance of the DLPD rule is illustrated via
extensive simulation studies and the analysis of a breast cancer dataset.
| stat |
Nonnegative Matrix Factorization with Toeplitz Penalty | Nonnegative Matrix Factorization (NMF) is an unsupervised learning algorithm
that produces a linear, parts-based approximation of a data matrix. NMF
constructs a nonnegative low rank basis matrix and a nonnegative low rank
matrix of weights which, when multiplied together, approximate the data matrix
of interest using some cost function. The NMF algorithm can be modified to
include auxiliary constraints which impose task-specific penalties or
restrictions on the cost function of the matrix factorization. In this paper we
propose a new NMF algorithm that makes use of non-data dependent auxiliary
constraints which incorporate a Toeplitz matrix into the multiplicative
updating of the basis and weight matrices. We compare the facial recognition
performance of our new Toeplitz Nonnegative Matrix Factorization (TNMF)
algorithm to the performance of the Zellner Nonnegative Matrix Factorization
(ZNMF) algorithm which makes use of data-dependent auxiliary constraints. We
also compare the facial recognition performance of the two aforementioned
algorithms with the performance of several preexisting constrained NMF
algorithms that have non-data-dependent penalties. The facial recognition
performances are evaluated using the Cambridge ORL Database of Faces and the
Yale Database of Faces.
| stat |
Confidence Intervals for Policy Evaluation in Adaptive Experiments | Adaptive experiment designs can dramatically improve statistical efficiency
in randomized trials, but they also complicate statistical inference. For
example, it is now well known that the sample mean is biased in adaptive
trials. Inferential challenges are exacerbated when our parameter of interest
differs from the parameter the trial was designed to target, such as when we
are interested in estimating the value of a sub-optimal treatment after running
a trial to determine the optimal treatment using a stochastic bandit design. In
this context, typical estimators that use inverse propensity weighting to
eliminate sampling bias can be problematic: their distributions become skewed
and heavy-tailed as the propensity scores decay to zero. In this paper, we
present a class of estimators that overcome these issues. Our approach is to
adaptively reweight the terms of an augmented inverse propensity weighting
estimator to control the contribution of each term to the estimator's variance.
This adaptive weighting scheme prevents estimates from becoming heavy-tailed,
ensuring asymptotically correct coverage. It also reduces variance, allowing us
to test hypotheses with greater power - especially hypotheses that were not
targeted by the experimental design. We validate the accuracy of the resulting
estimates and their confidence intervals in numerical experiments and show our
methods compare favorably to existing alternatives in terms of RMSE and
coverage.
| stat |
The Lipschitz Constant of Self-Attention | Lipschitz constants of neural networks have been explored in various contexts
in deep learning, such as provable adversarial robustness, estimating
Wasserstein distance, stabilising training of GANs, and formulating invertible
neural networks. Such works have focused on bounding the Lipschitz constant of
fully connected or convolutional networks, composed of linear maps and
pointwise non-linearities. In this paper, we investigate the Lipschitz constant
of self-attention, a non-linear neural network module widely used in sequence
modelling. We prove that the standard dot-product self-attention is not
Lipschitz, and propose an alternative L2 self-attention that is Lipschitz. We
derive an upper bound on the Lipschitz constant of L2 self-attention and
provide empirical evidence for its asymptotic tightness. To demonstrate the
practical relevance of the theory, we formulate invertible self-attention and
use it in a Transformer-based architecture for a character-level language
modelling task.
| stat |
Posterior Inference for Sparse Hierarchical Non-stationary Models | Gaussian processes are valuable tools for non-parametric modelling, where
typically an assumption of stationarity is employed. While removing this
assumption can improve prediction, fitting such models is challenging. In this
work, hierarchical models are constructed based on Gaussian Markov random
fields with stochastic spatially varying parameters. Importantly, this allows
for non-stationarity while also addressing the computational burden through a
sparse banded representation of the precision matrix. In this setting,
efficient Markov chain Monte Carlo (MCMC) sampling is challenging due to the
strong coupling a posteriori of the parameters and hyperparameters. We develop
and compare three adaptive MCMC schemes and make use of banded matrix
operations for faster inference. Furthermore, a novel extension to
multi-dimensional settings is proposed through an additive structure that
retains the flexibility and scalability of the model, while also inheriting
interpretability from the additive approach. A thorough assessment of the
efficiency and accuracy of the methods in nonstationary settings is presented
for both simulated experiments and a computer emulation problem.
| stat |
Structured Machine Learning Tools for Modelling Characteristics of
Guided Waves | The use of ultrasonic guided waves to probe the materials/structures for
damage continues to increase in popularity for non-destructive evaluation (NDE)
and structural health monitoring (SHM). The use of high-frequency waves such as
these offers an advantage over low-frequency methods from their ability to
detect damage on a smaller scale. However, in order to assess damage in a
structure, and implement any NDE or SHM tool, knowledge of the behaviour of a
guided wave throughout the material/structure is important (especially when
designing sensor placement for SHM systems). Determining this behaviour is
extremely diffcult in complex materials, such as fibre-matrix composites, where
unique phenomena such as continuous mode conversion takes place. This paper
introduces a novel method for modelling the feature-space of guided waves in a
composite material. This technique is based on a data-driven model, where prior
physical knowledge can be used to create structured machine learning tools;
where constraints are applied to provide said structure. The method shown makes
use of Gaussian processes, a full Bayesian analysis tool, and in this paper it
is shown how physical knowledge of the guided waves can be utilised in
modelling using an ML tool. This paper shows that through careful consideration
when applying machine learning techniques, more robust models can be generated
which offer advantages such as extrapolation ability and physical
interpretation.
| stat |
Time-varying auto-regressive models for count time-series | Count-valued time series data are routinely collected in many application
areas. We are particularly motivated to study the count time series of daily
new cases, arising from COVID-19 spread. We propose two Bayesian models, a
time-varying semiparametric AR(p) model for count and then a time-varying
INGARCH model considering the rapid changes in the spread. We calculate
posterior contraction rates of the proposed Bayesian methods with respect to
average Hellinger metric. 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 |
Dealing with multiple testing: To adjust or not to adjust | Multiple testing problems arise naturally in scientific studies because of
the need to capture or convey more information with more variables. The
literature is enormous, but the emphasis is primarily methodological, providing
numerous methods with their mathematical justification and practical
implementation. Our aim is to highlight the logical issues involved in the
application of multiple testing adjustment.
| stat |
Hypocoercivity of Piecewise Deterministic Markov Process-Monte Carlo | In this work, we establish $\mathrm{L}^2$-exponential convergence for a broad
class of Piecewise Deterministic Markov Processes recently proposed in the
context of Markov Process Monte Carlo methods and covering in particular the
Randomized Hamiltonian Monte Carlo, the Zig-Zag process and the Bouncy Particle
Sampler. The kernel of the symmetric part of the generator of such processes is
non-trivial, and we follow the ideas recently introduced by (Dolbeault et al.,
2009, 2015) to develop a rigorous framework for hypocoercivity in a fairly
general and unifying set-up, while deriving tractable estimates of the
constants involved in terms of the parameters of the dynamics. As a by-product
we characterize the scaling properties of these algorithms with respect to the
dimension of classes of problems, therefore providing some theoretical evidence
to support their practical relevance.
| stat |
Causal inference with recurrent data via inverse probability treatment
weighting method (IPTW) | Propensity score methods are increasingly being used to reduce estimation
bias of treatment effects for observational studies. Previous research has
shown that propensity score methods consistently estimate the marginal hazard
ratio for time to event data. However, recurrent data frequently arise in the
biomedical literature and there is a paucity of research into the use of
propensity score methods when data are recurrent in nature. The objective of
this paper is to extend the existing propensity score methods to recurrent data
setting. We illustrate our methods through a series of Monte Carlo simulations.
The simulation results indicate that without the presence of censoring, the
IPTW estimators allow us to consistently estimate the marginal hazard ratio for
each event. Under administrative censoring regime, the stabilized IPTW
estimator yields biased estimate of the marginal hazard ratio, and the degree
of bias depends on the proportion of subjects being censored. For variance
estimation, the na\"ive variance estimator often tends to substantially
underestimate the variance of the IPTW estimator, while the robust variance
estimator significantly reduces the estimation bias of the variance.
| stat |
Hybrid Density- and Partition-based Clustering Algorithm for Data with
Mixed-type Variables | Clustering is an essential technique for discovering patterns in data. The
steady increase in amount and complexity of data over the years led to
improvements and development of new clustering algorithms. However, algorithms
that can cluster data with mixed variable types (continuous and categorical)
remain limited, despite the abundance of data with mixed types particularly in
the medical field. Among existing methods for mixed data, some posit
unverifiable distributional assumptions or that the contributions of different
variable types are not well balanced.
We propose a two-step hybrid density- and partition-based algorithm (HyDaP)
that can detect clusters after variables selection. The first step involves
both density-based and partition-based algorithms to identify the data
structure formed by continuous variables and recognize the important variables
for clustering; the second step involves partition-based algorithm together
with a novel dissimilarity measure we designed for mixed data to obtain
clustering results. Simulations across various scenarios and data structures
were conducted to examine the performance of the HyDaP algorithm compared to
commonly used methods. We also applied the HyDaP algorithm on electronic health
records to identify sepsis phenotypes.
| stat |
Estimating accuracy of the MCMC variance estimator: a central limit
theorem for batch means estimators | The batch means estimator of the MCMC variance is a simple and effective
measure of accuracy for MCMC based ergodic averages. Under various regularity
conditions, the estimator has been shown to be consistent for the true
variance. However, the estimator can be unstable in practice as it depends
directly on the raw MCMC output. A measure of accuracy of the batch means
estimator itself, ideally in the form of a confidence interval, is therefore
desirable. The asymptotic variance of the batch means estimator is known;
however, without any knowledge of asymptotic distribution, asymptotic variances
are in general insufficient to describe variability. In this article we prove a
central limit theorem for the batch means estimator that allows for the
construction of asymptotically accurate confidence intervals for the batch
means estimator. Additionally, our results provide a Markov chain analogue of
the classical CLT for the sample variance parameter for i.i.d. observations.
Our result assumes standard regularity conditions similar to the ones assumed
in the literature for proving consistency. Simulated and real data examples are
included as illustrations and applications of the CLT.
| stat |
lgpr: An interpretable nonparametric method for inferring covariate
effects from longitudinal data | Longitudinal study designs are indispensable for studying disease
progression. Inferring covariate effects from longitudinal data, however,
requires interpretable methods that can model complicated covariance structures
and detect nonlinear effects of both categorical and continuous covariates, as
well as their interactions. Detecting disease effects is hindered by the fact
that they often occur rapidly near the disease initiation time, and this time
point cannot be exactly observed. An additional challenge is that the effect
magnitude can be heterogeneous over the subjects. We present lgpr, a widely
applicable and interpretable method for nonparametric analysis of longitudinal
data using additive Gaussian processes. We demonstrate that it outperforms
previous approaches in identifying the relevant categorical and continuous
covariates in various settings. Furthermore, it implements important novel
features, including the ability to account for the heterogeneity of covariate
effects, their temporal uncertainty, and appropriate observation models for
different types of biomedical data. The lgpr tool is implemented as a
comprehensive and user-friendly R-package. lgpr is available at
jtimonen.github.io/lgpr-usage with documentation, tutorials, test data, and
code for reproducing the experiments of this paper.
| stat |
Importance of spatial predictor variable selection in machine learning
applications -- Moving from data reproduction to spatial prediction | Machine learning algorithms find frequent application in spatial prediction
of biotic and abiotic environmental variables. However, the characteristics of
spatial data, especially spatial autocorrelation, are widely ignored. We
hypothesize that this is problematic and results in models that can reproduce
training data but are unable to make spatial predictions beyond the locations
of the training samples. We assume that not only spatial validation strategies
but also spatial variable selection is essential for reliable spatial
predictions. We introduce two case studies that use remote sensing to predict
land cover and the leaf area index for the "Marburg Open Forest", an open
research and education site of Marburg University, Germany. We use the machine
learning algorithm Random Forests to train models using non-spatial and spatial
cross-validation strategies to understand how spatial variable selection
affects the predictions. Our findings confirm that spatial cross-validation is
essential in preventing overoptimistic model performance. We further show that
highly autocorrelated predictors (such as geolocation variables, e.g. latitude,
longitude) can lead to considerable overfitting and result in models that can
reproduce the training data but fail in making spatial predictions. The problem
becomes apparent in the visual assessment of the spatial predictions that show
clear artefacts that can be traced back to a misinterpretation of the spatially
autocorrelated predictors by the algorithm. Spatial variable selection could
automatically detect and remove such variables that lead to overfitting,
resulting in reliable spatial prediction patterns and improved statistical
spatial model performance. We conclude that in addition to spatial validation,
a spatial variable selection must be considered in spatial predictions of
ecological data to produce reliable predictions.
| stat |
Bayesian biclustering for microbial metagenomic sequencing data via
multinomial matrix factorization | High-throughput sequencing technology provides unprecedented opportunities to
quantitatively explore human gut microbiome and its relation to diseases.
Microbiome data are compositional, sparse, noisy, and heterogeneous, which pose
serious challenges for statistical modeling. We propose an identifiable
Bayesian multinomial matrix factorization model to infer overlapping clusters
on both microbes and hosts. The proposed method represents the observed
over-dispersed zero-inflated count matrix as Dirichlet-multinomial mixtures on
which latent cluster structures are built hierarchically. Under the Bayesian
framework, the number of clusters is automatically determined and available
information from a taxonomic rank tree of microbes is naturally incorporated,
which greatly improves the interpretability of our findings. We demonstrate the
utility of the proposed approach by comparing to alternative methods in
simulations. An application to a human gut microbiome dataset involving
patients with inflammatory bowel disease reveals interesting clusters, which
contain bacteria families Bacteroidaceae, Bifidobacteriaceae,
Enterobacteriaceae, Fusobacteriaceae, Lachnospiraceae, Ruminococcaceae,
Pasteurellaceae, and Porphyromonadaceae that are known to be related to the
inflammatory bowel disease and its subtypes according to biological literature.
Our findings can help generate potential hypotheses for future investigation of
the heterogeneity of the human gut microbiome.
| stat |
Two-sample inference for high-dimensional Markov networks | Markov networks are frequently used in sciences to represent conditional
independence relationships underlying observed variables arising from a complex
system. It is often of interest to understand how an underlying network differs
between two conditions. In this paper, we develop methods for comparing a pair
of high-dimensional Markov networks where we allow the number of observed
variables to increase with the sample sizes. By taking the density ratio
approach, we are able to learn the network difference directly and avoid
estimating the individual graphs. Our methods are thus applicable even when the
individual networks are dense as long as their difference is sparse. We prove
finite-sample Gaussian approximation error bounds for the estimator we
construct under significantly weaker assumptions than are typically required
for model selection consistency. Furthermore, we propose bootstrap procedures
for estimating quantiles of a max-type statistics based on our estimator, and
show how they can be used to test the equality of two Markov networks or
construct simultaneous confidence intervals. The performance of our methods is
demonstrated through extensive simulations. The scientific usefulness is
illustrated with an analysis of a new fMRI dataset.
| stat |
Stratification as a general variance reduction method for Markov chain
Monte Carlo | The Eigenvector Method for Umbrella Sampling (EMUS) belongs to a popular
class of methods in statistical mechanics which adapt the principle of
stratified survey sampling to the computation of free energies. We develop a
detailed theoretical analysis of EMUS. Based on this analysis, we show that
EMUS is an efficient general method for computing averages over arbitrary
target distributions. In particular, we show that EMUS can be dramatically more
efficient than direct MCMC when the target distribution is multimodal or when
the goal is to compute tail probabilities. To illustrate these theoretical
results, we present a tutorial application of the method to a problem from
Bayesian statistics.
| stat |
Comparison of non-parametric global envelopes | This study presents a simulation study to compare different non-parametric
global envelopes that are refinements of the rank envelope proposed by
Myllym\"aki et al. (2017, Global envelope tests for spatial processes, J. R.
Statist. Soc. B 79, 381-404, doi: 10.1111/rssb.12172). The global envelopes are
constructed for a set of functions or vectors. For a large number of vectors,
all the refinements lead to the same outcome as the global rank envelope. For
smaller numbers of vectors the refinement playes a role, where different
refinements are sensitive to different types of extremeness of a vector among
the set of vectors. The performance of the different alternatives are compared
in a simulation study with respect to the numbers of available vectors, the
dimensionality of the vectors, the amount of dependence between the vector
elements and the expected type of extremeness.
| stat |
The Linear Lasso: a location model resolution | We use location model methodology to guide the least squares analysis of the
Lasso problem of variable selection and inference. The nuisance parameter is
taken to be an indicator for the selection of explanatory variables and the
interest parameter is the response variable itself. Recent theory eliminates
the nuisance parameter by marginalization on the data space and then uses the
resulting distribution for inference concerning the interest parameter. We
develop this approach and find: that primary inference is essentially
one-dimensional rather than $n$-dimensional; that inference focuses on the
response variable itself rather than the least squares estimate (as variables
are removed); that first order probabilities are available; that computation is
relatively easy; that a scalar marginal model is available; and that
ineffective variables can be removed by distributional tilt or shift.
| stat |
Modification Indices for Diagnostic Classification Models | Diagnostic classification models (DCMs) are psychometric models for
evaluating a student's mastery of the essential skills in a content domain
based upon their responses to a set of test items. Currently, diagnostic model
and/or Q-matrix misspecification is a known problem with limited avenues for
remediation. To address this problem, this paper defines a one-sided score
statistic that is a computationally efficient method for detecting
under-specification of both the Q-matrix and the model parameters of the
particular DCM chosen in the analysis. This method is analogous to the
modification indices widely used in structural equation modeling. The results
of a simulation study show the Type I error rate of modification indices for
DCMs are acceptably close to the nominal significance level when the
appropriate mixture chi-squared reference distribution is used. The simulation
results indicate that modification indices are very powerful in the detection
of an under-specified Q-matrix and have ample power to detect the omission of
model parameters in large samples or when the items are highly discriminating.
An application of modification indices for DCMs to analysis of response data
from a large-scale administration of a diagnostic test demonstrates how they
can be useful in diagnostic model refinement.
| stat |
Regression Models for Order-of-Addition Experiments | The purpose of order-of-addition (OofA) experiments is to identify the best
order in a sequence of m components in a system or treatment. Such experiments
may be analysed by various regression models, the most popular ones being based
on pairwise ordering (PWO) factors or on component-position (CP) factors. This
paper reviews these models and extensions and proposes a new class of models
based on response surface (RS) regression using component position numbers as
predictor variables. Using two published examples, it is shown that RS models
can be quite competitive. In case of model uncertainty, we advocate the use of
model averaging for analysis. The averaging idea leads naturally to a design
approach based on a compound optimality criterion assigning weights to each
candidate model.
| stat |
Optimal Experimental Design for Mathematical Models of Hematopoiesis | The hematopoietic system has a highly regulated and complex structure in
which cells are organized to successfully create and maintain new blood cells.
Feedback regulation is crucial to tightly control this system, but the specific
mechanisms by which control is exerted are not completely understood. In this
work, we aim to uncover the underlying mechanisms in hematopoiesis by
conducting perturbation experiments, where animal subjects are exposed to an
external agent in order to observe the system response and evolution.
Developing a proper experimental design for these studies is an extremely
challenging task. To address this issue, we have developed a novel Bayesian
framework for optimal design of perturbation experiments. We model the numbers
of hematopoietic stem and progenitor cells in mice that are exposed to a low
dose of radiation. We use a differential equations model that accounts for
feedback and feedforward regulation. A significant obstacle is that the
experimental data are not longitudinal, rather each data point corresponds to a
different animal. This model is embedded in a hierarchical framework with
latent variables that capture unobserved cellular population levels. We select
the optimum design based on the amount of information gain, measured by the
Kullback-Leibler divergence between the probability distributions before and
after observing the data. We evaluate our approach using synthetic and
experimental data. We show that a proper design can lead to better estimates of
model parameters even with relatively few subjects. Additionally, we
demonstrate that the model parameters show a wide range of sensitivities to
design options. Our method should allow scientists to find the optimal design
by focusing on their specific parameters of interest and provide insight to
hematopoiesis. Our approach can be extended to more complex models where latent
components are used.
| stat |
Some Developments in Clustering Analysis on Stochastic Processes | We review some developments on clustering stochastic processes and come with
the conclusion that asymptotically consistent clustering algorithms can be
obtained when the processes are ergodic and the dissimilarity measure satisfies
the triangle inequality. Examples are provided when the processes are
distribution ergodic, covariance ergodic and locally asymptotically
self-similar, respectively.
| stat |
Calibration of wind speed ensemble forecasts for power generation | In the last decades wind power became the second largest energy source in the
EU covering 16% of its electricity demand. However, due to its volatility,
accurate short range wind power predictions are required for successful
integration of wind energy into the electrical grid. Accurate predictions of
wind power require accurate hub height wind speed forecasts, where the state of
the art method is the probabilistic approach based on ensemble forecasts
obtained from multiple runs of numerical weather prediction models.
Nonetheless, ensemble forecasts are often uncalibrated and might also be
biased, thus require some form of post-processing to improve their predictive
performance. We propose a novel flexible machine learning approach for
calibrating wind speed ensemble forecasts, which results in a truncated normal
predictive distribution. In a case study based on 100m wind speed forecasts
produced by the operational ensemble prediction system of the Hungarian
Meteorological Service, the forecast skill of this method is compared with the
predictive performance of three different ensemble model output statistics
approaches and the raw ensemble forecasts. We show that compared with the raw
ensemble, post-processing always improves the calibration of probabilistic and
accuracy of point forecasts and from the four competing methods the novel
machine learning based approach results in the best overall performance.
| stat |
Outlier-robust Kalman Filter in the Presence of Correlated Measurements | We consider the robust filtering problem for a state-space model with
outliers in correlated measurements. We propose a new robust filtering
framework to further improve the robustness of conventional robust filters.
Specifically, the measurement fitting error is processed separately during the
reweighting procedure, which differs from existing solutions where a jointly
processed scheme is involved. Simulation results reveal that, under the same
setup, the proposed method outperforms the existing robust filter when the
outlier-contaminated measurements are correlated, while it has the same
performance as the existing one in the presence of uncorrelated measurements
since these two types of robust filters are equivalent under such a
circumstance.
| stat |
(Sequential) Importance Sampling Bandits | This work extends existing multi-armed bandit (MAB) algorithms beyond their
original settings by leveraging advances in sequential Monte Carlo (SMC)
methods from the approximate inference community. We leverage Monte Carlo
estimation and, in particular, the flexibility of (sequential) importance
sampling to allow for accurate estimation of the statistics of interest within
the MAB problem. The MAB is a sequential allocation task where the goal is to
learn a policy that maximizes long term payoff, where only the reward of the
executed action is observed; i.e., sequential optimal decisions are made, while
simultaneously learning how the world operates. In the stochastic setting, the
reward for each action is generated from an unknown distribution. To decide the
next optimal action to take, one must compute sufficient statistics of this
unknown reward distribution, e.g., upper-confidence bounds (UCB), or
expectations in Thompson sampling. Closed-form expressions for these statistics
of interest are analytically intractable except for simple cases. By combining
SMC methods --- which estimate posterior densities and expectations in
probabilistic models that are analytically intractable --- with Bayesian
state-of-the-art MAB algorithms, we extend their applicability to complex
models: those for which sampling may be performed even if analytic computation
of summary statistics is infeasible --- nonlinear reward functions and dynamic
bandits. We combine SMC both for Thompson sampling and upper confident
bound-based (Bayes-UCB) policies, and study different bandit models: classic
Bernoulli and Gaussian distributed cases, as well as dynamic and context
dependent linear-Gaussian, logistic and categorical-softmax rewards.
| stat |
Multi-Rank Sparse and Functional PCA: Manifold Optimization and
Iterative Deflation Techniques | We consider the problem of estimating multiple principal components using the
recently-proposed Sparse and Functional Principal Components Analysis (SFPCA)
estimator. We first propose an extension of SFPCA which estimates several
principal components simultaneously using manifold optimization techniques to
enforce orthogonality constraints. While effective, this approach is
computationally burdensome so we also consider iterative deflation approaches
which take advantage of existing fast algorithms for rank-one SFPCA. We show
that alternative deflation schemes can more efficiently extract signal from the
data, in turn improving estimation of subsequent components. Finally, we
compare the performance of our manifold optimization and deflation techniques
in a scenario where orthogonality does not hold and find that they still lead
to significantly improved performance.
| stat |
Inferring the Direction of a Causal Link and Estimating Its Effect via a
Bayesian Mendelian Randomization Approach | The use of genetic variants as instrumental variables - an approach known as
Mendelian randomization - is a popular epidemiological method for estimating
the causal effect of an exposure (phenotype, biomarker, risk factor) on a
disease or health-related outcome from observational data. Instrumental
variables must satisfy strong, often untestable assumptions, which means that
finding good genetic instruments among a large list of potential candidates is
challenging. This difficulty is compounded by the fact that many genetic
variants influence more than one phenotype through different causal pathways, a
phenomenon called horizontal pleiotropy. This leads to errors not only in
estimating the magnitude of the causal effect but also in inferring the
direction of the putative causal link. In this paper, we propose a Bayesian
approach called BayesMR that is a generalization of the Mendelian randomization
technique in which we allow for pleiotropic effects and, crucially, for the
possibility of reverse causation. The output of the method is a posterior
distribution over the target causal effect, which provides an immediate and
easily interpretable measure of the uncertainty in the estimation. More
importantly, we use Bayesian model averaging to determine how much more likely
the inferred direction is relative to the reverse direction.
| stat |
Feature Gradients: Scalable Feature Selection via Discrete Relaxation | In this paper we introduce Feature Gradients, a gradient-based search
algorithm for feature selection. Our approach extends a recent result on the
estimation of learnability in the sublinear data regime by showing that the
calculation can be performed iteratively (i.e., in mini-batches) and in linear
time and space with respect to both the number of features D and the sample
size N . This, along with a discrete-to-continuous relaxation of the search
domain, allows for an efficient, gradient-based search algorithm among feature
subsets for very large datasets. Crucially, our algorithm is capable of finding
higher-order correlations between features and targets for both the N > D and N
< D regimes, as opposed to approaches that do not consider such interactions
and/or only consider one regime. We provide experimental demonstration of the
algorithm in small and large sample-and feature-size settings.
| stat |
Identification of Latent Variables From Graphical Model Residuals | Graph-based causal discovery methods aim to capture conditional
independencies consistent with the observed data and differentiate causal
relationships from indirect or induced ones. Successful construction of
graphical models of data depends on the assumption of causal sufficiency: that
is, that all confounding variables are measured. When this assumption is not
met, learned graphical structures may become arbitrarily incorrect and effects
implied by such models may be wrongly attributed, carry the wrong magnitude, or
mis-represent direction of correlation. Wide application of graphical models to
increasingly less curated "big data" draws renewed attention to the unobserved
confounder problem.
We present a novel method that aims to control for the latent space when
estimating a DAG by iteratively deriving proxies for the latent space from the
residuals of the inferred model. Under mild assumptions, our method improves
structural inference of Gaussian graphical models and enhances identifiability
of the causal effect. In addition, when the model is being used to predict
outcomes, it un-confounds the coefficients on the parents of the outcomes and
leads to improved predictive performance when out-of-sample regime is very
different from the training data. We show that any improvement of prediction of
an outcome is intrinsically capped and cannot rise beyond a certain limit as
compared to the confounded model. We extend our methodology beyond GGMs to
ordinal variables and nonlinear cases. Our R package provides both PCA and
autoencoder implementations of the methodology, suitable for GGMs with some
guarantees and for better performance in general cases but without such
guarantees.
| stat |
Bounding bias due to selection | When epidemiologic studies are conducted in a subset of the population,
selection bias can threaten the validity of causal inference. This bias can
occur whether or not that selected population is the target population, and can
occur even in the absence of exposure-outcome confounding. However, it is often
difficult to quantify the extent of selection bias, and sensitivity analysis
can be challenging to undertake and to understand. In this article we
demonstrate that the magnitude of the bias due to selection can be bounded by
simple expressions defined by parameters characterizing the relationships
between unmeasured factor(s) responsible for the bias and the measured
variables. No functional form assumptions are necessary about those unmeasured
factors. Using knowledge about the selection mechanism, researchers can account
for the possible extent of selection bias by specifying the size of the
parameters in the bounds. We also show that the bounds, which differ depending
on the target population, result in summary measures that can be used to
calculate the minimum magnitude of the parameters required to shift a risk
ratio to the null. The summary measure can be used to determine the overall
strength of selection that would be necessary to explain away a result. We then
show that the bounds and summary measures can be simplified in certain contexts
or with certain assumptions. Using examples with varying selection mechanisms,
we also demonstrate how researchers can implement these simple sensitivity
analyses.
| stat |
Bayesian interpretation of SGD as Ito process | The current interpretation of stochastic gradient descent (SGD) as a
stochastic process lacks generality in that its numerical scheme restricts
continuous-time dynamics as well as the loss function and the distribution of
gradient noise. We introduce a simplified scheme with milder conditions that
flexibly interprets SGD as a discrete-time approximation of an Ito process. The
scheme also works as a common foundation of SGD and stochastic gradient
Langevin dynamics (SGLD), providing insights into their asymptotic properties.
We investigate the convergence of SGD with biased gradient in terms of the
equilibrium mode and the overestimation problem of the second moment of SGLD.
| stat |
How robust are Structural Equation Models to model miss-specification? A
simulation study | Structural Equation Models (SEMs) are routinely used in the analysis of
empirical data by researchers from different scientific fields such as
psychologists or economists. In some fields, such as in ecology, SEMs have only
started recently to attract attention and thanks to dedicated software packages
the use of SEMs has steadily increased. Yet, common analysis practices in such
fields that might be transposed from other statistical techniques such as model
acceptance or rejection based on p-value screening might be poorly fitted for
SEMs especially when these models are used to confirm or reject hypotheses.
In this simulation study, SEMs were fitted via two commonly used R packages:
lavaan and piecewiseSEM. Five different data-generation scenarios were
explored: (i) random, (ii) exact, (iii) shuffled, (iv) underspecified and (v)
overspecified. In addition, sample size and model complexity were also varied
to explore their impact on various global and local model fitness indices.
The results showed that not one single model index should be used to decide
on model fitness but rather a combination of different model fitness indices is
needed. The global chi-square test for lavaan or the Fisher's C statistic for
piecewiseSEM were, in isolation, poor indicators of model fitness. In addition,
the simulations showed that to achieve sufficient power to detect individual
effects, adequate sample sizes are required. Finally, BIC showed good capacity
to select models closer to the truth especially for more complex models.
I provide, based on these results, a flowchart indicating how information
from different metrics may be combined to reveal model strength and weaknesses.
Researchers in scientific fields with little experience in SEMs, such as in
ecology, should consider and accept these limitations.
| stat |
Improving Estimation in Functional Linear Regression with Points of
Impact: Insights into Google AdWords | The functional linear regression model with points of impact is a recent
augmentation of the classical functional linear model with many practically
important applications. In this work, however, we demonstrate that the existing
data-driven procedure for estimating the parameters of this regression model
can be very instable and inaccurate. The tendency to omit relevant points of
impact is a particularly problematic aspect resulting in omitted-variable
biases. We explain the theoretical reason for this problem and propose a new
sequential estimation algorithm that leads to significantly improved estimation
results. Our estimation algorithm is compared with the existing estimation
procedure using an in-depth simulation study. The applicability is demonstrated
using data from Google AdWords, today's most important platform for online
advertisements. The \textsf{R}-package \texttt{FunRegPoI} and additional
\textsf{R}-codes are provided in the online supplementary material.
| stat |
Diagnostic tests can predict efficacy without randomization to
intervention and control by solving simultaneous equations based on different
test results: Required when assessing the effectiveness of test, trace and
isolation | The efficacy of an intervention can be assessed by randomising patients to
different diagnostic tests instead of to an intervention and control. This is
based on a familiar view that treatment will be more effective if given to
patients based on the results of appropriate diagnostic tests (i.e. the correct
diagnosis) than if it given to those based on the results of inappropriate
tests (i.e. the wrong diagnosis). The tests must have different characteristics
e.g. sensitivities with respect to the outcome. This principle is applied by
allocating an intervention to an individual if a test result is positive (or on
one side of a threshold) and allocating to a control to an individual if the
result is negative (or on the other side of the threshold). This can also be
done with different dichotomizing thresholds for one test. The frequencies of
the outcome in those with each of the four resulting observations are then used
to calculate the risk reduction by solving a pair of simultaneous equations.
This assumes that the risk reduction and the overall frequency of the outcome
is the same in both groups created by randomization. The calculations are
illustrated by using data from a randomized controlled trial that assessed the
efficacy of the angiotensin receptor blocker in lowering the risk of diabetic
nephropathy in patients conditional on different urinary albumin excretion
rates. They are also illustrated with simulated data based on a suggested
methodology for assessing the effectiveness of test, trace and isolation to
reduce transmission of the SARS-Cov-2 virus using RT-PCR and LFD tests. This
approach also allows the sensitivity and specificity of these tests with
respect of the outcome of viral spreading to be determined irrespective of the
efficacy of isolation.
| stat |
Modeling Network Populations via Graph Distances | This article introduces a new class of models for multiple networks. The core
idea is to parametrize a distribution on labelled graphs in terms of a
Fr\'{e}chet mean graph (which depends on a user-specified choice of metric or
graph distance) and a parameter that controls the concentration of this
distribution about its mean. Entropy is the natural parameter for such control,
varying from a point mass concentrated on the Fr\'{e}chet mean itself to a
uniform distribution over all graphs on a given vertex set. We provide a
hierarchical Bayesian approach for exploiting this construction, along with
straightforward strategies for sampling from the resultant posterior
distribution. We conclude by demonstrating the efficacy of our approach via
simulation studies and two multiple-network data analysis examples: one drawn
from systems biology and the other from neuroscience.
| stat |
Measurement Error Correction in Particle Tracking Microrheology | In diverse biological applications, particle tracking of passive microscopic
species has become the experimental measurement of choice -- when either the
materials are of limited volume, or so soft as to deform uncontrollably when
manipulated by traditional instruments. In a wide range of particle tracking
experiments, a ubiquitous finding is that the mean squared displacement (MSD)
of particle positions exhibits a power-law signature, the parameters of which
reveal valuable information about the viscous and elastic properties of various
biomaterials. However, MSD measurements are typically contaminated by complex
and interacting sources of instrumental noise. As these often affect the
high-frequency bandwidth to which MSD estimates are particularly sensitive,
inadequate error correction can lead to severe bias in power law estimation and
thereby, the inferred viscoelastic properties. In this article, we propose a
novel strategy to filter high-frequency noise from particle tracking
measurements. Our filters are shown theoretically to cover a broad spectrum of
high-frequency noises, and lead to a parametric estimator of MSD power-law
coefficients for which an efficient computational implementation is presented.
Based on numerous analyses of experimental and simulated data, results suggest
our methods perform very well compared to other denoising procedures.
| stat |
Statistical inference with anchored Bayesian mixture of regressions
models: A case study analysis of allometric data | We present a case study in which we use a mixture of regressions model to
improve on an ill-fitting simple linear regression model relating log brain
mass to log body mass for 100 placental mammalian species. The slope of this
regression model is of particular scientific interest because it corresponds to
a constant that governs a hypothesized allometric power law relating brain mass
to body mass. A specific line of investigation is to determine whether the
regression parameters vary across subgroups of related species.
We model these data using an anchored Bayesian mixture of regressions model,
which modifies the standard Bayesian Gaussian mixture by pre-assigning small
subsets of observations to given mixture components with probability one. These
observations (called anchor points) break the relabeling invariance typical of
exchangeable model specifications (the so-called label-switching problem). A
careful choice of which observations to pre-classify to which mixture
components is key to the specification of a well-fitting anchor model.
In the article we compare three strategies for the selection of anchor
points. The first assumes that the underlying mixture of regressions model
holds and assigns anchor points to different components to maximize the
information about their labeling. The second makes no assumption about the
relationship between x and y and instead identifies anchor points using a
bivariate Gaussian mixture model. The third strategy begins with the assumption
that there is only one mixture regression component and identifies anchor
points that are representative of a clustering structure based on case-deletion
importance sampling weights. We compare the performance of the three strategies
on the allometric data set and use auxiliary taxonomic information about the
species to evaluate the model-based classifications estimated from these
models.
| stat |
Invertible DenseNets with Concatenated LipSwish | We introduce Invertible Dense Networks (i-DenseNets), a more parameter
efficient alternative to Residual Flows. The method relies on an analysis of
the Lipschitz continuity of the concatenation in DenseNets, where we enforce
invertibility of the network by satisfying the Lipschitz constant. We extend
this method by proposing a learnable concatenation, which not only improves the
model performance but also indicates the importance of the concatenated
representation. Additionally, we introduce the Concatenated LipSwish as
activation function, for which we show how to enforce the Lipschitz condition
and which boosts performance. The new architecture, i-DenseNet, out-performs
Residual Flow and other flow-based models on density estimation evaluated in
bits per dimension, where we utilize an equal parameter budget. Moreover, we
show that the proposed model out-performs Residual Flows when trained as a
hybrid model where the model is both a generative and a discriminative model.
| stat |
Separating common and idiosyncratic components without moment condition
and on the weighted $L_1$ minimization path | In large-dimensional factor analysis, existing methods, such as principal
component analysis (PCA), assumed finite fourth moment of the idiosyncratic
components, in order to derive the convergence rates of the estimated factor
loadings and scores. However, in many areas, such as finance and
macroeconomics, many variables are heavy-tailed. In this case, PCA-based
estimators and their variations are not theoretically underpinned. In this
paper, we investigate into the weighted $L_1$ minimization on the factor
loadings and scores, which amounts to assuming a temporal and cross-sectional
quantile structure for panel observations instead of the mean pattern in $L_2$
minimization. Without any moment constraint on the idiosyncratic errors, we
correctly identify the common and idiosyncratic components for each variable.
We obtained the convergence rates of a computationally feasible weighted $L_1$
minimization estimators via iteratively alternating the quantile regression
cross-sectionally and serially. Bahardur representations for the estimated
factor loadings and scores are provided under some mild conditions. In
addition, a robust method is proposed to estimate the number of factors
consistently. Simulation experiments checked the validity of the theory. Our
analysis on a financial data set shows the superiority of the proposed method
over other state-of-the-art methods.
| stat |
Inferring medication adherence from time-varying health measures | Medication adherence is a problem of widespread concern in clinical care.
Poor adherence is a particular problem for patients with chronic diseases
requiring long-term medication because poor adherence can result in less
successful treatment outcomes and even preventable deaths. Existing methods to
collect information about patient adherence are resource-intensive or do not
successfully detect low-adherers with high accuracy. Acknowledging that health
measures recorded at clinic visits are more reliably recorded than a patient's
adherence, we have developed an approach to infer medication adherence rates
based on longitudinally recorded health measures that are likely impacted by
time-varying adherence behaviors. Our framework permits the inclusion of
baseline health characteristics and socio-demographic data. We employ a modular
inferential approach. First, we fit a two-component model on a training set of
patients who have detailed adherence data obtained from electronic medication
monitoring. One model component predicts adherence behaviors only from baseline
health and socio-demographic information, and the other predicts longitudinal
health measures given the adherence and baseline health measures. Posterior
draws of relevant model parameters are simulated from this model using Markov
chain Monte Carlo methods. Second, we develop an approach to infer medication
adherence from the time-varying health measures using a Sequential Monte Carlo
algorithm applied to a new set of patients for whom no adherence data are
available. We apply and evaluate the method on a cohort of hypertensive
patients, using baseline health comorbidities, socio-demographic measures, and
blood pressure measured over time to infer patients' adherence to
antihypertensive medication.
| stat |
A two-way factor model for high-dimensional matrix data | In this article, we introduce a two-way factor model for a high-dimensional
data matrix and study the properties of the maximum likelihood estimation
(MLE). The proposed model assumes separable effects of row and column
attributes and captures the correlation across rows and columns with
low-dimensional hidden factors. The model inherits the dimension-reduction
feature of classical factor models but introduces a new framework with
separable row and column factors, representing the covariance or correlation
structure in the data matrix. We propose a block alternating, maximizing
strategy to compute the MLE of factor loadings as well as other model
parameters. We discuss model identifiability, obtain consistency and the
asymptotic distribution for the MLE as the numbers of rows and columns in the
data matrix increase. One interesting phenomenon that we learned from our
analysis is that the variance of the estimates in the two-way factor model
depends on the distance of variances of row factors and column factors in a way
that is not expected in classical factor analysis. We further demonstrate the
performance of the proposed method through simulation and real data analysis.
| stat |
Differentially Private Federated Variational Inference | In many real-world applications of machine learning, data are distributed
across many clients and cannot leave the devices they are stored on.
Furthermore, each client's data, computational resources and communication
constraints may be very different. This setting is known as federated learning,
in which privacy is a key concern. Differential privacy is commonly used to
provide mathematical privacy guarantees. This work, to the best of our
knowledge, is the first to consider federated, differentially private, Bayesian
learning. We build on Partitioned Variational Inference (PVI) which was
recently developed to support approximate Bayesian inference in the federated
setting. We modify the client-side optimisation of PVI to provide an
(${\epsilon}$, ${\delta}$)-DP guarantee. We show that it is possible to learn
moderately private logistic regression models in the federated setting that
achieve similar performance to models trained non-privately on centralised
data.
| stat |
Enhanced Bayesian Model Updating with Incomplete Modal Information Using
Parallel, Interactive and Adaptive Markov Chains | Finite element model updating is challenging because 1) the problem is
oftentimes underdetermined while the measurements are limited and/or
incomplete; 2) many combinations of parameters may yield responses that are
similar with respect to actual measurements; and 3) uncertainties inevitably
exist. The aim of this research is to leverage upon computational intelligence
through statistical inference to facilitate an enhanced, probabilistic finite
element model updating using incomplete modal response measurement. This new
framework is built upon efficient inverse identification through optimization,
whereas Bayesian inference is employed to account for the effect of
uncertainties. To overcome the computational cost barrier, we adopt Markov
chain Monte Carlo (MCMC) to characterize the target function/distribution.
Instead of using single Markov chain in conventional Bayesian approach, we
develop a new sampling theory with multiple parallel, interactive and adaptive
Markov chains and incorporate into Bayesian inference. This can harness the
collective power of these Markov chains to realize the concurrent search of
multiple local optima. The number of required Markov chains and their
respective initial model parameters are automatically determined via Monte
Carlo simulation-based sample pre-screening followed by K-means clustering
analysis. These enhancements can effectively address the aforementioned
challenges in finite element model updating. The validity of this framework is
systematically demonstrated through case studies.
| stat |
Couplings of the Random-Walk Metropolis algorithm | Couplings play a central role in contemporary Markov chain Monte Carlo
methods and in the analysis of their convergence to stationarity. In most
cases, a coupling must induce relatively fast meeting between chains to ensure
good performance. In this paper we fix attention on the random walk Metropolis
algorithm and examine a range of coupling design choices. We introduce proposal
and acceptance step couplings based on geometric, optimal transport, and
maximality considerations. We consider the theoretical properties of these
choices and examine their implication for the meeting time of the chains. We
conclude by extracting a few general principles and hypotheses on the design of
effective couplings.
| stat |
Collective Spectral Density Estimation and Clustering for
Spatially-Correlated Data | In this paper, we develop a method for estimating and clustering
two-dimensional spectral density functions (2D-SDFs) for spatial data from
multiple subregions. We use a common set of adaptive basis functions to explain
the similarities among the 2D-SDFs in a low-dimensional space and estimate the
basis coefficients by maximizing the Whittle likelihood with two penalties. We
apply these penalties to impose the smoothness of the estimated 2D-SDFs and the
spatial dependence of the spatially-correlated subregions. The proposed
technique provides a score matrix, that is comprised of the estimated
coefficients associated with the common set of basis functions representing the
2D-SDFs. {Instead of clustering the estimated SDFs directly, we propose to
employ the score matrix for clustering purposes, taking advantage of its
low-dimensional property.} In a simulation study, we demonstrate that our
proposed method outperforms other competing estimation procedures used for
clustering. Finally, to validate the described clustering method, we apply the
procedure to soil moisture data from the Mississippi basin to produce
homogeneous spatial clusters. We produce animations to dynamically show the
estimation procedure, including the estimated 2D-SDFs and the score matrix,
which provide an intuitive illustration of the proposed method.
| stat |
Statistical and Computational Trade-Offs in Kernel K-Means | We investigate the efficiency of k-means in terms of both statistical and
computational requirements. More precisely, we study a Nystr\"om approach to
kernel k-means. We analyze the statistical properties of the proposed method
and show that it achieves the same accuracy of exact kernel k-means with only a
fraction of computations. Indeed, we prove under basic assumptions that
sampling $\sqrt{n}$ Nystr\"om landmarks allows to greatly reduce computational
costs without incurring in any loss of accuracy. To the best of our knowledge
this is the first result of this kind for unsupervised learning.
| stat |
Weight-of-evidence 2.0 with shrinkage and spline-binning | In many practical applications, such as fraud detection, credit risk modeling
or medical decision making, classification models for assigning instances to a
predefined set of classes are required to be both precise as well as
interpretable. Linear modeling methods such as logistic regression are often
adopted, since they offer an acceptable balance between precision and
interpretability. Linear methods, however, are not well equipped to handle
categorical predictors with high-cardinality or to exploit non-linear relations
in the data. As a solution, data preprocessing methods such as
weight-of-evidence are typically used for transforming the predictors. The
binning procedure that underlies the weight-of-evidence approach, however, has
been little researched and typically relies on ad-hoc or expert driven
procedures. The objective in this paper, therefore, is to propose a formalized,
data-driven and powerful method.
To this end, we explore the discretization of continuous variables through
the binning of spline functions, which allows for capturing non-linear effects
in the predictor variables and yields highly interpretable predictors taking
only a small number of discrete values. Moreover, we extend upon the
weight-of-evidence approach and propose to estimate the proportions using
shrinkage estimators. Together, this offers an improved ability to exploit both
non-linear and categorical predictors for achieving increased classification
precision, while maintaining interpretability of the resulting model and
decreasing the risk of overfitting.
We present the results of a series of experiments in a fraud detection
setting, which illustrate the effectiveness of the presented approach. We
facilitate reproduction of the presented results and adoption of the proposed
approaches by providing both the dataset and the code for implementing the
experiments and the presented approach.
| stat |
A new asymmetric $\epsilon$-insensitive pinball loss function based
support vector quantile regression model | In this paper, we propose a novel asymmetric $\epsilon$-insensitive pinball
loss function for quantile estimation. There exists some pinball loss functions
which attempt to incorporate the $\epsilon$-insensitive zone approach in it
but, they fail to extend the $\epsilon$-insensitive approach for quantile
estimation in true sense. The proposed asymmetric $\epsilon$-insensitive
pinball loss function can make an asymmetric $\epsilon$- insensitive zone of
fixed width around the data and divide it using $\tau$ value for the estimation
of the $\tau$th quantile. The use of the proposed asymmetric
$\epsilon$-insensitive pinball loss function in Support Vector Quantile
Regression (SVQR) model improves its prediction ability significantly. It also
brings the sparsity back in SVQR model. Further, the numerical results obtained
by several experiments carried on artificial and real world datasets
empirically show the efficacy of the proposed `$\epsilon$-Support Vector
Quantile Regression' ($\epsilon$-SVQR) model over other existing SVQR models.
| stat |
Uncertainty quantification using martingales for misspecified Gaussian
processes | We address uncertainty quantification for Gaussian processes (GPs) under
misspecified priors, with an eye towards Bayesian Optimization (BO). GPs are
widely used in BO because they easily enable exploration based on posterior
uncertainty bands. However, this convenience comes at the cost of robustness: a
typical function encountered in practice is unlikely to have been drawn from
the data scientist's prior, in which case uncertainty estimates can be
misleading, and the resulting exploration can be suboptimal. We present a
frequentist approach to GP/BO uncertainty quantification. We utilize the GP
framework as a working model, but do not assume correctness of the prior. We
instead construct a confidence sequence (CS) for the unknown function using
martingale techniques. There is a necessary cost to achieving robustness: if
the prior was correct, posterior GP bands are narrower than our CS.
Nevertheless, when the prior is wrong, our CS is statistically valid and
empirically outperforms standard GP methods, in terms of both coverage and
utility for BO. Additionally, we demonstrate that powered likelihoods provide
robustness against model misspecification.
| stat |
Predicting Clinical Outcomes in Glioblastoma: An Application of
Topological and Functional Data Analysis | Glioblastoma multiforme (GBM) is an aggressive form of human brain cancer
that is under active study in the field of cancer biology. Its rapid
progression and the relative time cost of obtaining molecular data make other
readily-available forms of data, such as images, an important resource for
actionable measures in patients. Our goal is to utilize information given by
medical images taken from GBM patients in statistical settings. To do this, we
design a novel statistic---the smooth Euler characteristic transform
(SECT)---that quantifies magnetic resonance images (MRIs) of tumors. Due to its
well-defined inner product structure, the SECT can be used in a wider range of
functional and nonparametric modeling approaches than other previously proposed
topological summary statistics. When applied to a cohort of GBM patients, we
find that the SECT is a better predictor of clinical outcomes than both
existing tumor shape quantifications and common molecular assays. Specifically,
we demonstrate that SECT features alone explain more of the variance in GBM
patient survival than gene expression, volumetric features, and morphometric
features. The main takeaways from our findings are thus twofold. First, they
suggest that images contain valuable information that can play an important
role in clinical prognosis and other medical decisions. Second, they show that
the SECT is a viable tool for the broader study of medical imaging informatics.
| stat |
Sparse Variational Inference: Bayesian Coresets from Scratch | The proliferation of automated inference algorithms in Bayesian statistics
has provided practitioners newfound access to fast, reproducible data analysis
and powerful statistical models. Designing automated methods that are also both
computationally scalable and theoretically sound, however, remains a
significant challenge. Recent work on Bayesian coresets takes the approach of
compressing the dataset before running a standard inference algorithm,
providing both scalability and guarantees on posterior approximation error. But
the automation of past coreset methods is limited because they depend on the
availability of a reasonable coarse posterior approximation, which is difficult
to specify in practice. In the present work we remove this requirement by
formulating coreset construction as sparsity-constrained variational inference
within an exponential family. This perspective leads to a novel construction
via greedy optimization, and also provides a unifying information-geometric
view of present and past methods. The proposed Riemannian coreset construction
algorithm is fully automated, requiring no problem-specific inputs aside from
the probabilistic model and dataset. In addition to being significantly easier
to use than past methods, experiments demonstrate that past coreset
constructions are fundamentally limited by the fixed coarse posterior
approximation; in contrast, the proposed algorithm is able to continually
improve the coreset, providing state-of-the-art Bayesian dataset summarization
with orders-of-magnitude reduction in KL divergence to the exact posterior.
| stat |
Tutorial on logistic-regression calibration and fusion: Converting a
score to a likelihood ratio | Logistic-regression calibration and fusion are potential steps in the
calculation of forensic likelihood ratios. The present paper provides a
tutorial on logistic-regression calibration and fusion at a practical
conceptual level with minimal mathematical complexity. A score is
log-likelihood-ratio like in that it indicates the degree of similarity of a
pair of samples while taking into consideration their typicality with respect
to a model of the relevant population. A higher-valued score provides more
support for the same-origin hypothesis over the different-origin hypothesis
than does a lower-valued score; however, the absolute values of scores are not
interpretable as log likelihood ratios. Logistic-regression calibration is a
procedure for converting scores to log likelihood ratios, and
logistic-regression fusion is a procedure for converting parallel sets of
scores from multiple forensic-comparison systems to log likelihood ratios.
Logistic-regression calibration and fusion were developed for automatic speaker
recognition and are popular in forensic voice comparison. They can also be
applied in other branches of forensic science, a fingerprint/fingermark example
is provided.
| stat |
Tree-based Node Aggregation in Sparse Graphical Models | High-dimensional graphical models are often estimated using regularization
that is aimed at reducing the number of edges in a network. In this work, we
show how even simpler networks can be produced by aggregating the nodes of the
graphical model. We develop a new convex regularized method, called the
tree-aggregated graphical lasso or tag-lasso, that estimates graphical models
that are both edge-sparse and node-aggregated. The aggregation is performed in
a data-driven fashion by leveraging side information in the form of a tree that
encodes node similarity and facilitates the interpretation of the resulting
aggregated nodes. We provide an efficient implementation of the tag-lasso by
using the locally adaptive alternating direction method of multipliers and
illustrate our proposal's practical advantages in simulation and in
applications in finance and biology.
| stat |
A Bayesian Spatial Modeling Approach to Mortality Forecasting | This paper extends Bayesian mortality projection models for multiple
populations considering the stochastic structure and the effect of spatial
autocorrelation among the observations. We explain high levels of
overdispersion according to adjacent locations based on the conditional
autoregressive model. In an empirical study, we compare different hierarchical
projection models for the analysis of geographical diversity in mortality
between the Japanese counties in multiple years, according to age. By a Markov
chain Monte Carlo (MCMC) computation, results have demonstrated the flexibility
and predictive performance of our proposed model.
| stat |
Observational nonidentifiability, generalized likelihood and free energy | We study the parameter estimation problem in mixture models with
observational nonidentifiability: the full model (also containing hidden
variables) is identifiable, but the marginal (observed) model is not. Hence
global maxima of the marginal likelihood are (infinitely) degenerate and
predictions of the marginal likelihood are not unique. We show how to
generalize the marginal likelihood by introducing an effective temperature, and
making it similar to the free energy. This generalization resolves the
observational nonidentifiability, since its maximization leads to unique
results that are better than a random selection of one degenerate maximum of
the marginal likelihood or the averaging over many such maxima. The generalized
likelihood inherits many features from the usual likelihood, e.g. it holds the
conditionality principle, and its local maximum can be searched for via
suitably modified expectation-maximization method. The maximization of the
generalized likelihood relates to entropy optimization.
| stat |
Lasso for hierarchical polynomial models | In a polynomial regression model, the divisibility conditions implicit in
polynomial hierarchy give way to a natural construction of constraints for the
model parameters. We use this principle to derive versions of strong and weak
hierarchy and to extend existing work in the literature, which at the moment is
only concerned with models of degree two. We discuss how to estimate parameters
in lasso using standard quadratic programming techniques and apply our proposal
to both simulated data and examples from the literature. The proposed
methodology compares favorably with existing techniques in terms of low
validation error and model size.
| stat |
Copula-based measures of asymmetry between the lower and upper tail
probabilities | We propose a copula-based measure of asymmetry between the lower and upper
tail probabilities of bivariate distributions. The proposed measure has a
simple form and possesses some desirable properties as a measure of asymmetry.
The limit of the proposed measure as the index goes to the boundary of its
domain can be expressed in a simple form under certain conditions on copulas. A
sample analogue of the proposed measure for a sample from a copula is presented
and its weak convergence to a Gaussian process is shown. Another sample
analogue of the presented measure, which is based on a sample from a
distribution on $\mathbb{R}^2$, is given. Simple methods for interval
estimation and nonparametric testing based on the two sample analogues are
presented. As an example, the presented measure is applied to daily returns of
S&P500 and Nikkei225.
| stat |
Framework for the Treatment And Reporting of Missing data in
Observational Studies: The TARMOS framework | Missing data are ubiquitous in medical research. Although there is increasing
guidance on how to handle missing data, practice is changing slowly and
misapprehensions abound, particularly in observational research. We present a
practical framework for handling and reporting the analysis of incomplete data
in observational studies, which we illustrate using a case study from the Avon
Longitudinal Study of Parents and Children. The framework consists of three
steps: 1) Develop an analysis plan specifying the analysis model and how
missing data are going to be addressed. An important consideration is whether a
complete records analysis is likely to be valid, whether multiple imputation or
an alternative approach is likely to offer benefits, and whether a sensitivity
analysis regarding the missingness mechanism is required. 2) Explore the data,
checking the methods outlined in the analysis plan are appropriate, and conduct
the pre-planned analysis. 3) Report the results, including a description of the
missing data, details on how the missing data were addressed, and the results
from all analyses, interpreted in light of the missing data and the clinical
relevance. This framework seeks to support researchers in thinking
systematically about missing data, and transparently reporting the potential
effect on the study results.
| stat |
Categorical Co-Frequency Analysis: Clustering Diagnosis Codes to Predict
Hospital Readmissions | Accurately predicting patients' risk of 30-day hospital readmission would
enable hospitals to efficiently allocate resource-intensive interventions. We
develop a new method, Categorical Co-Frequency Analysis (CoFA), for clustering
diagnosis codes from the International Classification of Diseases (ICD)
according to the similarity in relationships between covariates and readmission
risk. CoFA measures the similarity between diagnoses by the frequency with
which two diagnoses are split in the same direction versus split apart in
random forests to predict readmission risk. Applying CoFA to de-identified data
from Berkshire Medical Center, we identified three groups of diagnoses that
vary in readmission risk. To evaluate CoFA, we compared readmission risk models
using ICD majors and CoFA groups to a baseline model without diagnosis
variables. We found substituting ICD majors for the CoFA-identified clusters
simplified the model without compromising the accuracy of predictions. Fitting
separate models for each ICD major and CoFA group did not improve predictions,
suggesting that readmission risk may be more homogeneous that heterogeneous
across diagnosis groups.
| stat |
RSVP-graphs: Fast High-dimensional Covariance Matrix Estimation under
Latent Confounding | In this work we consider the problem of estimating a high-dimensional $p
\times p$ covariance matrix $\Sigma$, given $n$ observations of confounded data
with covariance $\Sigma + \Gamma \Gamma^T$, where $\Gamma$ is an unknown $p
\times q$ matrix of latent factor loadings. We propose a simple and scalable
estimator based on the projection on to the right singular vectors of the
observed data matrix, which we call RSVP. Our theoretical analysis of this
method reveals that in contrast to PCA-based approaches, RSVP is able to cope
well with settings where the smallest eigenvalue of $\Gamma^T \Gamma$ is close
to the largest eigenvalue of $\Sigma$, as well as settings where the
eigenvalues of $\Gamma^T \Gamma$ are diverging fast. It is also able to handle
data that may have heavy tails and only requires that the data has an
elliptical distribution. RSVP does not require knowledge or estimation of the
number of latent factors $q$, but only recovers $\Sigma$ up to an unknown
positive scale factor. We argue this suffices in many applications, for example
if an estimate of the correlation matrix is desired. We also show that by using
subsampling, we can further improve the performance of the method. We
demonstrate the favourable performance of RSVP through simulation experiments
and an analysis of gene expression datasets collated by the GTEX consortium.
| stat |
The True Sample Complexity of Identifying Good Arms | We consider two multi-armed bandit problems with $n$ arms: (i) given an
$\epsilon > 0$, identify an arm with mean that is within $\epsilon$ of the
largest mean and (ii) given a threshold $\mu_0$ and integer $k$, identify $k$
arms with means larger than $\mu_0$. Existing lower bounds and algorithms for
the PAC framework suggest that both of these problems require $\Omega(n)$
samples. However, we argue that these definitions not only conflict with how
these algorithms are used in practice, but also that these results disagree
with intuition that says (i) requires only $\Theta(\frac{n}{m})$ samples where
$m = |\{ i : \mu_i > \max_{i \in [n]} \mu_i - \epsilon\}|$ and (ii) requires
$\Theta(\frac{n}{m}k)$ samples where $m = |\{ i : \mu_i > \mu_0 \}|$. We
provide definitions that formalize these intuitions, obtain lower bounds that
match the above sample complexities, and develop explicit, practical algorithms
that achieve nearly matching upper bounds.
| stat |
Testing parametric models in linear-directional regression | This paper presents a goodness-of-fit test for parametric regression models
with scalar response and directional predictor, that is, a vector on a sphere
of arbitrary dimension. The testing procedure is based on the weighted squared
distance between a smooth and a parametric regression estimator, where the
smooth regression estimator is obtained by a projected local approach.
Asymptotic behavior of the test statistic under the null hypothesis and local
alternatives is provided, jointly with a consistent bootstrap algorithm for
application in practice. A simulation study illustrates the performance of the
test in finite samples. The procedure is applied to test a linear model in text
mining.
| stat |
RecurJac: An Efficient Recursive Algorithm for Bounding Jacobian Matrix
of Neural Networks and Its Applications | The Jacobian matrix (or the gradient for single-output networks) is directly
related to many important properties of neural networks, such as the function
landscape, stationary points, (local) Lipschitz constants and robustness to
adversarial attacks. In this paper, we propose a recursive algorithm, RecurJac,
to compute both upper and lower bounds for each element in the Jacobian matrix
of a neural network with respect to network's input, and the network can
contain a wide range of activation functions. As a byproduct, we can
efficiently obtain a (local) Lipschitz constant, which plays a crucial role in
neural network robustness verification, as well as the training stability of
GANs. Experiments show that (local) Lipschitz constants produced by our method
is of better quality than previous approaches, thus providing better robustness
verification results. Our algorithm has polynomial time complexity, and its
computation time is reasonable even for relatively large networks.
Additionally, we use our bounds of Jacobian matrix to characterize the
landscape of the neural network, for example, to determine whether there exist
stationary points in a local neighborhood. Source code available at
\url{http://github.com/huanzhang12/RecurJac-Jacobian-bounds}.
| stat |
Point and interval estimation of the target dose using weighted
regression modelling | In a Phase II dose-finding study with a placebo control, a new drug with
several dose levels is compared with a placebo to test for the effectiveness of
the new drug. The main focus of such studies often lies in the characterization
of the dose-response relationship followed by the estimation of a target dose
that leads to a clinically relevant effect over the placebo. This target dose
is known as the minimum effective dose (MED) in a drug development study.
Several approaches exist that combine multiple comparison procedures with
modeling techniques to efficiently estimate the dose-response model and
thereafter select the target dose. Despite the flexibility of the existing
approaches, they cannot completely address the model uncertainty in the
model-selection step and may lead to target dose estimates that are biased. In
this article, we propose two new MED estimation approaches based on weighted
regression modeling that are robust against deviations from the dose-response
model assumptions. These approaches are compared with existing approaches with
regard to their accuracy in point and interval estimation of the MED. We
illustrate by a simulation study that by integrating one of the new dose
estimation approaches with the existing dose-response profile estimation
approaches one can take into account the uncertainty of the model selection
step.
| stat |
Adaptive, Distribution-Free Prediction Intervals for Deep Networks | The machine learning literature contains several constructions for prediction
intervals that are intuitively reasonable but ultimately ad-hoc in that they do
not come with provable performance guarantees. We present methods from the
statistics literature that can be used efficiently with neural networks under
minimal assumptions with guaranteed performance. We propose a neural network
that outputs three values instead of a single point estimate and optimizes a
loss function motivated by the standard quantile regression loss. We provide
two prediction interval methods with finite sample coverage guarantees solely
under the assumption that the observations are independent and identically
distributed. The first method leverages the conformal inference framework and
provides average coverage. The second method provides a new, stronger guarantee
by conditioning on the observed data. Lastly, our loss function does not
compromise the predictive accuracy of the network like other prediction
interval methods. We demonstrate the ease of use of our procedures as well as
its improvements over other methods on both simulated and real data. As most
deep networks can easily be modified by our method to output predictions with
valid prediction intervals, its use should become standard practice, much like
reporting standard errors along with mean estimates.
| stat |
Stable predictions for health related anticausal prediction tasks
affected by selection biases: the need to deconfound the test set features | In health related machine learning applications, the training data often
corresponds to a non-representative sample from the target populations where
the learners will be deployed. In anticausal prediction tasks, selection biases
often make the associations between confounders and the outcome variable
unstable across different target environments. As a consequence, the
predictions from confounded learners are often unstable, and might fail to
generalize in shifted test environments. Stable prediction approaches aim to
solve this problem by producing predictions that are stable across unknown test
environments. These approaches, however, are sometimes applied to the training
data alone with the hope that training an unconfounded model will be enough to
generate stable predictions in shifted test sets. Here, we show that this is
insufficient, and that improved stability can be achieved by deconfounding the
test set features as well. We illustrate these observations using both
synthetic data and real world data from a mobile health study.
| stat |
Impulse Response Analysis for Sparse High-Dimensional Time Series | We consider structural impulse response analysis for sparse high-dimensional
vector autoregressive (VAR) systems. Since standard procedures like the
delta-method do not lead to valid inference in the high-dimensional set-up, we
propose an alternative approach. First, we directly construct a de-sparsified
version of the regularized estimators of the moving average parameters that are
associated with the VAR process. Second, the obtained estimators are combined
with a de-sparsified estimator of the contemporaneous impact matrix in order to
estimate the structural impulse response coefficients of interest. We show that
the resulting estimator of the impulse response coefficients has a Gaussian
limiting distribution. Valid inference is then implemented using an appropriate
bootstrap approach. Our inference procedure is illustrated by means of
simulations and real data applications.
| stat |
Analytic Evaluation of the Fractional Moments for the Quasi-Stationary
Distribution of the Shiryaev Martingale on an Interval | We consider the quasi-stationary distribution of the classical Shiryaev
diffusion restricted to the interval $[0,A]$ with absorption at a fixed $A>0$.
We derive analytically a closed-form formula for the distribution's fractional
moment of an {\em arbitrary} given order $s\in\mathbb{R}$; the formula is
consistent with that previously found by Polunchenko and Pepelyshev (2018) for
the case of $s\in\mathbb{N}$. We also show by virtue of the formula that, if
$s<1$, then the $s$-th fractional moment of the quasi-stationary distribution
becomes that of the exponential distribution (with mean $1/2$) in the limit as
$A\to+\infty$; the limiting exponential distribution is the stationary
distribution of the reciprocal of the Shiryaev diffusion.
| stat |
Structure learning for CTBN's via penalized maximum likelihood methods | The continuous-time Bayesian networks (CTBNs) represent a class of stochastic
processes, which can be used to model complex phenomena, for instance, they can
describe interactions occurring in living processes, in social science models
or in medicine. The literature on this topic is usually focused on the case
when the dependence structure of a system is known and we are to determine
conditional transition intensities (parameters of the network). In the paper,
we study the structure learning problem, which is a more challenging task and
the existing research on this topic is limited. The approach, which we propose,
is based on a penalized likelihood method. We prove that our algorithm, under
mild regularity conditions, recognizes the dependence structure of the graph
with high probability. We also investigate the properties of the procedure in
numerical studies to demonstrate its effectiveness.
| stat |
Connections between statistical practice in elementary particle physics
and the severity concept as discussed in Mayo's Statistical Inference as
Severe Testing | For many years, philosopher-of-statistics Deborah Mayo has been advocating
the concept of severe testing as a key part of hypothesis testing. Her recent
book, Statistical Inference as Severe Testing, is a comprehensive exposition of
her arguments in the context of a historical study of many threads of
statistical inference, both frequentist and Bayesian. Her foundational point of
view is called error statistics, emphasizing frequentist evaluation of the
errors called Type I and Type II in the Neyman-Pearson theory of frequentist
hypothesis testing. Since the field of elementary particle physics (also known
as high energy physics) has strong traditions in frequentist inference, one
might expect that something like the severity concept was independently
developed in the field. Indeed, I find that, at least operationally
(numerically), we high-energy physicists have long interpreted data in ways
that map directly onto severity. Whether or not we subscribe to Mayo's
philosophical interpretations of severity is a more complicated story that I do
not address here.
| stat |
Investigating some attributes of periodicity in DNA sequences via
semi-Markov modelling | DNA segments and sequences have been studied thoroughly during the past
decades. One of the main problems in computational biology is the
identification of exon-intron structures inside genes using mathematical
techniques. Previous studies have used different methods, such as Fourier
analysis and hidden-Markov models, in order to be able to predict which parts
of a gene correspond to a protein encoding area. In this paper, a semi-Markov
model is applied to 3-base periodic sequences, which characterize the
protein-coding regions of the gene. Analytic forms of the related probabilities
and the corresponding indexes are provided, which yield a description of the
underlying periodic pattern. Last, the previous theoretical results are
illustrated with DNA sequences of synthetic and real data.
| stat |
Robust parametric modeling of Alzheimer's disease progression | Quantitative characterization of disease progression using longitudinal data
can provide long-term predictions for the pathological stages of individuals.
This work studies the robust modeling of Alzheimer's disease progression using
parametric methods. The proposed method linearly maps the individual's age to a
disease progression score (DPS) and jointly fits constrained generalized
logistic functions to the longitudinal dynamics of biomarkers as functions of
the DPS using M-estimation. Robustness of the estimates is quantified using
bootstrapping via Monte Carlo resampling, and the estimated inflection points
of the fitted functions are used to temporally order the modeled biomarkers in
the disease course. Kernel density estimation is applied to the obtained DPSs
for clinical status classification using a Bayesian classifier. Different
M-estimators and logistic functions, including a novel type proposed in this
study, called modified Stannard, are evaluated on the data from the Alzheimer's
Disease Neuroimaging Initiative (ADNI) for robust modeling of volumetric MRI
and PET biomarkers, CSF measurements, as well as cognitive tests. The results
show that the modified Stannard function fitted using the logistic loss
achieves the best modeling performance with an average normalized MAE of 0.991
across all biomarkers and bootstraps. Applied to the ADNI test set, this model
achieves a multiclass AUC of 0.934 in clinical status classification. The
obtained results for the proposed model outperform almost all state-of-the-art
results in predicting biomarker values and classifying clinical status.
Finally, the experiments show that the proposed model, trained using abundant
ADNI data, generalizes well to data from the National Alzheimer's Coordinating
Center (NACC) with an average normalized MAE of 1.182 and a multiclass AUC of
0.929.
| stat |
Improved Design of Quadratic Discriminant Analysis Classifier in
Unbalanced Settings | The use of quadratic discriminant analysis (QDA) or its regularized version
(R-QDA) for classification is often not recommended, due to its
well-acknowledged high sensitivity to the estimation noise of the covariance
matrix. This becomes all the more the case in unbalanced data settings for
which it has been found that R-QDA becomes equivalent to the classifier that
assigns all observations to the same class. In this paper, we propose an
improved R-QDA that is based on the use of two regularization parameters and a
modified bias, properly chosen to avoid inappropriate behaviors of R-QDA in
unbalanced settings and to ensure the best possible classification performance.
The design of the proposed classifier builds on a refined asymptotic analysis
of its performance when the number of samples and that of features grow large
simultaneously, which allows to cope efficiently with the high-dimensionality
frequently met within the big data paradigm. The performance of the proposed
classifier is assessed on both real and synthetic data sets and was shown to be
much better than what one would expect from a traditional R-QDA.
| stat |
bayes4psy -- an Open Source R Package for Bayesian Statistics in
Psychology | Research in psychology generates interesting data sets and unique statistical
modelling tasks. However, these tasks, while important, are often very
specific, so appropriate statistical models and methods cannot be found in
accessible Bayesian tools. As a result, the use of Bayesian methods is limited
to those that have the technical and statistical fundamentals that are required
for probabilistic programming. Such knowledge is not part of the typical
psychology curriculum and is a difficult obstacle for psychology students and
researchers to overcome. The goal of the bayes4psy package is to bridge this
gap and offer a collection of models and methods to be used for data analysis
that arises from psychology experiments and as a teaching tool for Bayesian
statistics in psychology. The package contains Bayesian t-test and
bootstrapping and models for analyzing reaction times, success rates, and
colors. It also provides all the diagnostic, analytic and visualization tools
for the modern Bayesian data analysis workflow.
| stat |
Score predictor factor analysis: Reproducing observed covariances by
means of factor score predictors | The model implied by factor score predictors does not reproduce the
non-diagonal elements of the observed covariance matrix as well as the factor
loadings. It is therefore investigated whether it is possible to estimate
factor loadings for which the model implied by the factor score predictors
optimally reproduces the non-diagonal elements of the observed covariance
matrix. Accordingly, loading estimates are proposed for which the model implied
by the factor score predictors allows for a least-squares approximation of the
non-diagonal elements of the observed covariance matrix. This estimation method
is termed Score predictor factor analysis and algebraically compared with
Minres factor analysis as well as principal component analysis. A population
based and a sample based simulation study was performed in order to compare
Score predictor factor analysis, Minres factor analysis, and principal
component analysis. It turns out that the non-diagonal elements of the observed
covariance matrix can more exactly be reproduced from the factor score
predictors computed from Score predictor factor analysis than from the factor
score predictors computed from Minres factor analysis and from principal
components. Moreover, Score predictor factor analysis can be helpful to
identify factors when the factor model does not perfectly fit to the data
because of model error.
| stat |
White-box vs Black-box: Bayes Optimal Strategies for Membership
Inference | Membership inference determines, given a sample and trained parameters of a
machine learning model, whether the sample was part of the training set. In
this paper, we derive the optimal strategy for membership inference with a few
assumptions on the distribution of the parameters. We show that optimal attacks
only depend on the loss function, and thus black-box attacks are as good as
white-box attacks. As the optimal strategy is not tractable, we provide
approximations of it leading to several inference methods, and show that
existing membership inference methods are coarser approximations of this
optimal strategy. Our membership attacks outperform the state of the art in
various settings, ranging from a simple logistic regression to more complex
architectures and datasets, such as ResNet-101 and Imagenet.
| stat |
Thermostat-assisted continuously-tempered Hamiltonian Monte Carlo for
Bayesian learning | We propose a new sampling method, the thermostat-assisted
continuously-tempered Hamiltonian Monte Carlo, for Bayesian learning on large
datasets and multimodal distributions. It simulates the Nos\'e-Hoover dynamics
of a continuously-tempered Hamiltonian system built on the distribution of
interest. A significant advantage of this method is that it is not only able to
efficiently draw representative i.i.d. samples when the distribution contains
multiple isolated modes, but capable of adaptively neutralising the noise
arising from mini-batches and maintaining accurate sampling. While the
properties of this method have been studied using synthetic distributions,
experiments on three real datasets also demonstrated the gain of performance
over several strong baselines with various types of neural networks plunged in.
| stat |
Predicting with High Correlation Features | It has been shown that instead of learning actual object features, deep
networks tend to exploit non-robust (spurious) discriminative features that are
shared between training and test sets. Therefore, while they achieve state of
the art performance on such test sets, they achieve poor generalization on out
of distribution (OOD) samples where the IID (independent, identical
distribution) assumption breaks and the distribution of non-robust features
shifts. In this paper, we consider distribution shift as a shift in the
distribution of input features during test time that exhibit low correlation
with targets in the training set. Under this definition, we evaluate existing
robust feature learning methods and regularization methods and compare them
against a baseline designed to specifically capture high correlation features
in training set. As a controlled test-bed, we design a colored MNIST (C-MNIST)
dataset and find that existing methods trained on this set fail to generalize
well on an OOD version this dataset, showing that they overfit the low
correlation color features. This is avoided by the baseline method trained on
the same C-MNIST data, which is designed to learn high correlation features,
and is able to generalize on the test sets of vanilla MNIST, MNIST-M and SVHN
datasets. Our code is available at
\url{https://github.com/salesforce/corr_based_prediction}.
| stat |
The effect of stay-at-home orders on COVID-19 cases and fatalities in
the United States | Governments issue "stay at home" orders to reduce the spread of contagious
diseases, but the magnitude of such orders' effectiveness is uncertain. In the
United States these orders were not coordinated at the national level during
the coronavirus disease 2019 (COVID-19) pandemic, which creates an opportunity
to use spatial and temporal variation to measure the policies' effect with
greater accuracy. Here, we combine data on the timing of stay-at-home orders
with daily confirmed COVID-19 cases and fatalities at the county level in the
United States. We estimate the effect of stay-at-home orders using a
difference-in-differences design that accounts for unmeasured local variation
in factors like health systems and demographics and for unmeasured temporal
variation in factors like national mitigation actions and access to tests.
Compared to counties that did not implement stay-at-home orders, the results
show that the orders are associated with a 30.2 percent (11.0 to 45.2)
reduction in weekly cases after one week, a 40.0 percent (23.4 to 53.0)
reduction after two weeks, and a 48.6 percent (31.1 to 61.7) reduction after
three weeks. Stay-at-home orders are also associated with a 59.8 percent (18.3
to 80.2) reduction in weekly fatalities after three weeks. These results
suggest that stay-at-home orders reduced confirmed cases by 390,000 (170,000 to
680,000) and fatalities by 41,000 (27,000 to 59,000) within the first three
weeks in localities where they were implemented.
| stat |
Process Control with Highly Left Censored Data | The need to monitor industrial processes, detecting changes in process
parameters in order to promptly correct problems that may arise, generates a
particular area of interest. This is particularly critical and complex when the
measured value falls below the sensitivity limits of the measuring system or
below detection limits, causing much of their observations are incomplete. Such
observations to be called incomplete observations or left censored data. With a
high level of censorship, for example greater than 70%, the application of
traditional methods for monitoring processes is not appropriate. It is required
to use appropriate data analysis statistical techniques, to assess the actual
state of the process at any time. This paper proposes a way to estimate process
parameters in such cases and presents the corresponding control chart, from an
algorithm that is also presented.
| stat |
Spatial-Temporal-Textual Point Processes with Applications in Crime
Linkage Detection | Crimes emerge out of complex interactions of human behaviors and situations.
Linkages between crime events are highly complex. Detecting crime linkage given
a set of events is a highly challenging task since we only have limited
information, including text descriptions, event times, and locations; moreover,
the link among incidents is subtle. In practice, there are very few labeled
data about related cases. We present a statistical modeling framework for
police reports based on multivariate marked spatio-temporal Hawkes processes.
Inspired by the notion of modus operandi (M.O.) in crime analysis, we perform
text embedding for police reports and treat embedding vectors as marks of the
Hawkes process. The embedding is performed by regularized Restricted Boltzmann
Machine (RBM) to promote the probabilistic sparsity of selected keywords.
Numerical results using real data demonstrate the competitive performance and
interpretability of our method. The proposed method can be used in other
similar data in social networks, electronic health records.
| stat |
Flexible Bayesian Dynamic Modeling of Correlation and Covariance
Matrices | Modeling correlation (and covariance) matrices can be challenging due to the
positive-definiteness constraint and potential high-dimensionality. Our
approach is to decompose the covariance matrix into the correlation and
variance matrices and propose a novel Bayesian framework based on modeling the
correlations as products of unit vectors. By specifying a wide range of
distributions on a sphere (e.g. the squared-Dirichlet distribution), the
proposed approach induces flexible prior distributions for covariance matrices
(that go beyond the commonly used inverse-Wishart prior). For modeling
real-life spatio-temporal processes with complex dependence structures, we
extend our method to dynamic cases and introduce unit-vector Gaussian process
priors in order to capture the evolution of correlation among components of a
multivariate time series. To handle the intractability of the resulting
posterior, we introduce the adaptive $\Delta$-Spherical Hamiltonian Monte
Carlo. We demonstrate the validity and flexibility of our proposed framework in
a simulation study of periodic processes and an analysis of rat's local field
potential activity in a complex sequence memory task.
| stat |
Deep and interpretable regression models for ordinal outcomes | Outcomes with a natural order commonly occur in prediction tasks and often
the available input data are a mixture of complex data like images and tabular
predictors. Deep Learning (DL) models are state-of-the-art for image
classification tasks but frequently treat ordinal outcomes as unordered and
lack interpretability. In contrast, classical ordinal regression models
consider the outcome's order and yield interpretable predictor effects but are
limited to tabular data. We present ordinal neural network transformation
models (ONTRAMs), which unite DL with classical ordinal regression approaches.
ONTRAMs are a special case of transformation models and trade off flexibility
and interpretability by additively decomposing the transformation function into
terms for image and tabular data using jointly trained neural networks. The
performance of the most flexible ONTRAM is by definition equivalent to a
standard multi-class DL model trained with cross-entropy while being faster in
training when facing ordinal outcomes. Lastly, we discuss how to interpret
model components for both tabular and image data on two publicly available
datasets.
| stat |
Bayesian Analysis of Nonparanormal Graphical Models Using
Rank-Likelihood | Gaussian graphical models, where it is assumed that the variables of interest
jointly follow a multivariate normal distribution with a sparse precision
matrix, have been used to study intrinsic dependence among variables, but the
normality assumption may be restrictive in many settings. A nonparanormal
graphical model is a semiparametric generalization of a Gaussian graphical
model for continuous variables where it is assumed that the variables follow a
Gaussian graphical model only after some unknown smooth monotone
transformation. We consider a Bayesian approach for the nonparanormal graphical
model using a rank-likelihood which remains invariant under monotone
transformations, thereby avoiding the need to put a prior on the transformation
functions. On the underlying precision matrix of the transformed variables, we
consider a horseshoe prior on its Cholesky decomposition and use an efficient
posterior Gibbs sampling scheme. We present a posterior consistency result for
the precision matrix based on the rank-based likelihood. We study the numerical
performance of the proposed method through a simulation study and apply it on a
real dataset.
| stat |
Generalised Bayes Updates with $f$-divergences through Probabilistic
Classifiers | A stream of algorithmic advances has steadily increased the popularity of the
Bayesian approach as an inference paradigm, both from the theoretical and
applied perspective. Even with apparent successes in numerous application
fields, a rising concern is the robustness of Bayesian inference in the
presence of model misspecification, which may lead to undesirable extreme
behavior of the posterior distributions for large sample sizes. Generalized
belief updating with a loss function represents a central principle to making
Bayesian inference more robust and less vulnerable to deviations from the
assumed model. Here we consider such updates with $f$-divergences to quantify a
discrepancy between the assumed statistical model and the probability
distribution which generated the observed data. Since the latter is generally
unknown, estimation of the divergence may be viewed as an intractable problem.
We show that the divergence becomes accessible through the use of probabilistic
classifiers that can leverage an estimate of the ratio of two probability
distributions even when one or both of them is unknown. We demonstrate the
behavior of generalized belief updates for various specific choices under the
$f$-divergence family. We show that for specific divergence functions such an
approach can even improve on methods evaluating the correct model likelihood
function analytically.
| stat |
Semiparametric estimation for space-time max-stable processes: F
-madogram-based estimation approach | Max-stable processes have been expanded to quantify extremal dependence in
spatio-temporal data. Due to the interaction between space and time,
spatio-temporal data are often complex to analyze. So, characterizing these
dependencies is one of the crucial challenges in this field of statistics. This
paper suggests a semiparametric inference methodology based on the
spatio-temporal F-madogram for estimating the parameters of a space-time
max-stable process using gridded data. The performance of the method is
investigated through various simulation studies. Finally, we apply our
inferential procedure to quantify the extremal behavior of radar rainfall data
in a region in the State of Florida.
| stat |
A spatial causal analysis of wildland fire-contributed PM2.5 using
numerical model output | Wildland fire smoke contains hazardous levels of fine particulate matter
PM2.5, a pollutant shown to adversely effect health. Estimating fire
attributable PM2.5 concentrations is key to quantifying the impact on air
quality and subsequent health burden. This is a challenging problem since only
total PM2.5 is measured at monitoring stations and both fire-attributable PM2.5
and PM2.5 from all other sources are correlated in space and time. We propose a
framework for estimating fire-contributed PM2.5 and PM2.5 from all other
sources using a novel causal inference framework and bias-adjusted chemical
model representations of PM2.5 under counterfactual scenarios. The chemical
model representation of PM2.5 for this analysis is simulated using Community
Multi-Scale Air Quality Modeling System (CMAQ), run with and without fire
emissions across the contiguous U.S. for the 2008-2012 wildfire seasons. The
CMAQ output is calibrated with observations from monitoring sites for the same
spatial domain and time period. We use a Bayesian model that accounts for
spatial variation to estimate the effect of wildland fires on PM2.5 and state
assumptions under which the estimate has a valid causal interpretation. Our
results include estimates of absolute, relative and cumulative contributions of
wildfire smoke to PM2.5 for the contiguous U.S. Additionally, we compute the
health burden associated with the PM2.5 attributable to wildfire smoke.
| stat |
Neural Conditional Event Time Models | Event time models predict occurrence times of an event of interest based on
known features. Recent work has demonstrated that neural networks achieve
state-of-the-art event time predictions in a variety of settings. However,
standard event time models suppose that the event occurs, eventually, in all
cases. Consequently, no distinction is made between a) the probability of event
occurrence, and b) the predicted time of occurrence. This distinction is
critical when predicting medical diagnoses, equipment defects, social media
posts, and other events that or may not occur, and for which the features
affecting a) may be different from those affecting b). In this work, we develop
a conditional event time model that distinguishes between these components,
implement it as a neural network with a binary stochastic layer representing
finite event occurrence, and show how it may be learned from right-censored
event times via maximum likelihood estimation. Results demonstrate superior
event occurrence and event time predictions on synthetic data, medical events
(MIMIC-III), and social media posts (Reddit), comprising 21 total prediction
tasks.
| stat |
Efficient Estimation of COM-Poisson Regression and Generalized Additive
Model | The Conway-Maxwell-Poisson (CMP) or COM-Poison regression is a popular model
for count data due to its ability to capture both under dispersion and over
dispersion. However, CMP regression is limited when dealing with complex
nonlinear relationships. With today's wide availability of count data,
especially due to the growing collection of data on human and social behavior,
there is need for count data models that can capture complex nonlinear
relationships. One useful approach is additive models; but, there has been no
additive model implementation for the CMP distribution. To fill this void, we
first propose a flexible estimation framework for CMP regression based on
iterative reweighed least squares (IRLS) and then extend this model to allow
for additive components using a penalized splines approach. Because the CMP
distribution belongs to the exponential family, convergence of IRLS is
guaranteed under some regularity conditions. Further, it is also known that
IRLS provides smaller standard errors compared to gradient-based methods. We
illustrate the usefulness of this approach through extensive simulation studies
and using real data from a bike sharing system in Washington, DC.
| stat |
algcomparison: Comparing the Performance of Graphical Structure Learning
Algorithms with TETRAD | In this report we describe a tool for comparing the performance of graphical
causal structure learning algorithms implemented in the TETRAD freeware suite
of causal analysis methods. Currently the tool is available as package in the
TETRAD source code (written in Java). Simulations can be done varying the
number of runs, sample sizes, and data modalities. Performance on this
simulated data can then be compared for a number of algorithms, with parameters
varied and with performance statistics as selected, producing a publishable
report. The package presented here may also be used to compare structure
learning methods across platforms and programming languages, i.e., to compare
algorithms implemented in TETRAD with those implemented in MATLAB, Python, or
R.
| stat |