title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
A robust approach to quantifying uncertainty in matching problems of
causal inference | Unquantified sources of uncertainty in observational causal analyses can
break the integrity of the results. One would never want another analyst to
repeat a calculation with the same dataset, using a seemingly identical
procedure, only to find a different conclusion. However, as we show in this
work, there is a typical source of uncertainty that is essentially never
considered in observational causal studies: the choice of match assignment for
matched groups, that is, which unit is matched to which other unit before a
hypothesis test is conducted. The choice of match assignment is anything but
innocuous, and can have a surprisingly large influence on the causal
conclusions. Given that a vast number of causal inference studies test
hypotheses on treatment effects after treatment cases are matched with similar
control cases, we should find a way to quantify how much this extra source of
uncertainty impacts results. What we would really like to be able to report is
that \emph{no matter} which match assignment is made, as long as the match is
sufficiently good, then the hypothesis test result still holds. In this paper,
we provide methodology based on discrete optimization to create robust tests
that explicitly account for this possibility. We formulate robust tests for
binary and continuous data based on common test statistics as integer linear
programs solvable with common methodologies. We study the finite-sample
behavior of our test statistic in the discrete-data case. We apply our methods
to simulated and real-world datasets and show that they can produce useful
results in practical applied settings.
| stat |
State Space Emulation and Annealed Sequential Monte Carlo for High
Dimensional Optimization | Many high dimensional optimization problems can be reformulated into a
problem of finding theoptimal state path under an equivalent state space model
setting. In this article, we present a general emulation strategy for
developing a state space model whose likelihood function (or posterior
distribution) shares the same general landscape as the objective function of
the original optimization problem. Then the solution of the optimization
problem is the same as the optimal state path that maximizes the likelihood
function or the posterior distribution under the emulated system. To find such
an optimal path, we adapt a simulated annealing approach by inserting a
temperature control into the emulated dynamic system and propose a novel
annealed Sequential Monte Carlo (SMC) method that effectively generating Monte
Carlo sample paths utilizing samples obtained on the high temperature scale.
Compared to the vanilla simulated annealing implementation, annealed SMC is an
iterative algorithm for state space model optimization that directly generates
state paths from the equilibrium distributions with a decreasing sequence of
temperatures through sequential importance sampling which does not require
burn-in or mixing iterations to ensure quasi-equilibrium condition. Several
applications of state space model emulation and the corresponding annealed SMC
results are demonstrated.
| stat |
Graphite: Iterative Generative Modeling of Graphs | Graphs are a fundamental abstraction for modeling relational data. However,
graphs are discrete and combinatorial in nature, and learning representations
suitable for machine learning tasks poses statistical and computational
challenges. In this work, we propose Graphite, an algorithmic framework for
unsupervised learning of representations over nodes in large graphs using deep
latent variable generative models. Our model parameterizes variational
autoencoders (VAE) with graph neural networks, and uses a novel iterative graph
refinement strategy inspired by low-rank approximations for decoding. On a wide
variety of synthetic and benchmark datasets, Graphite outperforms competing
approaches for the tasks of density estimation, link prediction, and node
classification. Finally, we derive a theoretical connection between message
passing in graph neural networks and mean-field variational inference.
| stat |
Structure learning via unstructured kernel-based M-regression | In statistical learning, identifying underlying structures of true target
functions based on observed data plays a crucial role to facilitate subsequent
modeling and analysis. Unlike most of those existing methods that focus on some
specific settings under certain model assumptions, this paper proposes a
general and novel framework for recovering true structures of target functions
by using unstructured M-regression in a reproducing kernel Hilbert space
(RKHS). The proposed framework is inspired by the fact that gradient functions
can be employed as a valid tool to learn underlying structures, including
sparse learning, interaction selection and model identification, and it is easy
to implement by taking advantage of the nice properties of the RKHS. More
importantly, it admits a wide range of loss functions, and thus includes many
commonly used methods, such as mean regression, quantile regression,
likelihood-based classification, and margin-based classification, which is also
computationally efficient by solving convex optimization tasks. The asymptotic
results of the proposed framework are established within a rich family of loss
functions without any explicit model specifications. The superior performance
of the proposed framework is also demonstrated by a variety of simulated
examples and a real case study.
| stat |
Multifidelity probability estimation via fusion of estimators | This paper develops a multifidelity method that enables estimation of failure
probabilities for expensive-to-evaluate models via information fusion and
importance sampling. The presented general fusion method combines multiple
probability estimators with the goal of variance reduction. We use low-fidelity
models to derive biasing densities for importance sampling and then fuse the
importance sampling estimators such that the fused multifidelity estimator is
unbiased and has mean-squared error lower than or equal to that of any of the
importance sampling estimators alone. By fusing all available estimators, the
method circumvents the challenging problem of selecting the best biasing
density and using only that density for sampling. A rigorous analysis shows
that the fused estimator is optimal in the sense that it has minimal variance
amongst all possible combinations of the estimators. The asymptotic behavior of
the proposed method is demonstrated on a convection-diffusion-reaction partial
differential equation model for which $10^5$ samples can be afforded. To
illustrate the proposed method at scale, we consider a model of a free plane
jet and quantify how uncertainties at the flow inlet propagate to a quantity of
interest related to turbulent mixing. Compared to an importance sampling
estimator that uses the high-fidelity model alone, our multifidelity estimator
reduces the required CPU time by 65\% while achieving a similar coefficient of
variation.
| stat |
Analysis of KNN Information Estimators for Smooth Distributions | KSG mutual information estimator, which is based on the distances of each
sample to its k-th nearest neighbor, is widely used to estimate mutual
information between two continuous random variables. Existing work has analyzed
the convergence rate of this estimator for random variables whose densities are
bounded away from zero in its support. In practice, however, KSG estimator also
performs well for a much broader class of distributions, including not only
those with bounded support and densities bounded away from zero, but also those
with bounded support but densities approaching zero, and those with unbounded
support. In this paper, we analyze the convergence rate of the error of KSG
estimator for smooth distributions, whose support of density can be both
bounded and unbounded. As KSG mutual information estimator can be viewed as an
adaptive recombination of KL entropy estimators, in our analysis, we also
provide convergence analysis of KL entropy estimator for a broad class of
distributions.
| stat |
Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design | We introduce Deep Adaptive Design (DAD), a general method for amortizing the
cost of performing sequential adaptive experiments using the framework of
Bayesian optimal experimental design (BOED). Traditional sequential BOED
approaches require substantial computational time at each stage of the
experiment. This makes them unsuitable for most real-world applications, where
decisions must typically be made quickly. DAD addresses this restriction by
learning an amortized design network upfront and then using this to rapidly run
(multiple) adaptive experiments at deployment time. This network takes as input
the data from previous steps, and outputs the next design using a single
forward pass; these design decisions can be made in milliseconds during the
live experiment. To train the network, we introduce contrastive information
bounds that are suitable objectives for the sequential setting, and propose a
customized network architecture that exploits key symmetries. We demonstrate
that DAD successfully amortizes the process of experimental design,
outperforming alternative strategies on a number of problems.
| stat |
Adaptive use of replicated Latin Hypercube Designs for computing Sobol'
sensitivity indices | As recently pointed out in the field of Global Sensitivity Analysis (GSA) of
computer simulations, the use of replicated Latin Hypercube Designs (rLHDs) is
a cost-saving alternative to regular Monte Carlo sampling to estimate
first-order Sobol' indices. Indeed, two rLHDs are sufficient to compute the
whole set of those indices regardless of the number of input variables. This
relies on a permutation trick which, however, only works within the class of
estimators called Oracle 2. In the present paper, we show that rLHDs are still
beneficial to another class of estimators, called Oracle 1, which often
outperforms Oracle 2 for estimating small and moderate indices. Even though
unlike Oracle 2 the computation cost of Oracle 1 depends on the input
dimension, the permutation trick can be applied to construct an averaged
(triple) Oracle 1 estimator whose great accuracy is presented on a numerical
example.
Thus, we promote an adaptive rLHDs-based Sobol' sensitivity analysis where
the first stage is to compute the whole set of first-order indices by Oracle 2.
If needed, the accuracy of small and moderate indices can then be reevaluated
by the averaged Oracle 1 estimators. This strategy, cost-saving and
guaranteeing the accuracy of estimates, is applied to a computer model from the
nuclear field.
| stat |
Impacts of California Proposition 47 on Crime in Santa Monica, CA | We examine crime patterns in Santa Monica, California before and after
passage of Proposition 47, a 2014 initiative that reclassified some non-violent
felonies to misdemeanors. We also study how the 2016 opening of four new light
rail stations, and how more community-based policing starting in late 2018,
impacted crime. A series of statistical analyses are performed on reclassified
(larceny, fraud, possession of narcotics, forgery, receiving/possessing stolen
property) and non-reclassified crimes by probing publicly available databases
from 2006 to 2019. We compare data before and after passage of Proposition 47,
city-wide and within eight neighborhoods. Similar analyses are conducted within
a 450 meter radius of the new transit stations. Reports of monthly reclassified
crimes increased city-wide by approximately 15% after enactment of Proposition
47, with a significant drop observed in late 2018. Downtown exhibited the
largest overall surge. The reported incidence of larceny intensified throughout
the city. Two new train stations, including Downtown, reported significant
crime increases in their vicinity after service began. While the number of
reported reclassified crimes increased after passage of Proposition 47, those
not affected by the new law decreased or stayed constant, suggesting that
Proposition 47 strongly impacted crime in Santa Monica. Reported crimes
decreased in late 2018 concurrent with the adoption of new policing measures
that enhanced outreach and patrolling. These findings may be relevant to law
enforcement and policy-makers. Follow-up studies needed to confirm long-term
trends may be affected by the COVID-19 pandemic that drastically changed
societal conditions.
| stat |
Deep Fourier Kernel for Self-Attentive Point Processes | We present a novel attention-based model for discrete event data to capture
complex non-linear temporal dependence structures. We borrow the idea from the
attention mechanism and incorporate it into the point processes' conditional
intensity function. We further introduce a novel score function using Fourier
kernel embedding, whose spectrum is represented using neural networks, which
drastically differs from the traditional dot-product kernel and can capture a
more complex similarity structure. We establish our approach's theoretical
properties and demonstrate our approach's competitive performance compared to
the state-of-the-art for synthetic and real data.
| stat |
Sparse Gaussian Process Based On Hat Basis Functions | Gaussian process is one of the most popular non-parametric Bayesian
methodologies for modeling the regression problem. It is completely determined
by its mean and covariance functions. And its linear property makes it
relatively straightforward to solve the prediction problem. Although Gaussian
process has been successfully applied in many fields, it is still not enough to
deal with physical systems that satisfy inequality constraints. This issue has
been addressed by the so-called constrained Gaussian process in recent years.
In this paper, we extend the core ideas of constrained Gaussian process.
According to the range of training or test data, we redefine the hat basis
functions mentioned in the constrained Gaussian process. Based on hat basis
functions, we propose a new sparse Gaussian process method to solve the
unconstrained regression problem. Similar to the exact Gaussian process and
Gaussian process with Fully Independent Training Conditional approximation, our
method obtains satisfactory approximate results on open-source datasets or
analytical functions. In terms of performance, the proposed method reduces the
overall computational complexity from $O(n^{3})$ computation in exact Gaussian
process to $O(nm^{2})$ with $m$ hat basis functions and $n$ training data
points.
| stat |
Multilevel Emulation for Stochastic Computer Models with Application to
Large Offshore Wind farms | Large renewable energy projects, such as large offshore wind farms, are
critical to achieving low-emission targets set by governments. Stochastic
computer models allow us to explore future scenarios to aid decision making
whilst considering the most relevant uncertainties. Complex stochastic computer
models can be prohibitively slow and thus an emulator may be constructed and
deployed to allow for efficient computation. We present a novel heteroscedastic
Gaussian Process emulator which exploits cheap approximations to a stochastic
offshore wind farm simulator. We conduct a probabilistic sensitivity analysis
to understand the influence of key parameters in the wind farm simulator which
will help us to plan a probability elicitation in the future.
| stat |
Robust inference for nonlinear regression models from the Tsallis score:
application to Covid-19 contagion in Italy | We discuss an approach for fitting robust nonlinear regression models, which
can be employed to model and predict the contagion dynamics of the Covid-19 in
Italy. The focus is on the analysis of epidemic data using robust dose-response
curves, but the functionality is applicable to arbitrary nonlinear regression
models.
| stat |
Dynamic Sasvi: Strong Safe Screening for Norm-Regularized Least Squares | A recently introduced technique for a sparse optimization problem called
"safe screening" allows us to identify irrelevant variables in the early stage
of optimization. In this paper, we first propose a flexible framework for safe
screening based on the Fenchel-Rockafellar duality and then derive a strong
safe screening rule for norm-regularized least squares by the framework. We
call the proposed screening rule for norm-regularized least squares "dynamic
Sasvi" because it can be interpreted as a generalization of Sasvi. Unlike the
original Sasvi, it does not require the exact solution of a more strongly
regularized problem; hence, it works safely in practice. We show that our
screening rule can eliminate more features and increase the speed of the solver
in comparison with other screening rules both theoretically and experimentally.
| stat |
Explaining predictive models with mixed features using Shapley values
and conditional inference trees | It is becoming increasingly important to explain complex, black-box machine
learning models. Although there is an expanding literature on this topic,
Shapley values stand out as a sound method to explain predictions from any type
of machine learning model. The original development of Shapley values for
prediction explanation relied on the assumption that the features being
described were independent. This methodology was then extended to explain
dependent features with an underlying continuous distribution. In this paper,
we propose a method to explain mixed (i.e. continuous, discrete, ordinal, and
categorical) dependent features by modeling the dependence structure of the
features using conditional inference trees. We demonstrate our proposed method
against the current industry standards in various simulation studies and find
that our method often outperforms the other approaches. Finally, we apply our
method to a real financial data set used in the 2018 FICO Explainable Machine
Learning Challenge and show how our explanations compare to the FICO challenge
Recognition Award winning team.
| stat |
Entropic Causal Inference: Identifiability and Finite Sample Results | Entropic causal inference is a framework for inferring the causal direction
between two categorical variables from observational data. The central
assumption is that the amount of unobserved randomness in the system is not too
large. This unobserved randomness is measured by the entropy of the exogenous
variable in the underlying structural causal model, which governs the causal
relation between the observed variables. Kocaoglu et al. conjectured that the
causal direction is identifiable when the entropy of the exogenous variable is
not too large. In this paper, we prove a variant of their conjecture. Namely,
we show that for almost all causal models where the exogenous variable has
entropy that does not scale with the number of states of the observed
variables, the causal direction is identifiable from observational data. We
also consider the minimum entropy coupling-based algorithmic approach presented
by Kocaoglu et al., and for the first time demonstrate algorithmic
identifiability guarantees using a finite number of samples. We conduct
extensive experiments to evaluate the robustness of the method to relaxing some
of the assumptions in our theory and demonstrate that both the constant-entropy
exogenous variable and the no latent confounder assumptions can be relaxed in
practice. We also empirically characterize the number of observational samples
needed for causal identification. Finally, we apply the algorithm on Tuebingen
cause-effect pairs dataset.
| stat |
Resampling Methods for Detecting Anisotropic Correlation Structure | This paper proposes parametric and non-parametric hypothesis testing
algorithms for detecting anisotropy -- rotational variance of the covariance
function in random fields. Both algorithms are based on resampling mechanisms,
which enable avoiding relying on asymptotic assumptions, as is common in
previous algorithms. The algorithms' performance is illustrated numerically in
simulation experiments and on real datasets representing a variety of potential
challenges.
| stat |
Diagnostic Uncertainty Calibration: Towards Reliable Machine Predictions
in Medical Domain | We propose an evaluation framework for class probability estimates (CPEs) in
the presence of label uncertainty, which is commonly observed as diagnosis
disagreement between experts in the medical domain. We also formalize
evaluation metrics for higher-order statistics, including inter-rater
disagreement, to assess predictions on label uncertainty. Moreover, we propose
a novel post-hoc method called $alpha$-calibration, that equips neural network
classifiers with calibrated distributions over CPEs. Using synthetic
experiments and a large-scale medical imaging application, we show that our
approach significantly enhances the reliability of uncertainty estimates:
disagreement probabilities and posterior CPEs.
| stat |
Clustering Binary Data by Application of Combinatorial Optimization
Heuristics | We study clustering methods for binary data, first defining aggregation
criteria that measure the compactness of clusters. Five new and original
methods are introduced, using neighborhoods and population behavior
combinatorial optimization metaheuristics: first ones are simulated annealing,
threshold accepting and tabu search, and the others are a genetic algorithm and
ant colony optimization. The methods are implemented, performing the proper
calibration of parameters in the case of heuristics, to ensure good results.
From a set of 16 data tables generated by a quasi-Monte Carlo experiment, a
comparison is performed for one of the aggregations using L1 dissimilarity,
with hierarchical clustering, and a version of k-means: partitioning around
medoids or PAM. Simulated annealing perform very well, especially compared to
classical methods.
| stat |
High-Dimensional Bayesian Optimization via Tree-Structured Additive
Models | Bayesian Optimization (BO) has shown significant success in tackling
expensive low-dimensional black-box optimization problems. Many optimization
problems of interest are high-dimensional, and scaling BO to such settings
remains an important challenge. In this paper, we consider generalized additive
models in which low-dimensional functions with overlapping subsets of variables
are composed to model a high-dimensional target function. Our goal is to lower
the computational resources required and facilitate faster model learning by
reducing the model complexity while retaining the sample-efficiency of existing
methods. Specifically, we constrain the underlying dependency graphs to tree
structures in order to facilitate both the structure learning and optimization
of the acquisition function. For the former, we propose a hybrid graph learning
algorithm based on Gibbs sampling and mutation. In addition, we propose a novel
zooming-based algorithm that permits generalized additive models to be employed
more efficiently in the case of continuous domains. We demonstrate and discuss
the efficacy of our approach via a range of experiments on synthetic functions
and real-world datasets.
| stat |
A Framework for Sample Efficient Interval Estimation with Control
Variates | We consider the problem of estimating confidence intervals for the mean of a
random variable, where the goal is to produce the smallest possible interval
for a given number of samples. While minimax optimal algorithms are known for
this problem in the general case, improved performance is possible under
additional assumptions. In particular, we design an estimation algorithm to
take advantage of side information in the form of a control variate, leveraging
order statistics. Under certain conditions on the quality of the control
variates, we show improved asymptotic efficiency compared to existing
estimation algorithms. Empirically, we demonstrate superior performance on
several real world surveying and estimation tasks where we use the output of
regression models as the control variates.
| stat |
Adaptive County Level COVID-19 Forecast Models: Analysis and Improvement | Accurately forecasting county level COVID-19 confirmed cases is crucial to
optimizing medical resources. Forecasting emerging outbreaks pose a particular
challenge because many existing forecasting techniques learn from historical
seasons trends. Recurrent neural networks (RNNs) with LSTM-based cells are a
logical choice of model due to their ability to learn temporal dynamics. In
this paper, we adapt the state and county level influenza model, TDEFSI-LONLY,
proposed in Wang et a. [l2020] to national and county level COVID-19 data. We
show that this model poorly forecasts the current pandemic. We analyze the two
week ahead forecasting capabilities of the TDEFSI-LONLY model with combinations
of regularization techniques. Effective training of the TDEFSI-LONLY model
requires data augmentation, to overcome this challenge we utilize an SEIR model
and present an inter-county mixing extension to this model to simulate
sufficient training data. Further, we propose an alternate forecast model, {\it
County Level Epidemiological Inference Recurrent Network} (\alg{}) that trains
an LSTM backbone on national confirmed cases to learn a low dimensional time
pattern and utilizes a time distributed dense layer to learn individual county
confirmed case changes each day for a two weeks forecast. We show that the
best, worst, and median state forecasts made using CLEIR-Net model are
respectively New York, South Carolina, and Montana.
| stat |
Multiple imputation using dimension reduction techniques for
high-dimensional data | Missing data present challenges in data analysis. Naive analyses such as
complete-case and available-case analysis may introduce bias and loss of
efficiency, and produce unreliable results. Multiple imputation (MI) is one of
the most widely used methods for handling missing data which can be partly
attributed to its ease of use. However, existing MI methods implemented in most
statistical software are not applicable to or do not perform well in
high-dimensional settings where the number of predictors is large relative to
the sample size. To remedy this issue, we develop an MI approach that uses
dimension reduction techniques. Specifically, in constructing imputation models
in the presence of high-dimensional data our approach uses sure independent
screening followed by either sparse principal component analysis (sPCA) or
sufficient dimension reduction (SDR) techniques. Our simulation studies,
conducted for high-dimensional data, demonstrate that using SIS followed by
sPCA to perform MI achieves better performance than the other imputation
methods including several existing imputation approaches. We apply our approach
to analysis of gene expression data from a prostate cancer study.
| stat |
Data segmentation algorithms: Univariate mean change and beyond | Data segmentation a.k.a. multiple change point analysis has received
considerable attention due to its importance in time series analysis and signal
processing, with applications in a variety of fields including natural and
social sciences, medicine, engineering and finance.
In the first part of this survey, we review the existing literature on the
canonical data segmentation problem which aims at detecting and localising
multiple change points in the mean of univariate time series. We provide an
overview of popular methodologies on their computational complexity and
theoretical properties. In particular, our theoretical discussion focuses on
the separation rate relating to which change points are detectable by a given
procedure, and the localisation rate quantifying the precision of corresponding
change point estimators, and we distinguish between whether a homogeneous or
multiscale viewpoint has been adopted in their derivation. We further highlight
that the latter viewpoint provides the most general setting for investigating
the optimality of data segmentation algorithms.
Arguably, the canonical segmentation problem has been the most popular
framework to propose new data segmentation algorithms and study their
efficiency in the last decades. In the second part of this survey, we motivate
the importance of attaining an in-depth understanding of strengths and
weaknesses of methodologies for the change point problem in a simpler,
univariate setting, as a stepping stone for the development of methodologies
for more complex problems. We illustrate this with a range of examples
showcasing the connections between complex distributional changes and those in
the mean. We also discuss extensions towards high-dimensional change point
problems where we demonstrate that the challenges arising from high
dimensionality are orthogonal to those in dealing with multiple change points.
| stat |
On Lower Bounds for Standard and Robust Gaussian Process Bandit
Optimization | In this paper, we consider algorithm-independent lower bounds for the problem
of black-box optimization of functions having a bounded norm is some
Reproducing Kernel Hilbert Space (RKHS), which can be viewed as a non-Bayesian
Gaussian process bandit problem. In the standard noisy setting, we provide a
novel proof technique for deriving lower bounds on the regret, with benefits
including simplicity, versatility, and an improved dependence on the error
probability. In a robust setting in which every sampled point may be perturbed
by a suitably-constrained adversary, we provide a novel lower bound for
deterministic strategies, demonstrating an inevitable joint dependence of the
cumulative regret on the corruption level and the time horizon, in contrast
with existing lower bounds that only characterize the individual dependencies.
Furthermore, in a distinct robust setting in which the final point is perturbed
by an adversary, we strengthen an existing lower bound that only holds for
target success probabilities very close to one, by allowing for arbitrary
success probabilities in $(0,1)$.
| stat |
Sampling Permutations for Shapley Value Estimation | Game-theoretic attribution techniques based on Shapley values are used
extensively to interpret black-box machine learning models, but their exact
calculation is generally NP-hard, requiring approximation methods for
non-trivial models. As the computation of Shapley values can be expressed as a
summation over a set of permutations, a common approach is to sample a subset
of these permutations for approximation. Unfortunately, standard Monte Carlo
sampling methods can exhibit slow convergence, and more sophisticated quasi
Monte Carlo methods are not well defined on the space of permutations. To
address this, we investigate new approaches based on two classes of
approximation methods and compare them empirically. First, we demonstrate
quadrature techniques in a RKHS containing functions of permutations, using the
Mallows kernel to obtain explicit convergence rates of $O(1/n)$, improving on
$O(1/\sqrt{n})$ for plain Monte Carlo. The RKHS perspective also leads to quasi
Monte Carlo type error bounds, with a tractable discrepancy measure defined on
permutations. Second, we exploit connections between the hypersphere
$\mathbb{S}^{d-2}$ and permutations to create practical algorithms for
generating permutation samples with good properties. Experiments show the above
techniques provide significant improvements for Shapley value estimates over
existing methods, converging to a smaller RMSE in the same number of model
evaluations.
| stat |
A Stochastic Large-scale Machine Learning Algorithm for Distributed
Features and Observations | As the size of modern data sets exceeds the disk and memory capacities of a
single computer, machine learning practitioners have resorted to parallel and
distributed computing. Given that optimization is one of the pillars of machine
learning and predictive modeling, distributed optimization methods have
recently garnered ample attention, in particular when either observations or
features are distributed, but not both. We propose a general stochastic
algorithm where observations, features, and gradient components can be sampled
in a double distributed setting, i.e., with both features and observations
distributed. Very technical analyses establish convergence properties of the
algorithm under different conditions on the learning rate (diminishing to zero
or constant). Computational experiments in Spark demonstrate a superior
performance of our algorithm versus a benchmark in early iterations of the
algorithm, which is due to the stochastic components of the algorithm.
| stat |
General Identification of Dynamic Treatment Regimes Under Interference | In many applied fields, researchers are often interested in tailoring
treatments to unit-level characteristics in order to optimize an outcome of
interest. Methods for identifying and estimating treatment policies are the
subject of the dynamic treatment regime literature. Separately, in many
settings the assumption that data are independent and identically distributed
does not hold due to inter-subject dependence. The phenomenon where a subject's
outcome is dependent on his neighbor's exposure is known as interference. These
areas intersect in myriad real-world settings. In this paper we consider the
problem of identifying optimal treatment policies in the presence of
interference. Using a general representation of interference, via
Lauritzen-Wermuth-Freydenburg chain graphs (Lauritzen and Richardson, 2002), we
formalize a variety of policy interventions under interference and extend
existing identification theory (Tian, 2008; Sherman and Shpitser, 2018).
Finally, we illustrate the efficacy of policy maximization under interference
in a simulation study.
| stat |
High-dimensional Gaussian graphical model for network-linked data | Graphical models are commonly used to represent conditional dependence
relationships between variables. There are multiple methods available for
exploring them from high-dimensional data, but almost all of them rely on the
assumption that the observations are independent and identically distributed.
At the same time, observations connected by a network are becoming increasingly
common, and tend to violate these assumptions. Here we develop a Gaussian
graphical model for observations connected by a network with potentially
different mean vectors, varying smoothly over the network. We propose an
efficient estimation algorithm and demonstrate its effectiveness on both
simulated and real data, obtaining meaningful and interpretable results on a
statistics coauthorship network. We also prove that our method estimates both
the inverse covariance matrix and the corresponding graph structure correctly
under the assumption of network âcohesionâ, which refers to the empirically
observed phenomenon of network neighbors sharing similar traits.
| stat |
Efficient Bayesian generalized linear models with time-varying
coefficients: The walker package in R | The R package walker extends standard Bayesian general linear models to the
case where the effects of the explanatory variables can vary in time. This
allows, for example, to model the effects of interventions such as changes in
tax policy which gradually increases their effect over time. The Markov chain
Monte Carlo algorithms powering the Bayesian inference are based on Hamiltonian
Monte Carlo provided by Stan software, using a state space representation of
the model to marginalise over the regression coefficients for efficient
low-dimensional sampling.
| stat |
An introduction to computational complexity in Markov Chain Monte Carlo
methods | The aim of this work is to give an introduction to the theoretical background
and computational complexity of Markov chain Monte Carlo methods. Most of the
mathematical results related to the convergence are not found in most of the
statistical references, and computational complexity is still an open question
for most of the MCMC methods. In this work, we provide a general overview,
references, and discussion about all these theoretical subjects.
| stat |
Filaments of crime: Informing policing via thresholded ridge estimation | Objectives: We introduce a new method for reducing crime in hot spots and
across cities through ridge estimation. In doing so, our goal is to explore the
application of density ridges to hot spots and patrol optimization, and to
contribute to the policing literature in police patrolling and crime reduction
strategies.
Methods: We make use of the subspace-constrained mean shift algorithm, a
recently introduced approach for ridge estimation further developed in
cosmology, which we modify and extend for geospatial datasets and hot spot
analysis. Our experiments extract density ridges of Part I crime incidents from
the City of Chicago during the year 2018 and early 2019 to demonstrate the
application to current data.
Results: Our results demonstrate nonlinear mode-following ridges in agreement
with broader kernel density estimates. Using early 2019 incidents with
predictive ridges extracted from 2018 data, we create multi-run confidence
intervals and show that our patrol templates cover around 94% of incidents for
0.1-mile envelopes around ridges, quickly rising to near-complete coverage. We
also develop and provide researchers, as well as practitioners, with a
user-friendly and open-source software for fast geospatial density ridge
estimation.
Conclusions: We show that ridges following crime report densities can be used
to enhance patrolling capabilities. Our empirical tests show the stability of
ridges based on past data, offering an accessible way of identifying routes
within hot spots instead of patrolling epicenters. We suggest further research
into the application and efficacy of density ridges for patrolling.
| stat |
Machine Discovery of Partial Differential Equations from Spatiotemporal
Data | The study presents a general framework for discovering underlying Partial
Differential Equations (PDEs) using measured spatiotemporal data. The method,
called Sparse Spatiotemporal System Discovery ($\text{S}^3\text{d}$), decides
which physical terms are necessary and which can be removed (because they are
physically negligible in the sense that they do not affect the dynamics too
much) from a pool of candidate functions. The method is built on the recent
development of Sparse Bayesian Learning; which enforces the sparsity in the
to-be-identified PDEs, and therefore can balance the model complexity and
fitting error with theoretical guarantees. Without leveraging prior knowledge
or assumptions in the discovery process, we use an automated approach to
discover ten types of PDEs, including the famous Navier-Stokes and sine-Gordon
equations, from simulation data alone. Moreover, we demonstrate our data-driven
discovery process with the Complex Ginzburg-Landau Equation (CGLE) using data
measured from a traveling-wave convection experiment. Our machine discovery
approach presents solutions that has the potential to inspire, support and
assist physicists for the establishment of physical laws from measured
spatiotemporal data, especially in notorious fields that are often too complex
to allow a straightforward establishment of physical law, such as biophysics,
fluid dynamics, neuroscience or nonlinear optics.
| stat |
Study designs for extending causal inferences from a randomized trial to
a target population | We examine study designs for extending (generalizing or transporting) causal
inferences from a randomized trial to a target population. Specifically, we
consider nested trial designs, where randomized individuals are nested within a
sample from the target population, and non-nested trial designs, including
composite dataset designs, where a randomized trial is combined with a
separately obtained sample of non-randomized individuals from the target
population. We show that the causal quantities that can be identified in each
study design depend on what is known about the probability of sampling
non-randomized individuals. For each study design, we examine identification of
potential outcome means via the g-formula and inverse probability weighting.
Last, we explore the implications of the sampling properties underlying the
designs for the identification and estimation of the probability of trial
participation.
| stat |
Riemannian geometry for Compound Gaussian distributions: application to
recursive change detection | A new Riemannian geometry for the Compound Gaussian distribution is proposed.
In particular, the Fisher information metric is obtained, along with
corresponding geodesics and distance function. This new geometry is applied on
a change detection problem on Multivariate Image Times Series: a recursive
approach based on Riemannian optimization is developed. As shown on simulated
data, it allows to reach optimal performance while being computationally more
efficient.
| stat |
Spatial homogeneity learning for spatially correlated functional data
with application to COVID-19 Growth rate curves | We study the spatial heterogeneity effect on regional COVID-19 pandemic
timing and severity by analyzing the COVID-19 growth rate curves in the United
States. We propose a geographically detailed functional data grouping method
equipped with a functional conditional autoregressive (CAR) prior to fully
capture the spatial correlation in the pandemic curves. The spatial homogeneity
pattern can then be detected by a geographically weighted Chinese restaurant
process prior which allows both locally spatially contiguous groups and
globally discontiguous groups. We design an efficient Markov chain Monte Carlo
(MCMC) algorithm to simultaneously infer the posterior distributions of the
number of groups and the grouping configuration of spatial functional data. The
superior numerical performance of the proposed method over competing methods is
demonstrated using simulated studies and an application to COVID-19 state-level
and county-level data study in the United States.
| stat |
Penalized least squares and sign constraints with modified
Newton-Raphson algorithms: application to EEG source imaging | We propose a modified Newton-Raphson (MNR) algorithm to estimate multiple
penalized least squares (MPLS) models, and its extension to perform efficient
optimization over the active set of selected features (AMNR). MPLS models are a
more flexible approach to find adaptive least squares solutions that can be
simultaneously required to be sparse and smooth. This is particularly important
when addressing real-life inverse problems where there is no ground truth
available, such as electrophysiological source imaging. The proposed MNR
technique can be interpreted as a generalization of the Majorize-Minimize (MM)
algorithm to include combinations of constraints. The AMNR algorithm allows to
extend some penalized least squares methods to the p much greater than n case,
as well as considering sign constraints. We show that these algorithms provide
solutions with acceptable reconstruction in simulated scenarios that do not
cope with model assumptions, for low n/p ratios. We then use both algorithms
for estimating known and new electroencephalography (EEG) inverse models with
multiple penalties. Synthetic data were used for a preliminary comparison with
the corresponding solutions using the least angle regression (LARS) algorithm
according to well-known quality measures; while a visual event-related EEG was
used to illustrate its usefulness in the analysis of real experimental data.
| stat |
A power one test for unit roots based on sample autocovariances | We propose a new unit-root test for a stationary null hypothesis $H_0$
against a unit-root alternative $H_1$. Our approach is nonparametric as the
null hypothesis only assumes that the process concerned is $I(0)$ without
specifying any parametric forms. The new test is based on the fact that the
sample autocovariance function (ACF) converges to the finite population ACF for
an $I(0)$ process while it diverges to infinity with probability approaching
one for a process with unit-roots. Therefore the new test rejects the null
hypothesis for the large values of the sample ACF. To address the technical
challenge `how large is large', we split the sample and establish an
appropriate normal approximation for the null-distribution of the test
statistic. The substantial discriminative power of the new test statistic is
rooted from the fact that it takes finite value under $H_0$ and diverges to
infinity almost surely under $H_1$. This allows us to truncate the critical
values of the test to make it with the asymptotic power one. It also alleviates
the loss of power due to the sample-splitting. The finite sample properties of
the test are illustrated by simulation which shows its stable and more powerful
performance in comparison with the KPSS test (Kwiatkowski et al., 1992). The
test is implemented in a user-friendly R-function.
| stat |
Robust Differential Abundance Test in Compositional Data | Differential abundance tests in the compositional data are essential and
fundamental tasks in various biomedical applications, such as single-cell, bulk
RNA-seq, and microbiome data analysis. Despite the recent developments in the
fields, differential abundance analysis in the compositional data is still a
complicated and unsolved statistical problem because of the compositional
constraint and prevalent zero counts in the dataset. A new differential
abundance test is introduced in this paper to address these challenges,
referred to as the robust differential abundance (RDB) test. Compared with
existing methods, the RDB test 1) is simple and computationally efficient, 2)
is robust to prevalent zero counts in compositional datasets, 3) can take the
data's compositional nature into account, and 4) has a theoretical guarantee to
control false discoveries in a general setting. Furthermore, in the presence of
observed covariates, the RDB test can work with the covariate balancing
techniques to remove the potential confounding effects and draw reliable
conclusions. To demonstrate its practical merits, we apply the new test to
several numerical examples using both simulated and real datasets.
| stat |
Scalable Bayesian dynamic covariance modeling with variational Wishart
and inverse Wishart processes | We implement gradient-based variational inference routines for Wishart and
inverse Wishart processes, which we apply as Bayesian models for the dynamic,
heteroskedastic covariance matrix of a multivariate time series. The Wishart
and inverse Wishart processes are constructed from i.i.d. Gaussian processes,
existing variational inference algorithms for which form the basis of our
approach. These methods are easy to implement as a black-box and scale
favorably with the length of the time series, however, they fail in the case of
the Wishart process, an issue we resolve with a simple modification into an
additive white noise parameterization of the model. This modification is also
key to implementing a factored variant of the construction, allowing inference
to additionally scale to high-dimensional covariance matrices. Through
experimentation, we demonstrate that some (but not all) model variants
outperform multivariate GARCH when forecasting the covariances of returns on
financial instruments.
| stat |
Smoothing methods to estimate the hazard rate under double truncation | In Survival Analysis, the observed lifetimes often correspond to individuals
for which the event occurs within a specific calendar time interval. With such
interval sampling, the lifetimes are doubly truncated at values determined by
the birth dates and the sampling interval. This double truncation may induce a
systematic bias in estimation, so specific corrections are needed. A relevant
target in Survival Analysis is the hazard rate function, which represents the
instantaneous probability for the event of interest. In this work we introduce
a flexible estimation approach for the hazard rate under double truncation.
Specifically, a kernel smoother is considered, in both a fully nonparametric
setting and a semiparametric setting in which the incidence process fits a
given parametric model. Properties of the kernel smoothers are investigated
both theoretically and through simulations. In particular, an asymptotic
expression of the mean integrated squared error is derived, leading to a
data-driven bandwidth for the estimators. The relevance of the semiparametric
approach is emphasized, in that it is generally more accurate and, importantly,
it avoids the potential issues of nonexistence or nonuniqueness of the fully
nonparametric estimator. Applications to the age of diagnosis of Acute Coronary
Syndrome (ACS) and AIDS incubation times are included.
| stat |
Quantum Implementation of Risk Analysis-relevant Copulas | Modern quantitative risk management relies on an adequate modeling of the
tail dependence and a possibly accurate quantification of risk measures, like
Value at Risk (VaR), at high confidence levels like 1 in 100 or even 1 in 2000.
Quantum computing makes such a quantification quadratically more efficient than
the Monte Carlo method; see (Woerner and Egger, 2018) and, for a broader
perspective, (Or\'us et al., 2018). An important element of the risk analysis
toolbox is copula, see (Jouanin et al., 2004) regarding financial applications.
However, to the best knowledge of the author, no quantum computing
implementation for sampling from a risk modeling-relevant copula in explicit
form has been published so far. Our focus here is implementation of simple yet
powerful copula models, capable of a satisfactory capturing the joint tail
behaviour of the modelled risk factors. This paper deals with a few simple
copula families, including Multivariate B11 (MB11) copula family, presented in
(Milek, 2014). We will show that this copula family is suitable for the risk
aggregation as it is exceptionally able to reproduce tail dependence
structures; see (Embrechts et al., 2016) for a relevant benchmark as well as
necessary and sufficient conditions regarding the ultimate feasible bivariate
tail dependence structures. It turns out that such a discretized copula can be
expressed using simple constructs present in the quantum computing: binary
fraction expansion format, comonotone/independent random variables, controlled
gates, and convex combinations, and is therefore suitable for a quantum
computer implementation. This paper presents design behind the quantum
implementation circuits, numerical and symbolic simulation results, and
experimental validation on IBM quantum computer. The paper proposes also a
generic method for quantum implementation of any discretized copula.
| stat |
A Simple Algorithm for Scalable Monte Carlo Inference | The methods of statistical physics are widely used for modelling complex
networks. Building on the recently proposed Equilibrium Expectation approach,
we derive a simple and efficient algorithm for maximum likelihood estimation
(MLE) of parameters of exponential family distributions - a family of
statistical models, that includes Ising model, Markov Random Field and
Exponential Random Graph models. Computational experiments and analysis of
empirical data demonstrate that the algorithm increases by orders of magnitude
the size of network data amenable to Monte Carlo based inference. We report
results suggesting that the applicability of the algorithm may readily be
extended to the analysis of large samples of dependent observations commonly
found in biology, sociology, astrophysics, and ecology.
| stat |
Modelling reporting delays for disease surveillance data | One difficulty for real-time tracking of epidemics is related to reporting
delay. The reporting delay may be due to laboratory confirmation, logistic
problems, infrastructure difficulties and so on. The ability to correct the
available information as quickly as possible is crucial, in terms of decision
making such as issuing warnings to the public and local authorities. A Bayesian
hierarchical modelling approach is proposed as a flexible way of correcting the
reporting delays and to quantify the associated uncertainty. Implementation of
the model is fast, due to the use of the integrated nested Laplace
approximation (INLA). The approach is illustrated on dengue fever incidence
data in Rio de Janeiro, and Severe Acute Respiratory Illness (SARI) data in
Paran\'a state, Brazil.
| stat |
Approximate Variational Estimation for a Model of Network Formation | We develop approximate estimation methods for exponential random graph models
(ERGMs), whose likelihood is proportional to an intractable normalizing
constant. The usual approach approximates this constant with Monte Carlo
simulations, however convergence may be exponentially slow. We propose a
deterministic method, based on a variational mean-field approximation of the
ERGM's normalizing constant. We compute lower and upper bounds for the
approximation error for any network size, adapting nonlinear large deviations
results. This translates into bounds on the distance between true likelihood
and mean-field likelihood. Monte Carlo simulations suggest that in practice our
deterministic method performs better than our conservative theoretical
approximation bounds imply, for a large class of models.
| stat |
Gaussian-Dirichlet Random Fields for Inference over High Dimensional
Categorical Observations | We propose a generative model for the spatio-temporal distribution of high
dimensional categorical observations. These are commonly produced by robots
equipped with an imaging sensor such as a camera, paired with an image
classifier, potentially producing observations over thousands of categories.
The proposed approach combines the use of Dirichlet distributions to model
sparse co-occurrence relations between the observed categories using a latent
variable, and Gaussian processes to model the latent variable's spatio-temporal
distribution. Experiments in this paper show that the resulting model is able
to efficiently and accurately approximate the temporal distribution of high
dimensional categorical measurements such as taxonomic observations of
microscopic organisms in the ocean, even in unobserved (held out) locations,
far from other samples. This work's primary motivation is to enable deployment
of informative path planning techniques over high dimensional categorical
fields, which until now have been limited to scalar or low dimensional vector
observations.
| stat |
Variance Reduction on Adaptive Stochastic Mirror Descent | In this work, we study the idea of variance reduction applied to adaptive
stochastic mirror descent algorithms in the nonsmooth nonconvex finite-sum
optimization problems. We propose a simple yet generalized adaptive mirror
descent algorithm with variance reduction named SVRAMD and provide its
convergence analysis in different settings. We prove that variance reduction
reduces the SFO complexity of most adaptive mirror descent algorithms and
accelerates their convergence. In particular, our general theory implies that
variance reduction can be applied to algorithms using time-varying step sizes
and self-adaptive algorithms such as AdaGrad and RMSProp. Moreover, the
convergence rates of SVRAMD recover the best existing rates of non-adaptive
variance reduced mirror descent algorithms. We check the validity of our claims
using experiments in deep learning.
| stat |
Simultaneous Transformation and Rounding (STAR) Models for
Integer-Valued Data | We propose a simple yet powerful framework for modeling integer-valued data,
such as counts, scores, and rounded data. The data-generating process is
defined by Simultaneously Transforming and Rounding (STAR) a continuous-valued
process, which produces a flexible family of integer-valued distributions
capable of modeling zero-inflation, bounded or censored data, and over- or
underdispersion. The transformation is modeled as unknown for greater
distributional flexibility, while the rounding operation ensures a coherent
integer-valued data-generating process. An efficient MCMC algorithm is
developed for posterior inference and provides a mechanism for adaptation of
successful Bayesian models and algorithms for continuous data to the
integer-valued data setting. Using the STAR framework, we design a new Bayesian
Additive Regression Tree (BART) model for integer-valued data, which
demonstrates impressive predictive distribution accuracy for both synthetic
data and a large healthcare utilization dataset. For interpretable
regression-based inference, we develop a STAR additive model, which offers
greater flexibility and scalability than existing integer-valued models. The
STAR additive model is applied to study the recent decline in Amazon river
dolphins.
| stat |
Estimating and Forecasting the Smoking-Attributable Mortality Fraction
for Both Genders Jointly in Over 60 Countries | Smoking is one of the preventable threats to human health and is a major risk
factor for lung cancer, upper aero-digestive cancer, and chronic obstructive
pulmonary disease. Estimating and forecasting the smoking attributable fraction
(SAF) of mortality can yield insights into smoking epidemics and also provide a
basis for more accurate mortality and life expectancy projection. Peto et al.
(1992) proposed a method to estimate the SAF using the lung cancer mortality
rate as an indicator of exposure to smoking in the population of interest. Here
we use the same method to estimate the all-age SAF (ASAF) for both genders for
over 60 countries. We document a strong and cross-nationally consistent pattern
of the evolution of the SAF over time. We use this as the basis for a new
Bayesian hierarchical model to project future male and female ASAF from over 60
countries simultaneously. This gives forecasts as well as predictive
distributions that can be used to find uncertainty intervals for any quantities
of interest. We assess the model using out-of-sample predictive validation, and
find that it provides good forecasts and well calibrated forecast intervals.
| stat |
A functional-data approach to the Argo data | The Argo data is a modern oceanography dataset that provides unprecedented
global coverage of temperature and salinity measurements in the upper 2,000
meters of depth of the ocean. We study the Argo data from the perspective of
functional data analysis (FDA). We develop spatio-temporal functional kriging
methodology for mean and covariance estimation to predict temperature and
salinity at a fixed location as a smooth function of depth. By combining tools
from FDA and spatial statistics, including smoothing splines, local regression,
and multivariate spatial modeling and prediction, our approach provides
advantages over current methodology that consider pointwise estimation at fixed
depths. Our approach naturally leverages the irregularly-sampled data in space,
time, and depth to fit a space-time functional model for temperature and
salinity. The developed framework provides new tools to address fundamental
scientific problems involving the entire upper water column of the oceans such
as the estimation of ocean heat content, stratification, and thermohaline
oscillation. For example, we show that our functional approach yields more
accurate ocean heat content estimates than ones based on discrete integral
approximations in pressure. Further, using the derivative function estimates,
we obtain a new product of a global map of the mixed layer depth, a key
component in the study of heat absorption and nutrient circulation in the
oceans. The derivative estimates also reveal evidence for density inversions in
areas distinguished by mixing of particularly different water masses.
| stat |
Component-Wise Boosting of Targets for Multi-Output Prediction | Multi-output prediction deals with the prediction of several targets of
possibly diverse types. One way to address this problem is the so called
problem transformation method. This method is often used in multi-label
learning, but can also be used for multi-output prediction due to its
generality and simplicity. In this paper, we introduce an algorithm that uses
the problem transformation method for multi-output prediction, while
simultaneously learning the dependencies between target variables in a sparse
and interpretable manner. In a first step, predictions are obtained for each
target individually. Target dependencies are then learned via a component-wise
boosting approach. We compare our new method with similar approaches in a
benchmark using multi-label, multivariate regression and mixed-type datasets.
| stat |
Blinded sample size re-calculation in multiple composite population
designs with normal data and baseline adjustments | The increasing interest in subpopulation analysis has led to the development
of various new trial designs and analysis methods in the fields of personalized
medicine and targeted therapies. In this paper, subpopulations are defined in
terms of an accumulation of disjoint population subsets and will therefore be
called composite populations. The proposed trial design is applicable to any
set of composite populations, considering normally distributed endpoints and
random baseline covariates. Treatment effects for composite populations are
tested by combining $p$-values, calculated on the subset levels, using the
inverse normal combination function to generate test statistics for those
composite populations. The family-wise type I error rate for simultaneous
testing is controlled in the strong sense by the application of the closed
testing procedure. Critical values for intersection hypothesis tests are
derived using multivariate normal distributions, reflecting the joint
distribution of composite population test statistics under the null hypothesis.
For sample size calculation and sample size re-calculation multivariate normal
distributions are derived which describe the joint distribution of composite
population test statistics under an assumed alternative hypothesis. Simulations
demonstrate the strong control of the family-wise type I error rate in fixed
designs and re-calculation designs with blinded sample size re-calculation. The
target power after sample size re-calculation is typically met or close to
being met.
| stat |
Spatial modeling of shot conversion in soccer to single out goalscoring
ability | Goals are results of pin-point shots and it is a pivotal decision in soccer
when, how and where to shoot. The main contribution of this study is two-fold.
At first, after showing that there exists high spatial correlation in the data
of shots across games, we introduce a spatial process in the error structure to
model the probability of conversion from a shot depending on positional and
situational covariates. The model is developed using a full Bayesian framework.
Secondly, based on the proposed model, we define two new measures that can
appropriately quantify the impact of an individual in soccer, by evaluating the
positioning senses and shooting abilities of the players. As a practical
application, the method is implemented on Major League Soccer data from 2016/17
season.
| stat |
Group testing and PCR: a tale of charge value | The original problem of group testing consists in the identification of
defective items in a collection, by applying tests on groups of items that
detect the presence of at least one defective item in the group. The aim is
then to identify all defective items of the collection with as few tests as
possible. This problem is relevant in several fields, among which biology and
computer sciences. It recently gained attraction as a potential tool to solve a
shortage of COVID-19 test kits, in particular for RT-qPCR. However, the problem
of group testing is not an exact match to this implementation. Indeed,
contrarily to the original problem, PCR testing employed for the detection of
COVID-19 returns more than a simple binary contaminated/non-contaminated value
when applied to a group of samples collected on different individuals. It gives
a real value representing the viral load in the sample instead. We study here
the use that can be made of this extra piece of information to construct a
one-stage pool testing algorithms on an idealize version of this model. We show
that under the right conditions, the total number of tests needed to detect
contaminated samples diminishes drastically.
| stat |
Quadruply Stochastic Gaussian Processes | We introduce a stochastic variational inference procedure for training
scalable Gaussian process (GP) models whose per-iteration complexity is
independent of both the number of training points, $n$, and the number basis
functions used in the kernel approximation, $m$. Our central contributions
include an unbiased stochastic estimator of the evidence lower bound (ELBO) for
a Gaussian likelihood, as well as a stochastic estimator that lower bounds the
ELBO for several other likelihoods such as Laplace and logistic. Independence
of the stochastic optimization update complexity on $n$ and $m$ enables
inference on huge datasets using large capacity GP models. We demonstrate
accurate inference on large classification and regression datasets using GPs
and relevance vector machines with up to $m = 10^7$ basis functions.
| stat |
Uncertainty-Aware (UNA) Bases for Bayesian Regression Using Multi-Headed
Auxiliary Networks | Neural Linear Models (NLM) are deep Bayesian models that produce predictive
uncertainty by learning features from the data and then performing Bayesian
linear regression over these features. Despite their popularity, few works have
focused on formally evaluating the predictive uncertainties of these models.
Furthermore, existing works point out the difficulties of encoding domain
knowledge in models like NLMs, making them unsuitable for applications where
interpretability is required. In this work, we show that traditional training
procedures for NLMs can drastically underestimate uncertainty in data-scarce
regions. We identify the underlying reasons for this behavior and propose a
novel training method that can both capture useful predictive uncertainties as
well as allow for incorporation of domain knowledge.
| stat |
The Local Randomization Framework for Regression Discontinuity Designs:
A Review and Some Extensions | Regression discontinuity designs (RDDs) are a common quasi-experiment in
economics and statistics. The most popular methodologies for analyzing RDDs
utilize continuity-based assumptions and local polynomial regression, but
recent works have developed alternative assumptions based on local
randomization. The local randomization framework avoids modeling assumptions by
instead placing assumptions on the assignment mechanism near the cutoff.
However, most works have focused on completely randomized assignment
mechanisms, which posit that propensity scores are equal for all units near the
cutoff. In our review of the local randomization framework, we extend the
framework to allow for any assignment mechanism, such that propensity scores
may differ. We outline randomization tests that can be used to select a window
around the cutoff where a particular assignment mechanism is most plausible, as
well as methodologies for estimating causal effects after a window and
assignment mechanism are chosen. We apply our methodology to a fuzzy RDD
assessing the effects of financial aid on college dropout rates in Italy. We
find that positing different assignment mechanisms within a single RDD can
provide more nuanced sensitivity analyses as well as more precise inferences
for causal effects.
| stat |
Nonparametric Pattern-Mixture Models for Inference with Missing Data | Pattern-mixture models provide a transparent approach for handling missing
data, where the full-data distribution is factorized in a way that explicitly
shows the parts that can be estimated from observed data alone, and the parts
that require identifying restrictions. We introduce a nonparametric estimator
of the full-data distribution based on the pattern-mixture model factorization.
Our approach uses the empirical observed-data distribution and augments it with
a nonparametric estimator of the missing-data distributions under a given
identifying restriction. Our results apply to a large class of donor-based
identifying restrictions that encompasses commonly used ones and can handle
both monotone and nonmonotone missingness. We propose a Monte Carlo procedure
to derive point estimates of functionals of interest, and the bootstrap to
construct confidence intervals.
| stat |
Fiducial Matching for the Approximate Posterior: F-ABC | F-ABC is introduced, using universal sufficient statistics, unlike previous
ABC papers, e.g. Bernton et al. (2019), and avoiding in the approximate
posterior artifacts due to a Kernel. The nature of matching tolerance is
examined and indications for determining its values are presented. F-ABC does
not face concerns associated with ABC. Asymptotics and simulation results are
also presented.
| stat |
$\sigma$-Ridge: group regularized ridge regression via empirical Bayes
noise level cross-validation | Features in predictive models are not exchangeable, yet common supervised
models treat them as such. Here we study ridge regression when the analyst can
partition the features into $K$ groups based on external side-information. For
example, in high-throughput biology, features may represent gene expression,
protein abundance or clinical data and so each feature group represents a
distinct modality. The analyst's goal is to choose optimal regularization
parameters $\lambda = (\lambda_1, \dotsc, \lambda_K)$ -- one for each group. In
this work, we study the impact of $\lambda$ on the predictive risk of
group-regularized ridge regression by deriving limiting risk formulae under a
high-dimensional random effects model with $p\asymp n$ as $n \to \infty$.
Furthermore, we propose a data-driven method for choosing $\lambda$ that
attains the optimal asymptotic risk: The key idea is to interpret the residual
noise variance $\sigma^2$, as a regularization parameter to be chosen through
cross-validation. An empirical Bayes construction maps the one-dimensional
parameter $\sigma$ to the $K$-dimensional vector of regularization parameters,
i.e., $\sigma \mapsto \widehat{\lambda}(\sigma)$. Beyond its theoretical
optimality, the proposed method is practical and runs as fast as
cross-validated ridge regression without feature groups ($K=1$).
| stat |
An RKHS-Based Semiparametric Approach to Nonlinear Sufficient Dimension
Reduction | Based on the theory of reproducing kernel Hilbert space (RKHS) and
semiparametric method, we propose a new approach to nonlinear dimension
reduction. The method extends the semiparametric method into a more generalized
domain where both the interested parameters and nuisance parameters to be
infinite dimensional. By casting the nonlinear dimensional reduction problem in
a generalized semiparametric framework, we calculate the orthogonal complement
space of generalized nuisance tangent space to derive the estimating equation.
Solving the estimating equation by the theory of RKHS and regularization, we
obtain the estimation of dimension reduction directions of the sufficient
dimension reduction (SDR) subspace and also show the asymptotic property of
estimator. Furthermore, the proposed method does not rely on the linearity
condition and constant variance condition. Simulation and real data studies are
conducted to demonstrate the finite sample performance of our method in
comparison with several existing methods.
| stat |
Distance Metric Learning for Graph Structured Data | Graphs are versatile tools for representing structured data. Therefore, a
variety of machine learning methods have been studied for graph data analysis.
Although many of those learning methods depend on the measurement of
differences between input graphs, defining an appropriate distance metric for a
graph remains a controversial issue. Hence, we propose a supervised distance
metric learning method for the graph classification problem. Our method, named
interpretable graph metric learning (IGML), learns discriminative metrics in a
subgraph-based feature space, which has a strong graph representation
capability. By introducing a sparsity-inducing penalty on a weight of each
subgraph, IGML can identify a small number of important subgraphs that can
provide insight about the given classification task. Since our formulation has
a large number of optimization variables, an efficient algorithm is also
proposed by using pruning techniques based on safe screening and working set
selection methods. An important property of IGML is that the optimality of the
solution is guaranteed because the problem is formulated as a convex problem
and our pruning strategies only discard unnecessary subgraphs. Further, we show
that IGML is also applicable to other structured data such as item-set and
sequence data, and that it can incorporate vertex-label similarity by using a
transportation-based subgraph feature. We empirically evaluate the
computational efficiency and classification performance on several benchmark
datasets and show some illustrative examples demonstrating that IGML identifies
important subgraphs from a given graph dataset.
| stat |
Estimation and Inference for the Mediation Effect in a Time-varying
Mediation Model | Traditional mediation analysis typically examines the relations among an
intervention, a time-invariant mediator, and a time-invariant outcome variable.
Although there may be a direct effect of the intervention on the outcome, there
is a need to understand the process by which the intervention affects the
outcome (i.e. the indirect effect through the mediator). This indirect effect
is frequently assumed to be time-invariant. With improvements in data
collection technology, it is possible to obtain repeated assessments over time
resulting in intensive longitudinal data. This calls for an extension of
traditional mediation analysis to incorporate time-varying variables as well as
time-varying effects. In this paper, we focus on estimation and inference for
the time-varying mediation model, which allows mediation effects to vary as a
function of time. We propose a two-step approach to estimate the time-varying
mediation effect. Moreover, we use a simulation based approach to derive the
corresponding point-wise confidence band for the time-varying mediation effect.
Simulation studies show that the proposed procedures perform well when
comparing the confidence band and the true underlying model. We further apply
the proposed model and the statistical inference procedure to real-world data
collected from a smoking cessation study.
| stat |
Discovering Conditionally Salient Features with Statistical Guarantees | The goal of feature selection is to identify important features that are
relevant to explain an outcome variable. Most of the work in this domain has
focused on identifying globally relevant features, which are features that are
related to the outcome using evidence across the entire dataset. We study a
more fine-grained statistical problem: conditional feature selection, where a
feature may be relevant depending on the values of the other features. For
example in genetic association studies, variant $A$ could be associated with
the phenotype in the entire dataset, but conditioned on variant $B$ being
present it might be independent of the phenotype. In this sense, variant $A$ is
globally relevant, but conditioned on $B$ it is no longer locally relevant in
that region of the feature space. We present a generalization of the knockoff
procedure that performs conditional feature selection while controlling a
generalization of the false discovery rate (FDR) to the conditional setting. By
exploiting the feature/response model-free framework of the knockoffs, the
quality of the statistical FDR guarantee is not degraded even when we perform
conditional feature selections. We implement this method and present an
algorithm that automatically partitions the feature space such that it enhances
the differences between selected sets in different regions, and validate the
statistical theoretical results with experiments.
| stat |
Observation on F.W.E.R. and F.D.R. for correlated normal | In this paper, we have attempted to study the behaviour of the family wise
error rate (FWER) for Bonferroni's procedure and false discovery rate (FDR) of
the Benjamini-Hodgeberg procedure for simultaneous testing problem with
equicorrelated normal observations. By simulation study, we have shown that
F.W.E.R. is a concave function for small no. of hypotheses and asymptotically
becomes a convex function of the correlation. The plots of F.W.E.R. and F.D.R.
confirms that if non-negative correlation is present, then these procedures
control the type-I error rate at a much smaller rate than the desired level of
significance. This confirms the conservative nature of these popular methods
when correlation is present and provides a scope for improvement in power by
appropriate adjustment for correlation.
| stat |
Uncertainty in the Design Stage of Two-Stage Bayesian Propensity Score
Analysis | The two-stage process of propensity score analysis (PSA) includes a design
stage where propensity scores are estimated and implemented to approximate a
randomized experiment and an analysis stage where treatment effects are
estimated conditional upon the design. This paper considers how uncertainty
associated with the design stage impacts estimation of causal effects in the
analysis stage. Such design uncertainty can derive from the fact that the
propensity score itself is an estimated quantity, but also from other features
of the design stage tied to choice of propensity score implementation. This
paper offers a procedure for obtaining the posterior distribution of causal
effects after marginalizing over a distribution of design-stage outputs,
lending a degree of formality to Bayesian methods for PSA (BPSA) that have
gained attention in recent literature. Formulation of a probability
distribution for the design-stage output depends on how the propensity score is
implemented in the design stage, and propagation of uncertainty into causal
estimates depends on how the treatment effect is estimated in the analysis
stage. We explore these differences within a sample of commonly-used propensity
score implementations (quantile stratification, nearest-neighbor matching,
caliper matching, inverse probability of treatment weighting, and doubly robust
estimation) and investigate in a simulation study the impact of statistician
choice in PS model and implementation on the degree of between- and
within-design variability in the estimated treatment effect. The methods are
then deployed in an investigation of the association between levels of fine
particulate air pollution and elevated exposure to emissions from coal-fired
power plants.
| stat |
Multiscale Invertible Generative Networks for High-Dimensional Bayesian
Inference | We propose a Multiscale Invertible Generative Network (MsIGN) and associated
training algorithm that leverages multiscale structure to solve
high-dimensional Bayesian inference. To address the curse of dimensionality,
MsIGN exploits the low-dimensional nature of the posterior, and generates
samples from coarse to fine scale (low to high dimension) by iteratively
upsampling and refining samples. MsIGN is trained in a multi-stage manner to
minimize the Jeffreys divergence, which avoids mode dropping in
high-dimensional cases. On two high-dimensional Bayesian inverse problems, we
show superior performance of MsIGN over previous approaches in posterior
approximation and multiple mode capture. On the natural image synthesis task,
MsIGN achieves superior performance in bits-per-dimension over baseline models
and yields great interpret-ability of its neurons in intermediate layers.
| stat |
A Nonlinear Differential Equation for Generating Warping Function | Given set of functions $y_i(t)$ and $x(t)$ such that $y_i(t) = a_i
x\left[h_i(t)\right]$ with $a_i$ being an unknown amplitude with low changes in
time (or $\frac{\Delta a_i}{a^2_i} << 1$) and $h_i(t)$ an unknown warping
function, the paper shows that $h_i(t)$ can be described using a non-linear
differential equation. The differential equation then can be utilized to
estimate the warping function $h_i(t)$ using a nonlinear least-squares
optimization. This differential equation can also be useful for reducing and
analyzing phase variability in data sequences. Results, obtained on synthetic
curves, showed that the proposed method is effective in aligning the curves.
The obtained aligned curves exhibit variation only in amplitude, and phase
variation can be removed efficiently.
| stat |
Non-constant hazard ratios in randomized controlled trials with
composite endpoints | The hazard ratio is routinely used as a summary measure to assess the
treatment effect in clinical trials with time-to-event endpoints. It is
frequently assumed as constant over time although this assumption often does
not hold. When the hazard ratio deviates considerably from being constant, the
average of its plausible values is not a valid measure of the treatment effect,
can be clinically misleading and common sample size formulas are not
appropriate.
In this paper, we study the hazard ratio along time of a two-component
composite endpoint under the assumption that the hazard ratio for each
component is constant.
This work considers two measures for quantifying the non-proportionality of
the hazard ratio: the difference $D$ between the maximum and minimum values of
hazard ratio over time and the relative measure $R$ representing the ratio
between the sample sizes for the minimum detectable and the average effects. We
illustrate $D$ and $R$ by means of the ZODIAC trial where the primary endpoint
was progression-free survival.
We have run a simulation study deriving scenarios for different values of the
hazard ratios, different event rates and different degrees of association
between the components. We illustrate situations that yield non-constant hazard
ratios for the composite endpoints and consider the likely impact on sample
size.
Results show that the distance between the two component hazard ratios plays
an important role, especially when they are close to 1. Furthermore, even when
the treatment effects for each component are similar, if the two-component
hazards are markedly different, hazard ratio of the composite is often
non-constant.
| stat |
Stable Implementation of Probabilistic ODE Solvers | Probabilistic solvers for ordinary differential equations (ODEs) provide
efficient quantification of numerical uncertainty associated with simulation of
dynamical systems. Their convergence rates have been established by a growing
body of theoretical analysis. However, these algorithms suffer from numerical
instability when run at high order or with small step-sizes -- that is, exactly
in the regime in which they achieve the highest accuracy. The present work
proposes and examines a solution to this problem. It involves three components:
accurate initialisation, a coordinate change preconditioner that makes
numerical stability concerns step-size-independent, and square-root
implementation. Using all three techniques enables numerical computation of
probabilistic solutions of ODEs with algorithms of order up to 11, as
demonstrated on a set of challenging test problems. The resulting rapid
convergence is shown to be competitive to high-order, state-of-the-art,
classical methods. As a consequence, a barrier between analysing probabilistic
ODE solvers and applying them to interesting machine learning problems is
effectively removed.
| stat |
Two-Sample Testing for Event Impacts in Time Series | In many application domains, time series are monitored to detect extreme
events like technical faults, natural disasters, or disease outbreaks.
Unfortunately, it is often non-trivial to select both a time series that is
informative about events and a powerful detection algorithm: detection may fail
because the detection algorithm is not suitable, or because there is no shared
information between the time series and the events of interest. In this work,
we thus propose a non-parametric statistical test for shared information
between a time series and a series of observed events. Our test allows
identifying time series that carry information on event occurrences without
committing to a specific event detection methodology. In a nutshell, we test
for divergences of the value distributions of the time series at increasing
lags after event occurrences with a multiple two-sample testing approach. In
contrast to related tests, our approach is applicable for time series over
arbitrary domains, including multivariate numeric, strings or graphs. We
perform a large-scale simulation study to show that it outperforms or is on par
with related tests on our task for univariate time series. We also demonstrate
the real-world applicability of our approach on datasets from social media and
smart home environments.
| stat |
A primer on coupled state-switching models for multiple interacting time
series | State-switching models such as hidden Markov models or Markov-switching
regression models are routinely applied to analyse sequences of observations
that are driven by underlying non-observable states. Coupled state-switching
models extend these approaches to address the case of multiple observation
sequences whose underlying state variables interact. In this paper, we provide
an overview of the modelling techniques related to coupling in state-switching
models, thereby forming a rich and flexible statistical framework particularly
useful for modelling correlated time series. Simulation experiments demonstrate
the relevance of being able to account for an asynchronous evolution as well as
interactions between the underlying latent processes. The models are further
illustrated using two case studies related to a) interactions between a dolphin
mother and her calf as inferred from movement data; and b) electronic health
record data collected on 696 patients within an intensive care unit.
| stat |
Estimating population average treatment effects from experiments with
noncompliance | Randomized control trials (RCTs) are the gold standard for estimating causal
effects, but often use samples that are non-representative of the actual
population of interest. We propose a reweighting method for estimating
population average treatment effects in settings with noncompliance.
Simulations show the proposed compliance-adjusted population estimator
outperforms its unadjusted counterpart when compliance is relatively low and
can be predicted by observed covariates. We apply the method to evaluate the
effect of Medicaid coverage on health care use for a target population of
adults who may benefit from expansions to the Medicaid program. We draw RCT
data from the Oregon Health Insurance Experiment, where less than one-third of
those randomly selected to receive Medicaid benefits actually enrolled.
| stat |
Tighter Bounds on the Log Marginal Likelihood of Gaussian Process
Regression Using Conjugate Gradients | We propose a lower bound on the log marginal likelihood of Gaussian process
regression models that can be computed without matrix factorisation of the full
kernel matrix. We show that approximate maximum likelihood learning of model
parameters by maximising our lower bound retains many of the sparse variational
approach benefits while reducing the bias introduced into parameter learning.
The basis of our bound is a more careful analysis of the log-determinant term
appearing in the log marginal likelihood, as well as using the method of
conjugate gradients to derive tight lower bounds on the term involving a
quadratic form. Our approach is a step forward in unifying methods relying on
lower bound maximisation (e.g. variational methods) and iterative approaches
based on conjugate gradients for training Gaussian processes. In experiments,
we show improved predictive performance with our model for a comparable amount
of training time compared to other conjugate gradient based approaches.
| stat |
Hamiltonian Monte Carlo using an adjoint-differentiated Laplace
approximation: Bayesian inference for latent Gaussian models and beyond | Gaussian latent variable models are a key class of Bayesian hierarchical
models with applications in many fields. Performing Bayesian inference on such
models can be challenging as Markov chain Monte Carlo algorithms struggle with
the geometry of the resulting posterior distribution and can be prohibitively
slow. An alternative is to use a Laplace approximation to marginalize out the
latent Gaussian variables and then integrate out the remaining hyperparameters
using dynamic Hamiltonian Monte Carlo, a gradient-based Markov chain Monte
Carlo sampler. To implement this scheme efficiently, we derive a novel adjoint
method that propagates the minimal information needed to construct the gradient
of the approximate marginal likelihood. This strategy yields a scalable
differentiation method that is orders of magnitude faster than state of the art
differentiation techniques when the hyperparameters are high dimensional. We
prototype the method in the probabilistic programming framework Stan and test
the utility of the embedded Laplace approximation on several models, including
one where the dimension of the hyperparameter is $\sim$6,000. Depending on the
cases, the benefits can include an alleviation of the geometric pathologies
that frustrate Hamiltonian Monte Carlo and a dramatic speed-up.
| stat |
The FEDHC Bayesian network learning algorithm | The paper proposes a new hybrid Bayesian network learning algorithm, termed
Forward Early Dropping Hill Climbing (FEDHC), designed to work with either
continuous or categorical data. FEDHC consists of a skeleton identification
phase (learning the conditional associations among the variables) followed by
the scoring phase that assigns the (causal) directions. Specifically for the
case of continuous data, a robust to outliers version of FEDHC is also
proposed. Further, the paper manifests that the only implementation of MMHC in
the statistical software \textit{R}, is prohibitively expensive and a new
implementation is offered. The FEDHC is tested via Monte Carlo simulations that
distinctly show it is computationally efficient, and produces Bayesian networks
of similar to, or of higher accuracy than MMHC and PCHC. Specifically, FEDHC
yields more accurate Bayesian networks than PCHC with continuous data but less
accurate with categorical data. Finally, an application of FEDHC, PCHC and MMHC
algorithms to real data, from the field of economics, is demonstrated using the
statistical software \textit{R}.
| stat |
Bayesian semi-parametric G-computation for causal inference in a cohort
study with MNAR dropout and death | Causal inference with observational longitudinal data and time-varying
exposures is often complicated by time-dependent confounding and attrition. The
G-computation formula is one approach for estimating a causal effect in this
setting. The parametric modeling approach typically used in practice relies on
strong modeling assumptions for valid inference, and moreover depends on an
assumption of missing at random, which is not appropriate when the missingness
is missing not at random (MNAR) or due to death. In this work we develop a
flexible Bayesian semi-parametric G-computation approach for assessing the
causal effect on the subpopulation that would survive irrespective of exposure,
in a setting with MNAR dropout. The approach is to specify models for the
observed data using Bayesian additive regression trees, and then use
assumptions with embedded sensitivity parameters to identify and estimate the
causal effect. The proposed approach is motivated by a longitudinal cohort
study on cognition, health, and aging, and we apply our approach to study the
effect of becoming a widow on memory. We also compare our approach to several
standard methods.
| stat |
Estimation and inference for the indirect effect in high-dimensional
linear mediation models | Mediation analysis is difficult when the number of potential mediators is
larger than the sample size. In this paper we propose new inference procedures
for the indirect effect in the presence of high-dimensional mediators for
linear mediation models. We develop methods for both incomplete mediation,
where a direct effect may exist, as well as complete mediation, where the
direct effect is known to be absent. We prove consistency and asymptotic
normality of our indirect effect estimators. Under complete mediation, where
the indirect effect is equivalent to the total effect, we further prove that
our approach gives a more powerful test compared to directly testing for the
total effect. We confirm our theoretical results in simulations, as well as in
an integrative analysis of gene expression and genotype data from a
pharmacogenomic study of drug response. We present a novel analysis of gene
sets to understand the molecular mechanisms of drug response, and also identify
a genome-wide significant noncoding genetic variant that cannot be detected
using standard analysis methods.
| stat |
Demand Forecasting in the Presence of Systematic Events: Cases in
Capturing Sales Promotions | Reliable demand forecasts are critical for the effective supply chain
management. Several endogenous and exogenous variables can influence the
dynamics of demand, and hence a single statistical model that only consists of
historical sales data is often insufficient to produce accurate forecasts. In
practice, the forecasts generated by baseline statistical models are often
judgmentally adjusted by forecasters to incorporate factors and information
that are not incorporated in the baseline models. There are however systematic
events whose effect can be effectively quantified and modeled to help minimize
human intervention in adjusting the baseline forecasts. In this paper, we
develop and test a novel regime-switching approach to quantify systematic
information/events and objectively incorporate them into the baseline
statistical model. Our simple yet practical and effective model can help limit
forecast adjustments to only focus on the impact of less systematic events such
as sudden climate change or dynamic market activities. The proposed model and
approach is validated empirically using sales and promotional data from two
Australian companies. Discussions focus on a thorough analysis of the
forecasting and benchmarking results. Our analysis indicates that the proposed
model can successfully improve the forecast accuracy when compared to the
current industry practice which heavily relies on human judgment to factor in
all types of information/events.
| stat |
Zeroth-Order Hybrid Gradient Descent: Towards A Principled Black-Box
Optimization Framework | In this work, we focus on the study of stochastic zeroth-order (ZO)
optimization which does not require first-order gradient information and uses
only function evaluations. The problem of ZO optimization has emerged in many
recent machine learning applications, where the gradient of the objective
function is either unavailable or difficult to compute. In such cases, we can
approximate the full gradients or stochastic gradients through function value
based gradient estimates. Here, we propose a novel hybrid gradient estimator
(HGE), which takes advantage of the query-efficiency of random gradient
estimates as well as the variance-reduction of coordinate-wise gradient
estimates. We show that with a graceful design in coordinate importance
sampling, the proposed HGE-based ZO optimization method is efficient both in
terms of iteration complexity as well as function query cost. We provide a
thorough theoretical analysis of the convergence of our proposed method for
non-convex, convex, and strongly-convex optimization. We show that the
convergence rate that we derive generalizes the results for some prominent
existing methods in the nonconvex case, and matches the optimal result in the
convex case. We also corroborate the theory with a real-world black-box attack
generation application to demonstrate the empirical advantage of our method
over state-of-the-art ZO optimization approaches.
| stat |
Sample size re-estimation incorporating prior information on a nuisance
parameter | Prior information is often incorporated informally when planning a clinical
trial. Here, we present an approach on how to incorporate prior information,
such as data from historical clinical trials, into the nuisance parameter based
sample size re-estimation in a design with an internal pilot study. We focus on
trials with continuous endpoints in which the outcome variance is the nuisance
parameter. For planning and analyzing the trial frequentist methods are
considered. Moreover, the external information on the variance is summarized by
the Bayesian meta-analytic-predictive (MAP) approach. To incorporate external
information into the sample size re-estimation, we propose to update the MAP
prior based on the results of the internal pilot study and to re-estimate the
sample size using an estimator from the posterior. By means of a simulation
study, we compare the operating characteristics such as power and sample size
distribution of the proposed procedure with the traditional sample size
re-estimation approach which uses the pooled variance estimator. The simulation
study shows that, if no prior-data conflict is present, incorporating external
information into the sample size re-estimation improves the operating
characteristics compared to the traditional approach. In the case of a
prior-data conflict, that is when the variance of the ongoing clinical trial is
unequal to the prior location, the performance of the traditional sample size
re-estimation procedure is in general superior, even when the prior information
is robustified. When considering to include prior information in sample size
re-estimation, the potential gains should be balanced against the risks.
| stat |
Maximum likelihood estimation in the additive hazards model | The additive hazards model specifies the effect of covariates on the hazard
in an additive way, in contrast to the popular Cox model, in which it is
multiplicative. As non-parametric model, it offers a very flexible way of
modeling time-varying covariate effects. It is most commonly estimated by
ordinary least squares. In this paper we consider the case where covariates are
bounded, and derive the maximum likelihood estimator under the constraint that
the hazard is non-negative for all covariate values in their domain. We
describe an efficient algorithm to find the maximum likelihood estimator. The
method is contrasted with the ordinary least squares approach in a simulation
study, and the method is illustrated on a realistic data set.
| stat |
A review and evaluation of standard methods to handle missing data on
time-varying confounders in marginal structural models | Marginal structural models (MSMs) are commonly used to estimate causal
intervention effects in longitudinal non-randomised studies. A common issue
when analysing data from observational studies is the presence of incomplete
confounder data, which might lead to bias in the intervention effect estimates
if they are not handled properly in the statistical analysis. However, there is
currently no recommendation on how to address missing data on covariates in
MSMs under a variety of missingness mechanisms encountered in practice. We
reviewed existing methods to handling missing data in MSMs and performed a
simulation study to compare the performance of complete case (CC) analysis, the
last observation carried forward (LOCF), the missingness pattern approach
(MPA), multiple imputation (MI) and inverse-probability-of-missingness
weighting (IPMW). We considered three mechanisms for non-monotone missing data
which are common in observational studies using electronic health record data.
Whereas CC analysis lead to biased estimates of the intervention effect in
almost all scenarios, the performance of the other approaches varied across
scenarios. The LOCF approach led to unbiased estimates only under a specific
non-random mechanism in which confounder values were missing when their values
remained unchanged since the previous measurement. In this scenario, MI, the
MPA and IPMW were biased. MI and IPMW led to the estimation of unbiased effects
when data were missing at random, given the covariates or the treatment but
only MI was unbiased when the outcome was a predictor of missingness.
Furthermore, IPMW generally lead to very large standard errors. Lastly,
regardless of the missingness mechanism, the MPA led to unbiased estimates only
when the failure to record a confounder at a given time-point modified the
subsequent relationships between the partially observed covariate and the
outcome.
| stat |
Posterior-based proposals for speeding up Markov chain Monte Carlo | Markov chain Monte Carlo (MCMC) is widely used for Bayesian inference in
models of complex systems. Performance, however, is often unsatisfactory in
models with many latent variables due to so-called poor mixing, necessitating
development of application specific implementations. This paper introduces
"posterior-based proposals" (PBPs), a new type of MCMC update applicable to a
huge class of statistical models (whose conditional dependence structures are
represented by directed acyclic graphs). PBPs generates large joint updates in
parameter and latent variable space, whilst retaining good acceptance rates
(typically 33%). Evaluation against other approaches (from standard Gibbs /
random walk updates to state-of-the-art Hamiltonian and particle MCMC methods)
was carried out for widely varying model types: an individual-based model for
disease diagnostic test data, a financial stochastic volatility model, a mixed
model used in statistical genetics and a population model used in ecology.
Whilst different methods worked better or worse in different scenarios, PBPs
were found to be either near to the fastest or significantly faster than the
next best approach (by up to a factor of 10). PBPs therefore represent an
additional general purpose technique that can be usefully applied in a wide
variety of contexts.
| stat |
Modeling Spatial Dependence with Cauchy Convolution Processes | We study the class of dependence models for spatial data obtained from Cauchy
convolution processes based on different types of kernel functions. We show
that the resulting spatial processes have appealing tail dependence properties,
such as tail dependence at short distances and independence at long distances
with suitable kernel functions. We derive the extreme-value limits of these
processes, study their smoothness properties, and detail some interesting
special cases. To get higher flexibility at sub-asymptotic levels and
separately control the bulk and the tail dependence properties, we further
propose spatial models constructed by mixing a Cauchy convolution process with
a Gaussian process. We demonstrate that this framework indeed provides a rich
class of models for the joint modeling of the bulk and the tail behaviors. Our
proposed inference approach relies on matching model-based and empirical
summary statistics, and an extensive simulation study shows that it yields
accurate estimates. We demonstrate our new methodology by application to a
temperature dataset measured at 97 monitoring stations in the state of
Oklahoma, US. Our results indicate that our proposed model provides a very good
fit to the data, and that it captures both the bulk and the tail dependence
structures accurately.
| stat |
A Robust Generalization of the Rao Test | This paper presents new families of Rao-type test statistics based on the
minimum density power divergence estimators which provide robust
generalizations for testing simple and composite null hypotheses. The
asymptotic null distributions of the proposed tests are obtained and their
robustness properties are also theoretically studied. Numerical illustrations
are provided to substantiate the theory developed. On the whole, the proposed
tests are seen to be excellent alternatives to the classical Rao test.
| stat |
A Deep Generative Model for Fragment-Based Molecule Generation | Molecule generation is a challenging open problem in cheminformatics.
Currently, deep generative approaches addressing the challenge belong to two
broad categories, differing in how molecules are represented. One approach
encodes molecular graphs as strings of text, and learns their corresponding
character-based language model. Another, more expressive, approach operates
directly on the molecular graph. In this work, we address two limitations of
the former: generation of invalid and duplicate molecules. To improve validity
rates, we develop a language model for small molecular substructures called
fragments, loosely inspired by the well-known paradigm of Fragment-Based Drug
Design. In other words, we generate molecules fragment by fragment, instead of
atom by atom. To improve uniqueness rates, we present a frequency-based masking
strategy that helps generate molecules with infrequent fragments. We show
experimentally that our model largely outperforms other language model-based
competitors, reaching state-of-the-art performances typical of graph-based
approaches. Moreover, generated molecules display molecular properties similar
to those in the training sample, even in absence of explicit task-specific
supervision.
| stat |
Making learning more transparent using conformalized performance
prediction | In this work, we study some novel applications of conformal inference
techniques to the problem of providing machine learning procedures with more
transparent, accurate, and practical performance guarantees. We provide a
natural extension of the traditional conformal prediction framework, done in
such a way that we can make valid and well-calibrated predictive statements
about the future performance of arbitrary learning algorithms, when passed an
as-yet unseen training set. In addition, we include some nascent empirical
examples to illustrate potential applications.
| stat |
Learning Markov models via low-rank optimization | Modeling unknown systems from data is a precursor of system optimization and
sequential decision making. In this paper, we focus on learning a Markov model
from a single trajectory of states. Suppose that the transition model has a
small rank despite of having a large state space, meaning that the system
admits a low-dimensional latent structure. We show that one can estimate the
full transition model accurately using a trajectory of length that is
proportional to the total number of states. We propose two maximum likelihood
estimation methods: a convex approach with nuclear-norm regularization and a
nonconvex approach with rank constraint. We explicitly derive the statistical
rates of both estimators in terms of the Kullback-Leiber divergence and the
$\ell_2$ error and also establish a minimax lower bound to assess the tightness
of these rates. For computing the nonconvex estimator, we develop a novel DC
(difference of convex function) programming algorithm that starts with the
convex M-estimator and then successively refines the solution till convergence.
Empirical experiments demonstrate consistent superiority of the nonconvex
estimator over the convex one.
| stat |
Integrative Sparse Partial Least Squares | Partial least squares, as a dimension reduction method, has become
increasingly important for its ability to deal with problems with a large
number of variables. Since noisy variables may weaken the performance of the
model, the sparse partial least squares (SPLS) technique has been proposed to
identify important variables and generate more interpretable results. However,
the small sample size of a single dataset limits the performance of
conventional methods. An effective solution comes from gathering information
from multiple comparable studies. The integrative analysis holds an important
status among multi-datasets analyses. The main idea is to improve estimation
results by assembling raw datasets and analyzing them jointly. In this paper,
we develop an integrative SPLS (iSPLS) method using penalization based on the
SPLS technique. The proposed approach consists of two penalties. The first
penalty conducts variable selection under the context of integrative analysis;
The second penalty, a contrasted one, is imposed to encourage the similarity of
estimates across datasets and generate more reasonable and accurate results.
Computational algorithms are provided. Simulation experiments are conducted to
compare iSPLS with alternative approaches. The practical utility of iSPLS is
shown in the analysis of two TCGA gene expression data.
| stat |
Shrinking characteristics of precision matrix estimators | We propose a framework to shrink a user-specified characteristic of a
precision matrix estimator that is needed to fit a predictive model. Estimators
in our framework minimize the Gaussian negative loglikelihood plus an $L_1$
penalty on a linear or affine function evaluated at the optimization variable
corresponding to the precision matrix. We establish convergence rate bounds for
these estimators and propose an alternating direction method of multipliers
algorithm for their computation. Our simulation studies show that our
estimators can perform better than competitors when they are used to fit
predictive models. In particular, we illustrate cases where our precision
matrix estimators perform worse at estimating the population precision matrix
but better at prediction.
| stat |
Practical Bayesian Optimization for Transportation Simulators | We provide a method to solve optimization problem when objective function is
a complex stochastic simulator of an urban transportation system. To reach this
goal, a Bayesian optimization framework is introduced. We show how the choice
of prior and inference algorithm effect the outcome of our optimization
procedure. We develop dimensionality reduction techniques that allow for our
optimization techniques to be applicable for real-life problems. We develop a
distributed, Gaussian Process Bayesian regression and active learning models
that allow parallel execution of our algorithms and enable usage of high
performance computing. We present a fully Bayesian approach that is more sample
efficient and reduces computational budget. Our framework is supported by
theoretical analysis and an empirical study. We demonstrate our framework on
the problem of calibrating a multi-modal transportation network of city of
Bloomington, Illinois. Finally, we discuss directions for further research.
| stat |
Reparameterization Gradients through Acceptance-Rejection Sampling
Algorithms | Variational inference using the reparameterization trick has enabled
large-scale approximate Bayesian inference in complex probabilistic models,
leveraging stochastic optimization to sidestep intractable expectations. The
reparameterization trick is applicable when we can simulate a random variable
by applying a differentiable deterministic function on an auxiliary random
variable whose distribution is fixed. For many distributions of interest (such
as the gamma or Dirichlet), simulation of random variables relies on
acceptance-rejection sampling. The discontinuity introduced by the
accept-reject step means that standard reparameterization tricks are not
applicable. We propose a new method that lets us leverage reparameterization
gradients even when variables are outputs of a acceptance-rejection sampling
algorithm. Our approach enables reparameterization on a larger class of
variational distributions. In several studies of real and synthetic data, we
show that the variance of the estimator of the gradient is significantly lower
than other state-of-the-art methods. This leads to faster convergence of
stochastic gradient variational inference.
| stat |
Large Margin Multi-modal Multi-task Feature Extraction for Image
Classification | The features used in many image analysis-based applications are frequently of
very high dimension. Feature extraction offers several advantages in
high-dimensional cases, and many recent studies have used multi-task feature
extraction approaches, which often outperform single-task feature extraction
approaches. However, most of these methods are limited in that they only
consider data represented by a single type of feature, even though features
usually represent images from multiple modalities. We therefore propose a novel
large margin multi-modal multi-task feature extraction (LM3FE) framework for
handling multi-modal features for image classification. In particular, LM3FE
simultaneously learns the feature extraction matrix for each modality and the
modality combination coefficients. In this way, LM3FE not only handles
correlated and noisy features, but also utilizes the complementarity of
different modalities to further help reduce feature redundancy in each
modality. The large margin principle employed also helps to extract strongly
predictive features so that they are more suitable for prediction (e.g.,
classification). An alternating algorithm is developed for problem optimization
and each sub-problem can be efficiently solved. Experiments on two challenging
real-world image datasets demonstrate the effectiveness and superiority of the
proposed method.
| stat |
Statistical analysis of two arm randomized pre-post design with one
post-treatment measurement | Randomized pre-post designs, with outcomes measured at baseline and
follow-ups, have been commonly used to compare the clinical effectiveness of
two competing treatments. There are vast, but often conflicting, amount of
information in current literature about the best analytic methods for pre-post
design. It is challenging for applied researchers to make an informed choice.
We discuss six methods commonly used in literature: one way analysis of
variance (ANOVA), analysis of covariance main effect and interaction models on
post-treatment measurement (ANCOVA I and II), ANOVA on change score between
baseline and post-treatment measurements, repeated measures and constrained
repeated measures models (cRM) on baseline and post-treatment measurements as
joint outcomes. We review a number of study endpoints in pre-post designs and
identify the difference in post-treatment measurement as the common treatment
effect that all six methods target. We delineate the underlying differences and
links between these competing methods in homogeneous and heterogeneous study
population. We demonstrate that ANCOVA and cRM outperform other alternatives
because their treatment effect estimators have the smallest variances. cRM has
comparable performance to ANCOVA I main effect model in homogeneous scenario
and to ANCOVA II interaction model in heterogeneous scenario. In spite of that,
ANCOVA has several advantages over cRM, including treating baseline measurement
as covariate because it is not an outcome by definition, the convenience of
incorporating other baseline variables and handling complex heteroscedasticity
patterns in a linear regression framework.
| stat |
Selection of proposal distributions for generalized importance sampling
estimators | The standard importance sampling (IS) estimator, generally does not work well
in examples involving simultaneous inference on several targets as the
importance weights can take arbitrarily large values making the estimator
highly unstable. In such situations, alternative generalized IS estimators
involving samples from multiple proposal distributions are preferred. Just like
the standard IS, the success of these multiple IS estimators crucially depends
on the choice of the proposal distributions. The selection of these proposal
distributions is the focus of this article. We propose three methods based on
(i) a geometric space filling coverage criterion, (ii) a minimax variance
approach, and (iii) a maximum entropy approach. The first two methods are
applicable to any multi-proposal IS estimator, whereas the third approach is
described in the context of Doss's (2010) two-stage IS estimator. For the first
method we propose a suitable measure of coverage based on the symmetric
Kullback-Leibler divergence, while the second and third approaches use
estimates of asymptotic variances of Doss's (2010) IS estimator and Geyer's
(1994) reverse logistic estimator, respectively. Thus, we provide consistent
spectral variance estimators for these asymptotic variances. The proposed
methods for selecting proposal densities are illustrated using various detailed
examples.
| stat |
Probabilistic HIV Recency Classification -- A Logistic Regression
without Labeled Individual Level Training Data | Accurate HIV incidence estimation based on individual recent infection status
(recent vs long-term infection) is important for monitoring the epidemic,
targeting interventions to those at greatest risk of new infection, and
evaluating existing programs of prevention and treatment. Starting from 2015,
the Population-based HIV Impact Assessment (PHIA) individual-level surveys are
implemented in the most-affected countries in sub-Saharan Africa. PHIA is a
nationally-representative HIV-focused survey that combines household visits
with key questions and cutting-edge technologies such as biomarker tests for
HIV antibody and HIV viral load which offer the unique opportunity of
distinguishing between recent infection and long-term infection, and providing
relevant HIV information by age, gender, and location. In this article, we
propose a semi-supervised logistic regression model for estimating individual
level HIV recency status. It incorporates information from multiple data
sources -- the PHIA survey where the true HIV recency status is unknown, and
the cohort studies provided in the literature where the relationship between
HIV recency status and the covariates are presented in the form of a
contingency table. It also utilizes the national level HIV incidence estimates
from the epidemiology model. Applying the proposed model to Malawi PHIA data,
we demonstrate that our approach is more accurate for the individual level
estimation and more appropriate for estimating HIV recency rates at aggregated
levels than the current practice -- the binary classification tree (BCT).
| stat |
Relaxed Softmax for learning from Positive and Unlabeled data | In recent years, the softmax model and its fast approximations have become
the de-facto loss functions for deep neural networks when dealing with
multi-class prediction. This loss has been extended to language modeling and
recommendation, two fields that fall into the framework of learning from
Positive and Unlabeled data. In this paper, we stress the different drawbacks
of the current family of softmax losses and sampling schemes when applied in a
Positive and Unlabeled learning setup. We propose both a Relaxed Softmax loss
(RS) and a new negative sampling scheme based on Boltzmann formulation. We show
that the new training objective is better suited for the tasks of density
estimation, item similarity and next-event prediction by driving uplifts in
performance on textual and recommendation datasets against classical softmax.
| stat |
A Convex Optimization Approach to High-Dimensional Sparse Quadratic
Discriminant Analysis | In this paper, we study high-dimensional sparse Quadratic Discriminant
Analysis (QDA) and aim to establish the optimal convergence rates for the
classification error. Minimax lower bounds are established to demonstrate the
necessity of structural assumptions such as sparsity conditions on the
discriminating direction and differential graph for the possible construction
of consistent high-dimensional QDA rules. We then propose a classification
algorithm called SDAR using constrained convex optimization under the sparsity
assumptions. Both minimax upper and lower bounds are obtained and this
classification rule is shown to be simultaneously rate optimal over a
collection of parameter spaces, up to a logarithmic factor. Simulation studies
demonstrate that SDAR performs well numerically. The algorithm is also
illustrated through an analysis of prostate cancer data and colon tissue data.
The methodology and theory developed for high-dimensional QDA for two groups in
the Gaussian setting are also extended to multi-group classification and to
classification under the Gaussian copula model.
| stat |
Applications of Quantum Annealing in Statistics | Quantum computation offers exciting new possibilities for statistics. This
paper explores the use of the D-Wave machine, a specialized type of quantum
computer, which performs quantum annealing. A general description of quantum
annealing through the use of the D-Wave is given, along with technical issues
to be encountered. Quantum annealing is used to perform maximum likelihood
estimation, generate an experimental design, and perform matrix inversion.
Though the results show that quantum computing is still at an early stage which
is not yet superior to classical computation, there is promise for quantum
computation in the future.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.