title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Sample Complexity Bounds for Recurrent Neural Networks with Application
to Combinatorial Graph Problems | Learning to predict solutions to real-valued combinatorial graph problems
promises efficient approximations. As demonstrated based on the NP-hard edge
clique cover number, recurrent neural networks (RNNs) are particularly suited
for this task and can even outperform state-of-the-art heuristics. However, the
theoretical framework for estimating real-valued RNNs is understood only
poorly. As our primary contribution, this is the first work that upper bounds
the sample complexity for learning real-valued RNNs. While such derivations
have been made earlier for feed-forward and convolutional neural networks, our
work presents the first such attempt for recurrent neural networks. Given a
single-layer RNN with $a$ rectified linear units and input of length $b$, we
show that a population prediction error of $\varepsilon$ can be realized with
at most $\tilde{\mathcal{O}}(a^4b/\varepsilon^2)$ samples. We further derive
comparable results for multi-layer RNNs. Accordingly, a size-adaptive RNN fed
with graphs of at most $n$ vertices can be learned in
$\tilde{\mathcal{O}}(n^6/\varepsilon^2)$, i.e., with only a polynomial number
of samples. For combinatorial graph problems, this provides a theoretical
foundation that renders RNNs competitive.
| stat |
New probability distributions in astrophysics: II. The generalized and
double truncated Lindley | The statistical parameters of five generalizations of the Lindley
distribution, such as the average, variance and moments, are reviewed. A new
double truncated Lindley distribution with three parameters is derived. The new
distributions are applied to model the initial mass function for stars.
| stat |
Stochastic Prototype Embeddings | Supervised deep-embedding methods project inputs of a domain to a
representational space in which same-class instances lie near one another and
different-class instances lie far apart. We propose a probabilistic method that
treats embeddings as random variables. Extending a state-of-the-art
deterministic method, Prototypical Networks (Snell et al., 2017), our approach
supposes the existence of a class prototype around which class instances are
Gaussian distributed. The prototype posterior is a product distribution over
labeled instances, and query instances are classified by marginalizing relative
prototype proximity over embedding uncertainty. We describe an efficient
sampler for approximate inference that allows us to train the model at roughly
the same space and time cost as its deterministic sibling. Incorporating
uncertainty improves performance on few-shot learning and gracefully handles
label noise and out-of-distribution inputs. Compared to the state-of-the-art
stochastic method, Hedged Instance Embeddings (Oh et al., 2019), we achieve
superior large- and open-set classification accuracy. Our method also aligns
class-discriminating features with the axes of the embedding space, yielding an
interpretable, disentangled representation.
| stat |
TiK-means: $K$-means clustering for skewed groups | The $K$-means algorithm is extended to allow for partitioning of skewed
groups. Our algorithm is called TiK-Means and contributes a $K$-means type
algorithm that assigns observations to groups while estimating their
skewness-transformation parameters. The resulting groups and transformation
reveal general-structured clusters that can be explained by inverting the
estimated transformation. Further, a modification of the jump statistic chooses
the number of groups. Our algorithm is evaluated on simulated and real-life
datasets and then applied to a long-standing astronomical dispute regarding the
distinct kinds of gamma ray bursts.
| stat |
A tutorial comparing different covariate balancing methods with an
application evaluating the causal effect of exercise on the progression of
Huntington's Disease | Randomized controlled trials are the gold standard for measuring the causal
effects of treatments on clinical outcomes. However, randomized trials are not
always feasible, and causal treatment effects must, therefore, often be
inferred from observational data. Observational study designs do not allow
conclusions about causal relationships to be drawn unless statistical
techniques are used to account for the imbalance of confounders across groups
while key assumptions hold. Propensity score (PS) and balance weighting are two
useful techniques that aim to reduce the imbalances between treatment groups by
weighting the groups to look alike on the observed confounders. There are many
methods available to estimate PS and balancing weights. However, it is unclear
a priori which will achieve the best trade-off between covariate balance and
effective sample size. Weighted analyses are further complicated by small
studies with limited sample sizes, which is common when studying rare diseases.
To address these issues, we present a step-by-step guide to covariate balancing
strategies, including how to evaluate overlap, obtain estimates of PS and
balancing weights, check for covariate balance, and assess sensitivity to
unobserved confounding. We compare the performance of a number of commonly used
estimation methods on a synthetic data set based on the Physical Activity and
Exercise Outcomes in Huntington Disease (PACE-HD) study, which explored whether
enhanced physical activity affects the progression and severity of the disease.
We provide general guidelines for the choice of method for estimation of PS and
balancing weights, interpretation, and sensitivity analysis of results. We also
present R code for implementing the different methods and assessing balance.
| stat |
Data Integration through outcome adaptive LASSO and a collaborative
propensity score approach | Administrative data, or non-probability sample data, are increasingly being
used to obtain official statistics due to their many benefits over survey
methods. In particular, they are less costly, provide a larger sample size, and
are not reliant on the response rate. However, it is difficult to obtain an
unbiased estimate of the population mean from such data due to the absence of
design weights. Several estimation approaches have been proposed recently using
an auxiliary probability sample which provides representative covariate
information of the target population. However, when this covariate information
is high-dimensional, variable selection is not a straight-forward task even for
a subject matter expert. In the context of efficient and doubly robust
estimation approaches for estimating a population mean, we develop two data
adaptive methods for variable selection using the outcome adaptive LASSO and a
collaborative propensity score, respectively. Simulation studies are performed
in order to verify the performance of the proposed methods versus competing
methods. Finally, we presented an anayisis of the impact of Covid-19 on
Canadians.
| stat |
Transfer learning enhanced physics informed neural network for
phase-field modeling of fracture | We present a new physics informed neural network (PINN) algorithm for solving
brittle fracture problems. While most of the PINN algorithms available in the
literature minimize the residual of the governing partial differential
equation, the proposed approach takes a different path by minimizing the
variational energy of the system. Additionally, we modify the neural network
output such that the boundary conditions associated with the problem are
exactly satisfied. Compared to conventional residual based PINN, the proposed
approach has two major advantages. First, the imposition of boundary conditions
is relatively simpler and more robust. Second, the order of derivatives present
in the functional form of the variational energy is of lower order than in the
residual form. Hence, training the network is faster. To compute the total
variational energy of the system, an efficient scheme that takes as input a
geometry described by spline based CAD model and employs Gauss quadrature rules
for numerical integration has been proposed. Moreover, we utilize the concept
of transfer learning to obtain the crack path in an efficient manner. The
proposed approach is used to solve four fracture mechanics problems. For all
the examples, results obtained using the proposed approach match closely with
the results available in the literature. For the first two examples, we compare
the results obtained using the proposed approach with the conventional residual
based neural network results. For both the problems, the proposed approach is
found to yield better accuracy compared to conventional residual based PINN
algorithms.
| stat |
Discovering causal factors of drought in Ethiopia | Drought is a costly natural hazard, many aspects of which remain poorly
understood. It has many contributory factors, driving its outset, duration, and
severity, including land surface, anthropogenic activities, and, most
importantly, meteorological anomalies. Prediction plays a crucial role in
drought preparedness and risk mitigation. However, this is a challenging task
at socio-economically critical lead times (1-2 years), because meteorological
anomalies operate at a wide range of temporal and spatial scales. Among them,
past studies have shown a correlation between the Sea Surface Temperature (SST)
anomaly and the amount of precipitation in various locations in Africa. In its
Eastern part, the cooling phase of El Nino-Southern Oscillation (ENSO) and SST
anomaly in the Indian ocean are correlated with the lack of rainfall. Given the
intrinsic shortcomings of correlation coefficients, we investigate the
association among SST modes of variability and the monthly fraction of grid
points in Ethiopia, which are in drought conditions in terms of causality.
Using the empirical extreme quantiles of precipitation distribution as a proxy
for drought, We show that the level of SST second mode of variability in the
prior year influences the occurrence of drought in Ethiopia. The causal link
between these two variables has a negative coefficient that verifies the
conclusion of past studies that rainfall deficiency in the Horn of Africa is
associated with ENSO's cooling phase.
| stat |
Sensitivity of quantum PageRank | In this paper, we discuss the sensitivity of quantum PageRank. By using the
finite dimensional perturbation theory, we estimate the change of the quantum
PageRank under a small analytical perturbation on the Google matrix. In
addition, we will show the way to estimate the lower bound of the convergence
radius as well as the error bound of the finite sum in the expansion of the
perturbed PageRank.
| stat |
Graphical approaches for the control of generalised error rates | When simultaneously testing multiple hypotheses, the usual approach in the
context of confirmatory clinical trials is to control the familywise error rate
(FWER), which bounds the probability of making at least one false rejection. In
many trial settings, these hypotheses will additionally have a hierarchical
structure that reflects the relative importance and links between different
clinical objectives. The graphical approach of Bretz et al. (2009) is a
flexible and easily communicable way of controlling the FWER while respecting
complex trial objectives and multiple structured hypotheses. However, the FWER
can be a very stringent criterion that leads to procedures with low power, and
may not be appropriate in exploratory trial settings. This motivates
controlling generalised error rates, particularly when the number of hypotheses
tested is no longer small. We consider the generalised familywise error rate
(k-FWER), which is the probability of making k or more false rejections, as
well as the tail probability of the false discovery proportion (FDP), which is
the probability that the proportion of false rejections is greater than some
threshold. We also consider asymptotic control of the false discovery rate
(FDR), which is the expectation of the FDP. In this paper, we show how to
control these generalised error rates when using the graphical approach and its
extensions. We demonstrate the utility of the resulting graphical procedures on
three clinical trial case studies.
| stat |
A Self-Penalizing Objective Function for Scalable Interaction Detection | We tackle the problem of nonparametric variable selection with a focus on
discovering interactions between variables. With $p$ variables there are
$O(p^s)$ possible order-$s$ interactions making exhaustive search infeasible.
It is nonetheless possible to identify the variables involved in interactions
with only linear computation cost, $O(p)$. The trick is to maximize a class of
parametrized nonparametric dependence measures which we call metric learning
objectives; the landscape of these nonconvex objective functions is sensitive
to interactions but the objectives themselves do not explicitly model
interactions. Three properties make metric learning objectives highly
attractive:
(a) The stationary points of the objective are automatically sparse (i.e.
performs selection) -- no explicit $\ell_1$ penalization is needed.
(b) All stationary points of the objective exclude noise variables with high
probability.
(c) Guaranteed recovery of all signal variables without needing to reach the
objective's global maxima or special stationary points.
The second and third properties mean that all our theoretical results apply
in the practical case where one uses gradient ascent to maximize the metric
learning objective. While not all metric learning objectives enjoy good
statistical power, we design an objective based on $\ell_1$ kernels that does
exhibit favorable power: it recovers (i) main effects with $n \sim \log p$
samples, (ii) hierarchical interactions with $n \sim \log p$ samples and (iii)
order-$s$ pure interactions with $n \sim p^{2(s-1)}\log p$ samples.
| stat |
Understanding Variational Inference in Function-Space | Recent work has attempted to directly approximate the `function-space' or
predictive posterior distribution of Bayesian models, without approximating the
posterior distribution over the parameters. This is appealing in e.g. Bayesian
neural networks, where we only need the former, and the latter is hard to
represent. In this work, we highlight some advantages and limitations of
employing the Kullback-Leibler divergence in this setting. For example, we show
that minimizing the KL divergence between a wide class of parametric
distributions and the posterior induced by a (non-degenerate) Gaussian process
prior leads to an ill-defined objective function. Then, we propose (featurized)
Bayesian linear regression as a benchmark for `function-space' inference
methods that directly measures approximation quality. We apply this methodology
to assess aspects of the objective function and inference scheme considered in
Sun, Zhang, Shi, and Grosse (2018), emphasizing the quality of approximation to
Bayesian inference as opposed to predictive performance.
| stat |
Covariate Selection Based on a Model-free Approach to Linear Regression
with Exact Probabilities | In this paper we propose a completely new approach to the problem of
covariate selection in linear regression. It is intuitive, very simple, fast
and powerful, non-frequentist and non-Bayesian. It does not overfit, there is
no shrinkage of the least squares coefficients, and it is model-free. A
covariate or a set of covariates is included only if they are better in the
sense of least squares than the same number of Gaussian covariates consisting
of i.i.d. $N(0,1)$ random variables. The degree to which they are better is
measured by the P-value which is the probability that the Gaussian covariates
are better. This probability is given in terms of the Beta distribution, it is
exact and it holds for the data at hand whatever this may be. The idea extends
to a stepwise procedure, the main contribution of the paper, where the best of
the remaining covariates is only accepted if it is better than the best of the
same number of random Gaussian covariates. Again this probability is given in
terms of the Beta distribution, it is exact and it holds for the data at hand
whatever this may be. We use a version with default parameters which works for
a large collection of known data sets with up to a few hundred thousand
covariates. The computing time for the largest data sets was about four
seconds, and it outperforms all other selection procedures of which we are
aware. The paper gives the results of simulations, applications to real data
sets and theorems on the asymptotic behaviour under the standard linear model.
An R-package {\it gausscov} is available. \
| stat |
Rank-based Lasso -- efficient methods for high-dimensional robust model
selection | We consider the problem of identifying significant predictors in large data
bases, where the response variable depends on the linear combination of
explanatory variables through an unknown link function, corrupted with the
noise from the unknown distribution. We utilize the natural, robust and
efficient approach, which relies on replacing values of the response variables
by their ranks and then identifying significant predictors by using well known
Lasso. We provide new consistency results for the proposed procedure (called
,,RankLasso") and extend the scope of its applications by proposing its
thresholded and adaptive versions. Our theoretical results show that these
modifications can identify the set of relevant predictors under much wider
range of data generating scenarios than regular RankLasso. Theoretical results
are supported by the simulation study and the real data analysis, which show
that our methods can properly identify relevant predictors, even when the error
terms come from the Cauchy distribution and the link function is nonlinear.
They also demonstrate the superiority of the modified versions of RankLasso in
the case when predictors are substantially correlated. The numerical study
shows also that RankLasso performs substantially better in model selection than
LADLasso, which is a well established methodology for robust model selection.
| stat |
Bayesian Analysis of Rank Data with Covariates and Heterogeneous Rankers | Data in the form of ranking lists are frequently encountered, and combining
ranking results from different sources can potentially generate a better
ranking list and help understand behaviors of the rankers. Of interest here are
the rank data under the following settings: (i) covariate information available
for the ranked entities; (ii) rankers of varying qualities or having different
opinions; and (iii) incomplete ranking lists for non-overlapping subgroups. We
review some key ideas built around the Thurstone model family by researchers in
the past few decades and provide a unifying approach for Bayesian Analysis of
Rank data with Covariates (BARC) and its extensions in handling heterogeneous
rankers. With this Bayesian framework, we can study rankers' varying quality,
cluster rankers' heterogeneous opinions, and measure the corresponding
uncertainties. To enable an efficient Bayesian inference, we advocate a
parameter-expanded Gibbs sampler to sample from the target posterior
distribution. The posterior samples also result in a Bayesian aggregated
ranking list, with credible intervals quantifying its uncertainty. We
investigate and compare performances of the proposed methods and other rank
aggregation methods in both simulation studies and two real-data examples.
| stat |
Learning drivers of climate-induced human migrations with Gaussian
processes | In the current context of climate change, extreme heatwaves, droughts, and
floods are not only impacting the biosphere and atmosphere but the
anthroposphere too. Human populations are forcibly displaced, which are now
referred to as climate-induced migrants. In this work, we investigate which
climate and structural factors forced major human displacements in the presence
of floods and storms during the years 2017-2019. We built, curated, and
harmonized a database of meteorological and remote sensing indicators along
with structural factors of 27 developing countries worldwide. We show how we
can use Gaussian Processes to learn what variables can explain the impact of
floods and storms in the context of forced displacements and to develop models
that reproduce migration flows. Our results at regional, global, and
disaster-specific scales show the importance of structural factors in the
determination of the magnitude of displacements. The study may have both
societal, political, and economical implications.
| stat |
Computational Limitations in Robust Classification and Win-Win Results | We continue the study of statistical/computational tradeoffs in learning
robust classifiers, following the recent work of Bubeck, Lee, Price and
Razenshteyn who showed examples of classification tasks where (a) an efficient
robust classifier exists, in the small-perturbation regime; (b) a non-robust
classifier can be learned efficiently; but (c) it is computationally hard to
learn a robust classifier, assuming the hardness of factoring large numbers.
The question of whether a robust classifier for their task exists in the large
perturbation regime seems related to important open questions in computational
number theory. In this work, we extend their work in three directions.
First, we demonstrate classification tasks where computationally efficient
robust classification is impossible, even when computationally unbounded robust
classifiers exist. For this, we rely on the existence of average-case hard
functions.
Second, we show hard-to-robustly-learn classification tasks in the
large-perturbation regime. Namely, we show that even though an efficient
classifier that is robust to large perturbations exists, it is computationally
hard to learn any non-trivial robust classifier. Our first construction relies
on the existence of one-way functions, and the second on the hardness of the
learning parity with noise problem. In the latter setting, not only does a
non-robust classifier exist, but also an efficient algorithm that generates
fresh new labeled samples given access to polynomially many training examples
(termed as generation by Kearns et. al. (1994)).
Third, we show that any such counterexample implies the existence of
cryptographic primitives such as one-way functions. This leads us to a win-win
scenario: either we can learn an efficient robust classifier, or we can
construct new instances of cryptographic primitives.
| stat |
Dependent relevance determination for smooth and structured sparse
regression | In many problem settings, parameter vectors are not merely sparse but
dependent in such a way that non-zero coefficients tend to cluster together. We
refer to this form of dependency as "region sparsity." Classical sparse
regression methods, such as the lasso and automatic relevance determination
(ARD), which model parameters as independent a priori, and therefore do not
exploit such dependencies. Here we introduce a hierarchical model for smooth,
region-sparse weight vectors and tensors in a linear regression setting. Our
approach represents a hierarchical extension of the relevance determination
framework, where we add a transformed Gaussian process to model the
dependencies between the prior variances of regression weights. We combine this
with a structured model of the prior variances of Fourier coefficients, which
eliminates unnecessary high frequencies. The resulting prior encourages weights
to be region-sparse in two different bases simultaneously. We develop Laplace
approximation and Monte Carlo Markov Chain (MCMC) sampling to provide efficient
inference for the posterior. Furthermore, a two-stage convex relaxation of the
Laplace approximation approach is also provided to relax the inevitable
non-convexity during the optimization. We finally show substantial improvements
over comparable methods for both simulated and real datasets from brain
imaging.
| stat |
Modeling Function-Valued Processes with Nonseparable and/or
Nonstationary Covariance Structure | We discuss a general Bayesian framework on modeling multidimensional
function-valued processes by using a Gaussian process or a heavy-tailed process
as a prior, enabling us to handle nonseparable and/or nonstationary covariance
structure. The nonstationarity is introduced by a convolution-based approach
through a varying anisotropy matrix, whose parameters vary along the input
space and are estimated via a local empirical Bayesian method. For the varying
matrix, we propose to use a spherical parametrization, leading to unconstrained
and interpretable parameters. The unconstrained nature allows the parameters to
be modeled as a nonparametric function of time, spatial location or other
covariates. The interpretation of the parameters is based on closed-form
expressions, providing valuable insights into nonseparable covariance
structures. Furthermore, to extract important information in data with complex
covariance structure, the Bayesian framework can decompose the function-valued
processes using the eigenvalues and eigensurfaces calculated from the estimated
covariance structure. The results are demonstrated by simulation studies and by
an application to wind intensity data. Supplementary materials for this article
are available online.
| stat |
On Proximal Causal Learning with Many Hidden Confounders | We generalize the proximal g-formula of Miao, Geng, and Tchetgen Tchetgen
(2018) for causal inference under unobserved confounding using proxy variables.
Specifically, we show that the formula holds true for all causal models in a
certain equivalence class, and this class contains models in which the total
number of levels for the set of unobserved confounders can be arbitrarily
larger than the number of levels of each proxy variable. Although
straightforward to obtain, the result can be significant for applications.
Simulations corroborate our formal arguments.
| stat |
An estimator for predictive regression: reliable inference for financial
economics | Estimating linear regression using least squares and reporting robust
standard errors is very common in financial economics, and indeed, much of the
social sciences and elsewhere. For thick tailed predictors under
heteroskedasticity this recipe for inference performs poorly, sometimes
dramatically so. Here, we develop an alternative approach which delivers an
unbiased, consistent and asymptotically normal estimator so long as the means
of the outcome and predictors are finite. The new method has standard errors
under heteroskedasticity which are easy to reliably estimate and tests which
are close to their nominal size. The procedure works well in simulations and in
an empirical exercise. An extension is given to quantile regression.
| stat |
P\'olygamma Data Augmentation to address Non-conjugacy in the Bayesian
Estimation of Mixed Multinomial Logit Models | The standard Gibbs sampler of Mixed Multinomial Logit (MMNL) models involves
sampling from conditional densities of utility parameters using
Metropolis-Hastings (MH) algorithm due to unavailability of conjugate prior for
logit kernel. To address this non-conjugacy concern, we propose the application
of P\'olygamma data augmentation (PG-DA) technique for the MMNL estimation. The
posterior estimates of the augmented and the default Gibbs sampler are similar
for two-alternative scenario (binary choice), but we encounter empirical
identification issues in the case of more alternatives ($J \geq 3$).
| stat |
Applicability and Interpretability of Hierarchical Agglomerative
Clustering With or Without Contiguity Constraints | Hierarchical Agglomerative Classification (HAC) with Ward's linkage has been
widely used since its introduction in Ward (1963). The present article reviews
the different extensions of the method to various input data and the
constrained framework, while providing applicability conditions. In addition,
various versions of the graphical representation of the results as a dendrogram
are also presented and their properties are clarified. While some of these
results can sometimes be found in an heteroclite literature, we clarify and
complete them all using a uniform background. In particular, this study reveals
an important distinction between a consistency property of the dendrogram and
the absence of crossover within it. Finally, a simulation study shows that the
constrained version of HAC can sometimes provide more relevant results than its
unconstrained version despite the fact that the latter optimizes the objective
criterion on a reduced set of solutions at each step. Overall, the article
provides comprehensive recommandations for the use of HAC and constrained HAC
depending on the input data as well as for the representation of the results.
| stat |
Learning Minimax Estimators via Online Learning | We consider the problem of designing minimax estimators for estimating the
parameters of a probability distribution. Unlike classical approaches such as
the MLE and minimum distance estimators, we consider an algorithmic approach
for constructing such estimators. We view the problem of designing minimax
estimators as finding a mixed strategy Nash equilibrium of a zero-sum game. By
leveraging recent results in online learning with non-convex losses, we provide
a general algorithm for finding a mixed-strategy Nash equilibrium of general
non-convex non-concave zero-sum games. Our algorithm requires access to two
subroutines: (a) one which outputs a Bayes estimator corresponding to a given
prior probability distribution, and (b) one which computes the worst-case risk
of any given estimator. Given access to these two subroutines, we show that our
algorithm outputs both a minimax estimator and a least favorable prior. To
demonstrate the power of this approach, we use it to construct provably minimax
estimators for classical problems such as estimation in the finite Gaussian
sequence model, and linear regression.
| stat |
A computationally efficient, high-dimensional multiple changepoint
procedure with application to global terrorism incidence | Detecting changepoints in datasets with many variates is a data science
challenge of increasing importance. Motivated by the problem of detecting
changes in the incidence of terrorism from a global terrorism database, we
propose a novel approach to multiple changepoint detection in multivariate time
series. Our method, which we call SUBSET, is a model-based approach which uses
a penalised likelihood to detect changes for a wide class of parametric
settings. We provide theory that guides the choice of penalties to use for
SUBSET, and that shows it has high power to detect changes regardless of
whether only a few variates or many variates change. Empirical results show
that SUBSET out-performs many existing approaches for detecting changes in mean
in Gaussian data; additionally, unlike these alternative methods, it can be
easily extended to non-Gaussian settings such as are appropriate for modelling
counts of terrorist events.
| stat |
Small increases in agent-based model complexity can result in large
increases in required calibration data | Agent-based models (ABMs) are widely used to model coupled natural-human
systems. Descriptive models require careful calibration with observed data.
However, ABMs are often not calibrated in a statistical sense. Here we examine
the impact of data record structure on the calibration of an ABM for housing
abandonment in the presence of flood risk. Using a perfect model experiment, we
examine the impact of data record structures on (i) model calibration and (ii)
the ability to distinguish a model with inter-agent interactions from one
without. We show how limited data sets may not constrain a model with just four
parameters. This indicates that many ABMs may require informative prior
distributions to be descriptive. We also illustrate how spatially-aggregated
data can be insufficient to identify the correct model structure. This
emphasizes the need for utilizing independent lines of evidence to select sound
and informative priors.
| stat |
A Bayesian Model for Bivariate Causal Inference | We address the problem of two-variable causal inference without intervention.
This task is to infer an existing causal relation between two random variables,
i.e. $X \rightarrow Y$ or $Y \rightarrow X$ , from purely observational data.
As the option to modify a potential cause is not given in many situations only
structural properties of the data can be used to solve this ill-posed problem.
We briefly review a number of state-of-the-art methods for this, including very
recent ones. A novel inference method is introduced, Bayesian Causal Inference
(BCI), which assumes a generative Bayesian hierarchical model to pursue the
strategy of Bayesian model selection. In the adopted model the distribution of
the cause variable is given by a Poisson lognormal distribution, which allows
to explicitly regard the discrete nature of datasets, correlations in the
parameter spaces, as well as the variance of probability densities on
logarithmic scales. We assume Fourier diagonal Field covariance operators. The
model itself is restricted to use cases where a direct causal relation $X
\rightarrow Y$ has to be decided against a relation $Y \rightarrow X$ ,
therefore we compare it other methods for this exact problem setting. The
generative model assumed provides synthetic causal data for benchmarking our
model in comparison to existing State-of-the-art models, namely LiNGAM ,
ANM-HSIC , ANM-MML , IGCI and CGNN . We explore how well the above methods
perform in case of high noise settings, strongly discretized data and very
sparse data. BCI performs generally reliable with synthetic data as well as
with the real world TCEP benchmark set, with an accuracy comparable to
state-of-the-art algorithms. We discuss directions for the future development
of BCI .
| stat |
Synthetic Dataset Generation of Driver Telematics | This article describes techniques employed in the production of a synthetic
dataset of driver telematics emulated from a similar real insurance dataset.
The synthetic dataset generated has 100,000 policies that included observations
about driver's claims experience together with associated classical risk
variables and telematics-related variables. This work is aimed to produce a
resource that can be used to advance models to assess risks for usage-based
insurance. It follows a three-stage process using machine learning algorithms.
The first stage is simulating values for the number of claims as multiple
binary classifications applying feedforward neural networks. The second stage
is simulating values for aggregated amount of claims as regression using
feedforward neural networks, with number of claims included in the set of
feature variables. In the final stage, a synthetic portfolio of the space of
feature variables is generated applying an extended $\texttt{SMOTE}$ algorithm.
The resulting dataset is evaluated by comparing the synthetic and real datasets
when Poisson and gamma regression models are fitted to the respective data.
Other visualization and data summarization produce remarkable similar
statistics between the two datasets. We hope that researchers interested in
obtaining telematics datasets to calibrate models or learning algorithms will
find our work valuable.
| stat |
Conservation of the $t$-digest Scale Invariant | A $t$-digest is a compact data structure that allows estimates of quantiles
which increased accuracy near $q = 0$ or $q=1$. This is done by clustering
samples from $\mathbb R$ subject to a constraint that the number of points
associated with any particular centroid is constrained so that the so-called
$k$-size of the centroid is always $\le 1$. The $k$-size is defined using a
scale function that maps quantile $q$ to index $k$. Since the centroids are
real numbers, they can be ordered and thus the quantile range of a centroid can
be mapped into an interval in $k$ whose size is the $k$-size of that centroid.
The accuracy of quantile estimates made using a $t$-digest depends on the
invariance of this constraint even as new data is added or $t$-digests are
merged. This paper provides proofs of this invariance for four practically
important scale functions.
| stat |
A Bayesian spatio-temporal nowcasting model for public health
decision-making and surveillance | As COVID-19 spread through the United States in 2020, states began to set up
alert systems to inform policy decisions and serve as risk communication tools
for the general public. Many of these systems, like in Ohio, included
indicators based on an assessment of trends in reported cases. However, when
cases are indexed by date of disease onset, reporting delays complicate the
interpretation of trends. Despite a foundation of statistical literature to
address this problem, these methods have not been widely applied in practice.
In this paper, we develop a Bayesian spatio-temporal nowcasting model for
assessing trends in county-level COVID-19 cases in Ohio. We compare the
performance of our model to the current approach used in Ohio and the approach
that was recommended by the Centers for Disease Control and Prevention. We
demonstrate gains in performance while still retaining interpretability using
our model. In addition, we are able to fully account for uncertainty in both
the time series of cases and in the reporting process. While we cannot
eliminate all of the uncertainty in public health surveillance and subsequent
decision-making, we must use approaches that embrace these challenges and
deliver more accurate and honest assessments to policymakers.
| stat |
An invitation to sequential Monte Carlo samplers | Sequential Monte Carlo samplers provide consistent approximations of
sequences of probability distributions and of their normalizing constants, via
particles obtained with a combination of importance weights and Markov
transitions. This article presents this class of methods and a number of recent
advances, with the goal of helping statisticians assess the applicability and
usefulness of these methods for their purposes. Our presentation emphasizes the
role of bridging distributions for computational and statistical purposes.
Numerical experiments are provided on simple settings such as multivariate
Normals, logistic regression and a basic susceptible-infected-recovered model,
illustrating the impact of the dimension, the ability to perform inference
sequentially and the estimation of normalizing constants.
| stat |
Cross-trait prediction accuracy of high-dimensional ridge-type
estimators in genome-wide association studies | Marginal association summary statistics have attracted great attention in
statistical genetics, mainly because the primary results of most genome-wide
association studies (GWAS) are produced by marginal screening. In this paper,
we study the prediction accuracy of marginal estimator in dense (or sparsity
free) high-dimensional settings with $(n,p,m) \to \infty$, $m/n \to \gamma \in
(0,\infty)$, and $p/n \to \omega \in (0,\infty)$. We consider a general
correlation structure among the $p$ features and allow an unknown subset $m$ of
them to be signals. As the marginal estimator can be viewed as a ridge
estimator with regularization parameter $\lambda \to \infty$, we further
investigate a class of ridge-type estimators in a unifying framework, including
the popular best linear unbiased prediction (BLUP) in genetics. We find that
the influence of $\lambda$ on out-of-sample prediction accuracy heavily depends
on $\omega$. Though selecting an optimal $\lambda$ can be important when $p$
and $n$ are comparable, it turns out that the out-of-sample $R^2$ of ridge-type
estimators becomes near-optimal for any $\lambda \in (0,\infty)$ as $\omega$
increases. For example, when features are independent, the out-of-sample $R^2$
is always bounded by $1/\omega$ from above and is largely invariant to
$\lambda$ given large $\omega$ (say, $\omega>5$). We also find that in-sample
$R^2$ has completely different patterns and depends much more on $\lambda$ than
out-of-sample $R^2$. In practice, our analysis delivers useful messages for
genome-wide polygenic risk prediction and computation-accuracy trade-off in
dense high-dimensions. We numerically illustrate our results in simulation
studies and a real data example.
| stat |
Inference on Breakdown Frontiers | Given a set of baseline assumptions, a breakdown frontier is the boundary
between the set of assumptions which lead to a specific conclusion and those
which do not. In a potential outcomes model with a binary treatment, we
consider two conclusions: First, that ATE is at least a specific value (e.g.,
nonnegative) and second that the proportion of units who benefit from treatment
is at least a specific value (e.g., at least 50\%). For these conclusions, we
derive the breakdown frontier for two kinds of assumptions: one which indexes
relaxations of the baseline random assignment of treatment assumption, and one
which indexes relaxations of the baseline rank invariance assumption. These
classes of assumptions nest both the point identifying assumptions of random
assignment and rank invariance and the opposite end of no constraints on
treatment selection or the dependence structure between potential outcomes.
This frontier provides a quantitative measure of robustness of conclusions to
relaxations of the baseline point identifying assumptions. We derive
$\sqrt{N}$-consistent sample analog estimators for these frontiers. We then
provide two asymptotically valid bootstrap procedures for constructing lower
uniform confidence bands for the breakdown frontier. As a measure of
robustness, estimated breakdown frontiers and their corresponding confidence
bands can be presented alongside traditional point estimates and confidence
intervals obtained under point identifying assumptions. We illustrate this
approach in an empirical application to the effect of child soldiering on
wages. We find that sufficiently weak conclusions are robust to simultaneous
failures of rank invariance and random assignment, while some stronger
conclusions are fairly robust to failures of rank invariance but not
necessarily to relaxations of random assignment.
| stat |
Prediction of Future Failures for Heterogeneous Reliability Field Data | This article introduces methods for constructing prediction bounds or
intervals for the number of future failures from heterogeneous reliability
field data. We focus on within-sample prediction where early data from a
failure-time process is used to predict future failures from the same process.
Early data from high-reliability products, however, often have limited
information due to some combination of small sample sizes, censoring, and
truncation. In such cases, we use a Bayesian hierarchical model to model
jointly multiple lifetime distributions arising from different subpopulations
of similar products. By borrowing information across subpopulations, our method
enables stable estimation and the computation of corresponding prediction
intervals, even in cases where there are few observed failures. Three
applications are provided to illustrate this methodology, and a simulation
study is used to validate the coverage performance of the prediction intervals.
| stat |
Fast robust correlation for high-dimensional data | The product moment covariance is a cornerstone of multivariate data analysis,
from which one can derive correlations, principal components, Mahalanobis
distances and many other results. Unfortunately the product moment covariance
and the corresponding Pearson correlation are very susceptible to outliers
(anomalies) in the data. Several robust measures of covariance have been
developed, but few are suitable for the ultrahigh dimensional data that are
becoming more prevalent nowadays. For that one needs methods whose computation
scales well with the dimension, are guaranteed to yield a positive semidefinite
covariance matrix, and are sufficiently robust to outliers as well as
sufficiently accurate in the statistical sense of low variability. We construct
such methods using data transformations. The resulting approach is simple, fast
and widely applicable. We study its robustness by deriving influence functions
and breakdown values, and computing the mean squared error on contaminated
data. Using these results we select a method that performs well overall. This
also allows us to construct a faster version of the DetectDeviatingCells method
(Rousseeuw and Van den Bossche, 2018) to detect cellwise outliers, that can
deal with much higher dimensions. The approach is illustrated on genomic data
with 12,000 variables and color video data with 920,000 dimensions.
| stat |
Quantifying the under-reporting of genital warts cases | Genital warts are a common and highly contagious sexually transmitted
disease. They have a large economic burden and affect several aspects of
quality of life. Incidence data underestimate the real occurrence of genital
warts because this infection is often under-reported, mostly due to their
specific characteristics such as the asymptomatic course. Genital warts cases
for the analysis were obtained from the catalan public health system database
(SIDIAP) for the period 2009-2016, covering 74\% of the Catalan population.
People under 15 and over 94 years old were excluded from the analysis as the
incidence of genital warts in this population is negligible. This work
introduces a time series model based on a mixture of two distributions, capable
of detecting the presence of under-reporting in the data. In order to identify
potential differences in the magnitude of the under-reporting issue depending
on sex and age, these covariates were included in the model. This work shows
that only about 80\% in average of genital warts incidence in Catalunya in the
period 2009-2016 was registered, although the frequency of under-reporting has
been decreasing over the study period. It can also be seen that the
under-reported issue has a deeper impact on women over 30 years old. The
registered incidence in the Catalan public health system is underestimating the
real burden in almost 10,000 cases in Catalunya, around 23\% of the registered
cases. The total annual cost in Catalunya is underestimated in at least about
10 million Euros respect the 54 million Euros annually devoted to genital warts
in Catalunya, representing 0.4\% of the total budget of the public health
system.
| stat |
Overlapping-sample Mendelian randomisation with multiple exposures: A
Bayesian approach | Background: Mendelian randomization (MR) has been widely applied to causal
inference in medical research. It uses genetic variants as instrumental
variables (IVs) to investigate putative causal relationship between an exposure
and an outcome. Traditional MR methods have dominantly focussed on a two-sample
setting in which IV-exposure association study and IV-outcome association study
are independent. However, it is not uncommon that participants from the two
studies fully overlap (one-sample) or partly overlap (overlapping-sample).
Methods: We proposed a method that is applicable to all the three sample
settings. In essence, we converted a two- or overlapping- sample problem to a
one-sample problem where data of some or all of the individuals were
incomplete. Assume that all individuals were drawn from the same population and
unmeasured data were missing at random. Then the unobserved data were treated
au pair with the model parameters as unknown quantities, and thus, could be
imputed iteratively conditioning on the observed data and estimated parameters
using Markov chain Monte Carlo. We generalised our model to allow for
pleiotropy and multiple exposures and assessed its performance by a number of
simulations using four metrics: mean, standard deviation, coverage and power.
Results: Higher sample overlapping rate and stronger instruments led to
estimates with higher precision and power. Pleiotropy had a notably negative
impact on the estimates. Nevertheless, overall the coverages were high and our
model performed well in all the sample settings. Conclusions: Our model offers
the flexibility of being applicable to any of the sample settings, which is an
important addition to the MR literature which has restricted to one- or two-
sample scenarios. Given the nature of Bayesian inference, it can be easily
extended to more complex MR analysis in medical research.
| stat |
Time-varying $\ell_0$ optimization for Spike Inference from Multi-Trial
Calcium Recordings | Optical imaging of genetically encoded calcium indicators is a powerful tool
to record the activity of a large number of neurons simultaneously over a long
period of time from freely behaving animals. However, determining the exact
time at which a neuron spikes and estimating the underlying firing rate from
calcium fluorescence data remains challenging, especially for calcium imaging
data obtained from a longitudinal study. We propose a multi-trial time-varying
$\ell_0$ penalized method to jointly detect spikes and estimate firing rates by
robustly integrating evolving neural dynamics across trials. Our simulation
study shows that the proposed method performs well in both spike detection and
firing rate estimation. We demonstrate the usefulness of our method on calcium
fluorescence trace data from two studies, with the first study showing
differential firing rate functions between two behaviors and the second study
showing evolving firing rate function across trials due to learning.
| stat |
Deep Latent Variable Model for Learning Longitudinal Multi-view Data | In many scientific problems such as video surveillance, modern genomic
analysis, and clinical studies, data are often collected from diverse domains
across time that exhibit time-dependent heterogeneous properties. It is
important to not only integrate data from multiple sources (called multiview
data), but also to incorporate time dependency for deep understanding of the
underlying system. Latent factor models are popular tools for exploring
multi-view data. However, it is frequently observed that these models do not
perform well for complex systems and they are not applicable to time-series
data. Therefore, we propose a generative model based on variational autoencoder
and recurrent neural network to infer the latent dynamic factors for
multivariate timeseries data. This approach allows us to identify the
disentangled latent embeddings across multiple modalities while accounting for
the time factor. We invoke our proposed model for analyzing three datasets on
which we demonstrate the effectiveness and the interpretability of the model.
| stat |
Linear Regression under Special Relativity | This study investigated the problem posed by using ordinary least squares
(OLS) to estimate parameters of simple linear regression under a specific
context of special relativity, where an independent variable is restricted to
an open interval, (-c, c). It is found that the OLS estimate for the slope
coefficient is not invariant under Lorentz velocity transformation.
Accordingly, an alternative estimator for the parameters of linear regression
under special relativity is proposed. This estimator can be considered a
generalization of the OLS estimator under special relativity; when c approaches
to infinity, the proposed estimator and its variance converges to the OLS
estimator and its variance, respectively. The variance of the proposed
estimator is larger than that of the OLS estimator, which implies that
hypothesis testing using the OLS estimator and its variance may result in a
liberal test under special relativity.
| stat |
Bayesian analysis of longitudinal studies with treatment by indication | It is often of interest in observational studies to measure the causal effect
of a treatment on time-to-event outcomes. In a medical setting, observational
studies commonly involve patients who initiate medication therapy and others
who do not, and the goal is to infer the effect of medication therapy on time
until recovery, a pre-defined level of improvement, or some other time-to-event
outcome. A difficulty with such studies is that the notion of a medication
initiation time does not exist in the control group. We propose an approach to
infer causal effects of an intervention in longitudinal observational studies
when the time of treatment assignment is only observed for treated units and
where treatment is given by indication. We present a framework for
conceptualizing an underlying randomized experiment in this setting based on
separating the process that governs the time of study arm assignment from the
mechanism that determines the assignment. Our approach involves inferring the
missing times of assignment followed by estimating treatment effects. This
approach allows us to incorporate uncertainty about the missing times of study
arm assignment, which induces uncertainty in both the selection of the control
group and the measurement of time-to-event outcomes for these controls. We
demonstrate our approach to study the effects on mortality of inappropriately
prescribing phosphodiesterase type 5 inhibitors (PDE5Is), a medication
contraindicated for groups 2 and 3 pulmonary hypertension, using administrative
data from the Veterans Affairs (VA) health care system.
| stat |
A mathematical take on the competitive balance of a football league | Competitive balance in a football league is extremely important from the
perspective of economic growth of the industry. Many researchers have earlier
proposed different measures of competitive balance, which are primarily adapted
from the standard economic theory. However, these measures fail to capture the
finer nuances of the game. In this work, we discuss a new framework which is
more suitable for a football league. First, we present a mathematical proof of
an ideal situation where a football league becomes perfectly balanced. Next, a
goal based index for competitive balance is developed. We present relevant
theoretical results and show how the proposed index can be used to formally
test for the presence of imbalance. The methods are implemented on the data
from top five European leagues, and it shows that the new approach can better
explain the changes in the seasonal competitive balance of the leagues.
Further, using appropriate panel data models, we show that the proposed index
is more suitable to analyze the variability in total revenues of the football
leagues.
| stat |
Deep learning for undersampled MRI reconstruction | This paper presents a deep learning method for faster magnetic resonance
imaging (MRI) by reducing k-space data with sub-Nyquist sampling strategies and
provides a rationale for why the proposed approach works well. Uniform
subsampling is used in the time-consuming phase-encoding direction to capture
high-resolution image information, while permitting the image-folding problem
dictated by the Poisson summation formula. To deal with the localization
uncertainty due to image folding, very few low-frequency k-space data are
added. Training the deep learning net involves input and output images that are
pairs of Fourier transforms of the subsampled and fully sampled k-space data.
Numerous experiments show the remarkable performance of the proposed method;
only 29% of k-space data can generate images of high quality as effectively as
standard MRI reconstruction with fully sampled data.
| stat |
Assessing the effectiveness of regional physical distancing measures of
COVID-19 in rural regions of British Columbia | We study the effects of physical distancing measures for the spread of
COVID-19 in regional areas within British Columbia, using the reported cases of
the five provincial Health Authorities. Building on the Bayesian
epidemiological model of Anderson et al. (2020), we propose a hierarchical
Bayesian model with time-varying regional parameters to account for the
relative reduction in contact due to physical distancing and increased testing
from March to December of 2020. In the absence of COVID-19 variants and
vaccinations during this period, we examine the regionalized basic reproduction
number, modelled prevalence, fraction of normal contacts, proportion of
anticipated cases, and we observed significant differences between the
provincial-wide and regional models.
| stat |
Lenient Regret and Good-Action Identification in Gaussian Process
Bandits | In this paper, we study the problem of Gaussian process (GP) bandits under
relaxed optimization criteria stating that any function value above a certain
threshold is "good enough". On the theoretical side, we study various
\emph{\lenient regret} notions in which all near-optimal actions incur zero
penalty, and provide upper bounds on the lenient regret for GP-UCB and an
elimination algorithm, circumventing the usual $O(\sqrt{T})$ term (with time
horizon $T$) resulting from zooming extremely close towards the function
maximum. In addition, we complement these upper bounds with
algorithm-independent lower bounds. On the practical side, we consider the
problem of finding a single "good action" according to a known pre-specified
threshold, and introduce several good-action identification algorithms that
exploit knowledge of the threshold. We experimentally find that such algorithms
can often find a good action faster than standard optimization-based
approaches.
| stat |
Evaluation of extremal properties of GARCH(p,q) processes | Generalized autoregressive conditionally heteroskedastic (GARCH) processes
are widely used for modelling features commonly found in observed financial
returns. The extremal properties of these processes are of considerable
interest for market risk management. For the simplest GARCH(p,q) process, with
max(p,q) = 1, all extremal features have been fully characterised. Although the
marginal features of extreme values of the process have been theoretically
characterised when max(p, q) >= 2, much remains to be found about both marginal
and dependence structure during extreme excursions. Specifically, a reliable
method is required for evaluating the tail index, which regulates the marginal
tail behaviour and there is a need for methods and algorithms for determining
clustering. In particular, for the latter, the mean number of extreme values in
a short-term cluster, i.e., the reciprocal of the extremal index, has only been
characterised in special cases which exclude all GARCH(p,q) processes that are
used in practice. Although recent research has identified the multivariate
regular variation property of stationary GARCH(p,q) processes, currently there
are no reliable methods for numerically evaluating key components of these
characterisations. We overcome these issues and are able to generate the
forward tail chain of the process to derive the extremal index and a range of
other cluster functionals for all GARCH(p, q) processes including integrated
GARCH processes and processes with unbounded and asymmetric innovations. The
new theory and methods we present extend to assessing the strict stationarity
and extremal properties for a much broader class of stochastic recurrence
equations.
| stat |
Causal Inference with Spatio-temporal Data: Estimating the Effects of
Airstrikes on Insurgent Violence in Iraq | Many causal processes have spatial and temporal dimensions. Yet the classic
causal inference framework is not directly applicable when the treatment and
outcome variables are generated by spatio-temporal processes with an infinite
number of possible event locations at each point in time. We take up the
challenge of extending the potential outcomes framework to these settings by
formulating the treatment point process as stochastic intervention. Our causal
estimands include the expected number of outcome events in a specified area of
interest under a particular stochastic treatment assignment strategy. We
develop an estimation technique that applies the inverse probability of
treatment weighting method to spatially-smoothed outcome surfaces. We
demonstrate that the proposed estimator is consistent and asymptotically normal
as the number of time period approaches infinity. A primary advantage of our
methodology is its ability to avoid structural assumptions about spatial
spillover and temporal carryover effects. We use the proposed methods to
estimate the effects of American airstrikes on insurgent violence in Iraq
(February 2007-July 2008). We find that increasing the average number of daily
airstrikes for up to one month increases insurgent attacks across Iraq and
within Baghdad. We also find evidence that airstrikes can displace attacks from
Baghdad to new locations up to 400 kilometers away.
| stat |
Path Analysis for Binary Random Variables | The decomposition of the overall effect of a treatment into direct and
indirect effects is here investigated with reference to a recursive system of
binary random variables. We show how, for the single mediator context, the
marginal effect measured on the log odds scale can be written as the sum of the
indirect and direct effects plus a residual term that vanishes under some
specific conditions. We then extend our definitions to situations involving
multiple mediators and address research questions concerning the decomposition
of the total effect when some mediators on the pathway from the treatment to
the outcome are marginalized over. Connections to the counterfactual
definitions of the effects are also made. Data coming from an encouragement
design on students' attitude to visit museums in Florence, Italy, are
reanalyzed. The estimates of the defined quantities are reported together with
their standard errors to compute p-values and form confidence intervals.
| stat |
Communication-Efficient Distributed Estimator for Generalized Linear
Models with a Diverging Number of Covariates | Distributed statistical inference has recently attracted immense attention.
The asymptotic efficiency of the maximum likelihood estimator (MLE), the
one-step MLE, and the aggregated estimating equation estimator are established
for generalized linear models under the "large $n$, diverging $p_n$" framework,
where the dimension of the covariates $p_n$ grows to infinity at a polynomial
rate $o(n^\alpha)$ for some $0<\alpha<1$. Then a novel method is proposed to
obtain an asymptotically efficient estimator for large-scale distributed data
by two rounds of communication. In this novel method, the assumption on the
number of servers is more relaxed and thus practical for real-world
applications. Simulations and a case study demonstrate the satisfactory
finite-sample performance of the proposed estimators.
| stat |
Mining geometric constraints from crowd-sourced radio signals and its
application to indoor positioning | Crowd-sourcing has become a promising way to build} a feature-based indoor
positioning system that has lower labour and time costs. It can make full use
of the widely deployed infrastructure as well as built-in sensors on mobile
devices. One of the key challenges is to generate the reference feature map
(RFM), a database used for localization, by {aligning crowd-sourced
{trajectories according to associations embodied in the data. In order to
facilitate the data fusion using crowd-sourced inertial sensors and radio
signals, this paper proposes an approach to adaptively mining geometric
information. This is the essential for generating spatial associations between
trajectories when employing graph-based optimization methods. The core idea is
to estimate the functional relationship to map the similarity/dissimilarity
between radio signals to the physical space based on the relative positions
obtained from inertial sensors and their associated radio signals. Namely, it
is adaptable to different modalities of data and can be implemented in a
self-supervised way. We verify the generality of the proposed approach through
comprehensive experimental analysis: i) qualitatively comparing the estimation
of geometric mapping models and the alignment of crowd-sourced trajectories;
ii) quantitatively evaluating the positioning performance. The 68\% of the
positioning error is less than 4.7 $\mathrm{m}$ using crowd-sourced RFM, which
is on a par with manually collected RFM, in a multi-storey shopping mall, which
covers more than 10, 000 $ \mathrm{m}^2 $.
| stat |
Computational Model for Urban Growth Using Socioeconomic Latent
Parameters | Land use land cover changes (LULCC) are generally modeled using multi-scale
spatio-temporal variables. Recently, Markov Chain (MC) has been used to model
LULCC. However, the model is derived from the proportion of LULCC observed over
a given period and it does not account for temporal factors such as
macro-economic, socio-economic, etc. In this paper, we present a richer model
based on Hidden Markov Model (HMM), grounded in the common knowledge that
economic, social and LULCC processes are tightly coupled. We propose a HMM
where LULCC classes represent hidden states and temporal fac-tors represent
emissions that are conditioned on the hidden states. To our knowledge, HMM has
not been used in LULCC models in the past. We further demonstrate its
integration with other spatio-temporal models such as Logistic Regression. The
integrated model is applied on the LULCC data of Pune district in the state of
Maharashtra (India) to predict and visualize urban LULCC over the past 14
years. We observe that the HMM integrated model has improved prediction
accuracy as compared to the corresponding MC integrated model
| stat |
Bayes-raking: Bayesian Finite Population Inference with Known Margins | Raking is widely used in categorical data modeling and survey practice but
faced with methodological and computational challenges. We develop a Bayesian
paradigm for raking by incorporating the marginal constraints as a prior
distribution via two main strategies: 1) constructing the solution subspaces
via basis functions or projection matrix and 2) modeling soft constraints. The
proposed Bayes-raking estimation integrates the models for the margins, the
sample selection and response mechanism, and the outcome, with the capability
to propagate all sources of uncertainty. Computation is done via Stan, and
codes are ready for public use. Simulation studies show that Bayes-raking can
perform as well as raking with large samples and outperform in terms of
validity and efficiency gains, especially with a sparse contingency table or
dependent raking factors. We apply the new method to the Longitudinal Study of
Wellbeing study and demonstrate that model-based approaches significantly
improve inferential reliability and substantive findings as a unified survey
inference framework.
| stat |
Model selection criteria for regression models with splines and the
automatic localization of knots | In this paper we propose a model selection approach to fit a regression model
using splines with a variable number of knots. We introduce a penalized
criterion to estimate the number and the position of the knots where to anchor
the splines basis. The method is evaluated on simulated data and applied to
covid-19 daily reported cases for short-term prediction.
| stat |
Bootstraps Regularize Singular Correlation Matrices | I show analytically that the average of $k$ bootstrapped correlation matrices
rapidly becomes positive-definite as $k$ increases, which provides a simple
approach to regularize singular Pearson correlation matrices. If $n$ is the
number of objects and $t$ the number of features, the averaged correlation
matrix is almost surely positive-definite if $k> \frac{e}{e-1}\frac{n}{t}\simeq
1.58\frac{n}{t}$ in the limit of large $t$ and $n$. The probability of
obtaining a positive-definite correlation matrix with $k$ bootstraps is also
derived for finite $n$ and $t$. Finally, I demonstrate that the number of
required bootstraps is always smaller than $n$. This method is particularly
relevant in fields where $n$ is orders of magnitude larger than the size of
data points $t$, e.g., in finance, genetics, social science, or image
processing.
| stat |
Beyond Backprop: Online Alternating Minimization with Auxiliary
Variables | Despite significant recent advances in deep neural networks, training them
remains a challenge due to the highly non-convex nature of the objective
function. State-of-the-art methods rely on error backpropagation, which suffers
from several well-known issues, such as vanishing and exploding gradients,
inability to handle non-differentiable nonlinearities and to parallelize
weight-updates across layers, and biological implausibility. These limitations
continue to motivate exploration of alternative training algorithms, including
several recently proposed auxiliary-variable methods which break the complex
nested objective function into local subproblems. However, those techniques are
mainly offline (batch), which limits their applicability to extremely large
datasets, as well as to online, continual or reinforcement learning. The main
contribution of our work is a novel online (stochastic/mini-batch) alternating
minimization (AM) approach for training deep neural networks, together with the
first theoretical convergence guarantees for AM in stochastic settings and
promising empirical results on a variety of architectures and datasets.
| stat |
Similarity-based Random Survival Forest | Predicting time-to-event outcomes in large databases can be a challenging but
important task. One example of this is in predicting the time to a clinical
outcome for patients in intensive care units (ICUs), which helps to support
critical medical treatment decisions. In this context, the time to an event of
interest could be, for example, survival time or time to recovery from a
disease/ailment observed within the ICU. The massive health datasets generated
from the uptake of Electronic Health Records (EHRs) are quite heterogeneous as
patients can be quite dissimilar in their relationship between the feature
vector and the outcome, adding more noise than information to prediction. In
this paper, we propose a modified random forest method for survival data that
identifies similar cases in an attempt to improve accuracy for predicting
time-to-event outcomes; this methodology can be applied in various settings,
including with ICU databases. We also introduce an adaptation of our
methodology in the case of dependent censoring. Our proposed method is
demonstrated in the Medical Information Mart for Intensive Care (MIMIC-III)
database, and, in addition, we present properties of our methodology through a
comprehensive simulation study. Introducing similarity to the random survival
forest method indeed provides improved predictive accuracy compared to random
survival forest alone across the various analyses we undertook.
| stat |
Adapting Auxiliary Losses Using Gradient Similarity | One approach to deal with the statistical inefficiency of neural networks is
to rely on auxiliary losses that help to build useful representations. However,
it is not always trivial to know if an auxiliary task will be helpful for the
main task and when it could start hurting. We propose to use the cosine
similarity between gradients of tasks as an adaptive weight to detect when an
auxiliary loss is helpful to the main loss. We show that our approach is
guaranteed to converge to critical points of the main task and demonstrate the
practical usefulness of the proposed algorithm in a few domains: multi-task
supervised learning on subsets of ImageNet, reinforcement learning on
gridworld, and reinforcement learning on Atari games.
| stat |
Utilizing Import Vector Machines to Identify Dangerous Pro-active
Traffic Conditions | Traffic accidents have been a severe issue in metropolises with the
development of traffic flow. This paper explores the theory and application of
a recently developed machine learning technique, namely Import Vector Machines
(IVMs), in real-time crash risk analysis, which is a hot topic to reduce
traffic accidents. Historical crash data and corresponding traffic data from
Shanghai Urban Expressway System were employed and matched. Traffic conditions
are labelled as dangerous (i.e. probably leading to a crash) and safe (i.e. a
normal traffic condition) based on 5-minute measurements of average speed,
volume and occupancy. The IVM algorithm is trained to build the classifier and
its performance is compared to the popular and successfully applied technique
of Support Vector Machines (SVMs). The main findings indicate that IVMs could
successfully be employed in real-time identification of dangerous pro-active
traffic conditions. Furthermore, similar to the "support points" of the SVM,
the IVM model uses only a fraction of the training data to index kernel basis
functions, typically a much smaller fraction than the SVM, and its
classification rates are similar to those of SVMs. This gives the IVM a
computational advantage over the SVM, especially when the size of the training
data set is large.
| stat |
Efficient Graph-Based Active Learning with Probit Likelihood via
Gaussian Approximations | We present a novel adaptation of active learning to graph-based
semi-supervised learning (SSL) under non-Gaussian Bayesian models. We present
an approximation of non-Gaussian distributions to adapt previously
Gaussian-based acquisition functions to these more general cases. We develop an
efficient rank-one update for applying "look-ahead" based methods as well as
model retraining. We also introduce a novel "model change" acquisition function
based on these approximations that further expands the available collection of
active learning acquisition functions for such methods.
| stat |
New ideas for method comparison: a Monte Carlo power analysis | In this paper some new proposals for method comparison are presented. On the
one hand, two new robust regressions, the M-Deming and the MM-Deming, have been
developed by modifying Linnet's method of the weighted Deming regression. The
M-Deming regression shows superior qualities to the Passing-Bablok regression;
it does not suffer from bias when the data to be validated have a reduced
precision, and therefore turns out to be much more reliable. On the other hand,
a graphical method (box and ellipses) for validations has been developed which
is also equipped with a unified statistical test. In this test the intercept
and slope pairs obtained from a bootstrap process are combined into a
multinomial distribution by robust determination of the covariance matrix. The
Mahalanobis distance from the point representing the null hypothesis is
evaluated using the $\chi^{2}$ distribution. It is emphasized that the
interpretation of the graph is more important than the probability obtained
from the test. The unified test has been evaluated through Monte Carlo
simulations, comparing the theoretical $\alpha$ levels with the empirical rate
of rejections (type-I errors). In addition, a power comparison of the various
(new and old) methods was conducted using the same techniques. This unified
method, regardless of the regression chosen, shows much higher power and allows
a significant reduction in the sample size required for validations.
| stat |
Stochastic modeling and estimation of COVID-19 population dynamics | The aim of the paper is to describe a model of the development of the
Covid-19 contamination of the population of a country or a region. For this
purpose a special branching process with two types of individuals is
considered. This model is intended to use only the observed daily statistics to
estimate the main parameter of the contamination and to give a prediction of
the mean value of the non-observed population of the contaminated individuals.
This is a serious advantage in comparison with other more complicated models
where the observed official statistics are not sufficient. In this way the
specific development of the Covid-19 epidemics is considered for different
countries.
| stat |
Boundary-free Kernel-smoothed Goodness-of-fit Tests for Data on General
Interval | We propose kernel-type smoothed Kolmogorov-Smirnov and Cram\'{e}r-von Mises
tests for data on general interval, using bijective transformations. Though not
as severe as in the kernel density estimation, utilizing naive kernel method
directly to those particular tests will result in boundary problem as well.
This happens mostly because the value of the naive kernel distribution function
estimator is still larger than $0$ (or less than $1$) when it is evaluated at
the boundary points. This situation can increase the errors of the tests
especially the second-type error. In this article, we use bijective
transformations to eliminate the boundary problem. Some simulation results
illustrating the estimator and the tests' performances will be presented in the
last part of this article.
| stat |
Meta-Learning for Koopman Spectral Analysis with Short Time-series | Koopman spectral analysis has attracted attention for nonlinear dynamical
systems since we can analyze nonlinear dynamics with a linear regime by
embedding data into a Koopman space by a nonlinear function. For the analysis,
we need to find appropriate embedding functions. Although several neural
network-based methods have been proposed for learning embedding functions,
existing methods require long time-series for training neural networks. This
limitation prohibits performing Koopman spectral analysis in applications where
only short time-series are available. In this paper, we propose a meta-learning
method for estimating embedding functions from unseen short time-series by
exploiting knowledge learned from related but different time-series. With the
proposed method, a representation of a given short time-series is obtained by a
bidirectional LSTM for extracting its properties. The embedding function of the
short time-series is modeled by a neural network that depends on the
time-series representation. By sharing the LSTM and neural networks across
multiple time-series, we can learn common knowledge from different time-series
while modeling time-series-specific embedding functions with the time-series
representation. Our model is trained such that the expected test prediction
error is minimized with the episodic training framework. We experimentally
demonstrate that the proposed method achieves better performance in terms of
eigenvalue estimation and future prediction than existing methods.
| stat |
Marginalised Gaussian Processes with Nested Sampling | Gaussian Process (GPs) models are a rich distribution over functions with
inductive biases controlled by a kernel function. Learning occurs through the
optimisation of kernel hyperparameters using the marginal likelihood as the
objective. This classical approach known as Type-II maximum likelihood (ML-II)
yields point estimates of the hyperparameters, and continues to be the default
method for training GPs. However, this approach risks underestimating
predictive uncertainty and is prone to overfitting especially when there are
many hyperparameters. Furthermore, gradient based optimisation makes ML-II
point estimates highly susceptible to the presence of local minima. This work
presents an alternative learning procedure where the hyperparameters of the
kernel function are marginalised using Nested Sampling (NS), a technique that
is well suited to sample from complex, multi-modal distributions. We focus on
regression tasks with the spectral mixture (SM) class of kernels and find that
a principled approach to quantifying model uncertainty leads to substantial
gains in predictive performance across a range of synthetic and benchmark data
sets. In this context, nested sampling is also found to offer a speed advantage
over Hamiltonian Monte Carlo (HMC), widely considered to be the gold-standard
in MCMC based inference.
| stat |
Locally induced Gaussian processes for large-scale simulation
experiments | Gaussian processes (GPs) serve as flexible surrogates for complex surfaces,
but buckle under the cubic cost of matrix decompositions with big training data
sizes. Geospatial and machine learning communities suggest pseudo-inputs, or
inducing points, as one strategy to obtain an approximation easing that
computational burden. However, we show how placement of inducing points and
their multitude can be thwarted by pathologies, especially in large-scale
dynamic response surface modeling tasks. As remedy, we suggest porting the
inducing point idea, which is usually applied globally, over to a more local
context where selection is both easier and faster. In this way, our proposed
methodology hybridizes global inducing point and data subset-based local GP
approximation. A cascade of strategies for planning the selection of local
inducing points is provided, and comparisons are drawn to related methodology
with emphasis on computer surrogate modeling applications. We show that local
inducing points extend their global and data-subset component parts on the
accuracy--computational efficiency frontier. Illustrative examples are provided
on benchmark data and a large-scale real-simulation satellite drag
interpolation problem.
| stat |
A Bayesian Hierarchical Model for Evaluating Forensic Footwear Evidence | When a latent shoeprint is discovered at a crime scene, forensic analysts
inspect it for distinctive patterns of wear such as scratches and holes (known
as accidentals) on the source shoe's sole. If its accidentals correspond to
those of a suspect's shoe, the print can be used as forensic evidence to place
the suspect at the crime scene. The strength of this evidence depends on the
random match probability---the chance that a shoe chosen at random would match
the crime scene print's accidentals. Evaluating random match probabilities
requires an accurate model for the spatial distribution of accidentals on shoe
soles. A recent report by the President's Council of Advisors in Science and
Technology criticized existing models in the literature, calling for new
empirically validated techniques. We respond to this request with a new spatial
point process model for accidental locations, developed within a hierarchical
Bayesian framework. We treat the tread pattern of each shoe as a covariate,
allowing us to pool information across large heterogeneous databases of shoes.
Existing models ignore this information; our results show that including it
leads to significantly better model fit. We demonstrate this by fitting our
model to one such database.
| stat |
Zero-inflated generalized extreme value regression model for binary data
and application in health study | Logistic regression model is widely used in many studies to investigate the
relationship between a binary response variable $Y$ and a set of potential
predictors $\mathbf X$. The binary response may represent, for example, the
occurrence of some outcome of interest ($Y=1$ if the outcome occurred and $Y=0$
otherwise). When the dependent variable $Y$ represents a rare event, the
logistic regression model shows relevant drawbacks. In order to overcome these
drawbacks we propose the Generalized Extreme Value (GEV) regression model. In
particular, we suggest the quantile function of the GEV distribution as link
function, so our attention is focused on the tail of the response curve for
values close to one. A sample of observations is said to contain a cure
fraction when a proportion of the study subjects (the so-called cured
individuals, as opposed to the susceptibles) cannot experience the outcome of
interest. One problem arising then is that it is usually unknown who are the
cured and the susceptible subjects, unless the outcome of interest has been
observed. In these settings, a logistic regression analysis of the relationship
between $\mathbf X$ and $Y$ among the susceptibles is no more straightforward.
We develop a maximum likelihood estimation procedure for this problem, based on
the joint modeling of the binary response of interest and the cure status. We
investigate the identifiability of the resulting model. Then, we conduct a
simulation study to investigate its finite-sample behavior, and application to
real data.
| stat |
FILDNE: A Framework for Incremental Learning of Dynamic Networks
Embeddings | Representation learning on graphs has emerged as a powerful mechanism to
automate feature vector generation for downstream machine learning tasks. The
advances in representation on graphs have centered on both homogeneous and
heterogeneous graphs, where the latter presenting the challenges associated
with multi-typed nodes and/or edges. In this paper, we consider the additional
challenge of evolving graphs. We ask the question of whether the advances in
representation learning for static graphs can be leveraged for dynamic graphs
and how? It is important to be able to incorporate those advances to maximize
the utility and generalization of methods. To that end, we propose the
Framework for Incremental Learning of Dynamic Networks Embedding (FILDNE),
which can utilize any existing static representation learning method for
learning node embeddings, while keeping the computational costs low. FILDNE
integrates the feature vectors computed using the standard methods over
different timesteps into a single representation by developing a convex
combination function and alignment mechanism. Experimental results on several
downstream tasks, over seven real-world data sets, show that FILDNE is able to
reduce memory and computational time costs while providing competitive quality
measure gains with respect to the contemporary methods for representation
learning on dynamic graphs.
| stat |
Simpson's Paradox and the implications for medical trials | This paper describes Simpson's paradox, and explains its serious implications
for randomised control trials. In particular, we show that for any number of
variables we can simulate the result of a controlled trial which uniformly
points to one conclusion (such as 'drug is effective') for every possible
combination of the variable states, but when a previously unobserved
confounding variable is included every possible combination of the variables
state points to the opposite conclusion ('drug is not effective'). In other
words no matter how many variables are considered, and no matter how
'conclusive' the result, one cannot conclude the result is truly 'valid' since
there is theoretically an unobserved confounding variable that could completely
reverse the result.
| stat |
Distributed Averaging Methods for Randomized Second Order Optimization | We consider distributed optimization problems where forming the Hessian is
computationally challenging and communication is a significant bottleneck. We
develop unbiased parameter averaging methods for randomized second order
optimization that employ sampling and sketching of the Hessian. Existing works
do not take the bias of the estimators into consideration, which limits their
application to massively parallel computation. We provide closed-form formulas
for regularization parameters and step sizes that provably minimize the bias
for sketched Newton directions. We also extend the framework of second order
averaging methods to introduce an unbiased distributed optimization framework
for heterogeneous computing systems with varying worker resources.
Additionally, we demonstrate the implications of our theoretical findings via
large scale experiments performed on a serverless computing platform.
| stat |
Gaussian Hierarchical Latent Dirichlet Allocation: Bringing Polysemy
Back | Topic models are widely used to discover the latent representation of a set
of documents. The two canonical models are latent Dirichlet allocation, and
Gaussian latent Dirichlet allocation, where the former uses multinomial
distributions over words, and the latter uses multivariate Gaussian
distributions over pre-trained word embedding vectors as the latent topic
representations, respectively. Compared with latent Dirichlet allocation,
Gaussian latent Dirichlet allocation is limited in the sense that it does not
capture the polysemy of a word such as ``bank.'' In this paper, we show that
Gaussian latent Dirichlet allocation could recover the ability to capture
polysemy by introducing a hierarchical structure in the set of topics that the
model can use to represent a given document. Our Gaussian hierarchical latent
Dirichlet allocation significantly improves polysemy detection compared with
Gaussian-based models and provides more parsimonious topic representations
compared with hierarchical latent Dirichlet allocation. Our extensive
quantitative experiments show that our model also achieves better topic
coherence and held-out document predictive accuracy over a wide range of corpus
and word embedding vectors.
| stat |
Neural Network Middle-Term Probabilistic Forecasting of Daily Power
Consumption | Middle-term horizon (months to a year) power consumption prediction is a main
challenge in the energy sector, in particular when probabilistic forecasting is
considered. We propose a new modelling approach that incorporates trend,
seasonality and weather conditions, as explicative variables in a shallow
Neural Network with an autoregressive feature. We obtain excellent results for
density forecast on the one-year test set applying it to the daily power
consumption in New England U.S.A.. The quality of the achieved power
consumption probabilistic forecasting has been verified, on the one hand,
comparing the results to other standard models for density forecasting and, on
the other hand, considering measures that are frequently used in the energy
sector as pinball loss and CI backtesting.
| stat |
Learning Randomly Perturbed Structured Predictors for Direct Loss
Minimization | Direct loss minimization is a popular approach for learning predictors over
structured label spaces. This approach is computationally appealing as it
replaces integration with optimization and allows to propagate gradients in a
deep net using loss-perturbed prediction. Recently, this technique was extended
to generative models, while introducing a randomized predictor that samples a
structure from a randomly perturbed score function. In this work, we learn the
variance of these randomized structured predictors and show that it balances
better between the learned score function and the randomized noise in
structured prediction. We demonstrate empirically the effectiveness of learning
the balance between the signal and the random noise in structured discrete
spaces.
| stat |
Analysis of Regularized Least Squares in Reproducing Kernel Krein Spaces | In this paper, we study the asymptotic properties of regularized least
squares with indefinite kernels in reproducing kernel Krein spaces (RKKS). By
introducing a bounded hyper-sphere constraint to such non-convex regularized
risk minimization problem, we theoretically demonstrate that this problem has a
globally optimal solution with a closed form on the sphere, which makes
approximation analysis feasible in RKKS. Regarding to the original regularizer
induced by the indefinite inner product, we modify traditional error
decomposition techniques, prove convergence results for the introduced
hypothesis error based on matrix perturbation theory, and derive learning rates
of such regularized regression problem in RKKS. Under some conditions, the
derived learning rates in RKKS are the same as that in reproducing kernel
Hilbert spaces (RKHS), which is actually the first work on approximation
analysis of regularized learning algorithms in RKKS.
| stat |
Local Interpretability of Calibrated Prediction Models: A Case of Type 2
Diabetes Mellitus Screening Test | Machine Learning (ML) models are often complex and difficult to interpret due
to their 'black-box' characteristics. Interpretability of a ML model is usually
defined as the degree to which a human can understand the cause of decisions
reached by a ML model. Interpretability is of extremely high importance in many
fields of healthcare due to high levels of risk related to decisions based on
ML models. Calibration of the ML model outputs is another issue often
overlooked in the application of ML models in practice. This paper represents
an early work in examination of prediction model calibration impact on the
interpretability of the results. We present a use case of a patient in diabetes
screening prediction scenario and visualize results using three different
techniques to demonstrate the differences between calibrated and uncalibrated
regularized regression model.
| stat |
P\'olya Urn Latent Dirichlet Allocation: a doubly sparse massively
parallel sampler | Latent Dirichlet Allocation (LDA) is a topic model widely used in natural
language processing and machine learning. Most approaches to training the model
rely on iterative algorithms, which makes it difficult to run LDA on big
corpora that are best analyzed in parallel and distributed computational
environments. Indeed, current approaches to parallel inference either don't
converge to the correct posterior or require storage of large dense matrices in
memory. We present a novel sampler that overcomes both problems, and we show
that this sampler is faster, both empirically and theoretically, than previous
Gibbs samplers for LDA. We do so by employing a novel P\'olya-urn-based
approximation in the sparse partially collapsed sampler for LDA. We prove that
the approximation error vanishes with data size, making our algorithm
asymptotically exact, a property of importance for large-scale topic models. In
addition, we show, via an explicit example, that - contrary to popular belief
in the topic modeling literature - partially collapsed samplers can be more
efficient than fully collapsed samplers. We conclude by comparing the
performance of our algorithm with that of other approaches on well-known
corpora.
| stat |
A Leisurely Look at Versions and Variants of the Cross Validation
Estimator | Many versions of cross-validation (CV) exist in the literature; and each
version though has different variants. All are used interchangeably by many
practitioners; yet, without explanation to the connection or difference among
them. This article has three contributions. First, it starts by mathematical
formalization of these different versions and variants that estimate the error
rate and the Area Under the ROC Curve (AUC) of a classification rule, to show
the connection and difference among them. Second, we prove some of their
properties and prove that many variants are either redundant or "not smooth".
Hence, we suggest to abandon all redundant versions and variants and only keep
the leave-one-out, the $K$-fold, and the repeated $K$-fold. We show that the
latter is the only among the three versions that is "smooth" and hence looks
mathematically like estimating the mean performance of the classification
rules. However, empirically, for the known phenomenon of "weak correlation",
which we explain mathematically and experimentally, it estimates both
conditional and mean performance almost with the same accuracy. Third, we
conclude the article with suggesting two research points that may answer the
remaining question of whether we can come up with a finalist among the three
estimators: (1) a comparative study, that is much more comprehensive than those
available in literature and conclude no overall winner, is needed to consider a
wide range of distributions, datasets, and classifiers including complex ones
obtained via the recent deep learning approach. (2) we sketch the path of
deriving a rigorous method for estimating the variance of the only "smooth"
version, repeated $K$-fold CV, rather than those ad-hoc methods available in
the literature that ignore the covariance structure among the folds of CV.
| stat |
Schr\"odinger Bridge Samplers | Consider a reference Markov process with initial distribution $\pi_{0}$ and
transition kernels $\{M_{t}\}_{t\in[1:T]}$, for some $T\in\mathbb{N}$. Assume
that you are given distribution $\pi_{T}$, which is not equal to the marginal
distribution of the reference process at time $T$. In this scenario,
Schr\"odinger addressed the problem of identifying the Markov process with
initial distribution $\pi_{0}$ and terminal distribution equal to $\pi_{T}$
which is the closest to the reference process in terms of Kullback--Leibler
divergence. This special case of the so-called Schr\"odinger bridge problem can
be solved using iterative proportional fitting, also known as the Sinkhorn
algorithm. We leverage these ideas to develop novel Monte Carlo schemes, termed
Schr\"odinger bridge samplers, to approximate a target distribution $\pi$ on
$\mathbb{R}^{d}$ and to estimate its normalizing constant. This is achieved by
iteratively modifying the transition kernels of the reference Markov chain to
obtain a process whose marginal distribution at time $T$ becomes closer to
$\pi_T = \pi$, via regression-based approximations of the corresponding
iterative proportional fitting recursion. We report preliminary experiments and
make connections with other problems arising in the optimal transport, optimal
control and physics literatures.
| stat |
Distance-distributed design for Gaussian process surrogates | A common challenge in computer experiments and related fields is to
efficiently explore the input space using a small number of samples, i.e., the
experimental design problem. Much of the recent focus in the computer
experiment literature, where modeling is often via Gaussian process (GP)
surrogates, has been on space-filling designs, via maximin distance, Latin
hypercube, etc. However, it is easy to demonstrate empirically that such
designs disappoint when the model hyperparameterization is unknown, and must be
estimated from data observed at the chosen design sites. This is true even when
the performance metric is prediction-based, or when the target of interest is
inherently or eventually sequential in nature, such as in blackbox (Bayesian)
optimization. Here we expose such inefficiencies, showing that in many cases
purely random design is superior to higher-powered alternatives. We then
propose a family of new schemes by reverse engineering the qualities of the
random designs which give the best estimates of GP lengthscales. Specifically,
we study the distribution of pairwise distances between design elements, and
develop a numerical scheme to optimize those distances for a given sample size
and dimension. We illustrate how our distance-based designs, and their hybrids
with more conventional space-filling schemes, outperform in both static
(one-shot design) and sequential settings.
| stat |
Ratio of counts vs ratio of rates in Poisson processes | The often debated issue of `ratios of small numbers of events' is approached
from a probabilistic perspective, making a clear distinction between the
predictive problem (forecasting numbers of events we might count under well
stated assumptions, and therefore of their ratios) and inferential problem
(learning about the relevant parameters of the related probability
distribution, in the light of the observed number of events). The quantities of
interests and their relations are visualized in a graphical model (`Bayesian
network'), very useful to understand how to approach the problem following the
rules of probability theory. In this paper, written with didactic intent, we
discuss in detail the basic ideas, however giving some hints of how real life
complications, like (uncertain) efficiencies and possible background and
systematics, can be included in the analysis, as well as the possibility that
the ratio of rates might depend on some physical quantity. The simple models
considered in this paper allow to obtain, under reasonable assumptions, closed
expressions for the rates and their ratios. Monte Carlo methods are also used,
both to cross check the exact results and to evaluate by sampling the ratios of
counts in the cases in which large number approximation does not hold. In
particular it is shown how to make approximate inferences using a Markov Chain
Monte Carlo using JAGS/rjags. Some examples of R and JAGS code are provided.
| stat |
Short-Term Covid-19 Forecast for Latecomers | The number of Covid-19 cases is increasing dramatically worldwide. Therefore,
the availability of reliable forecasts for the number of cases in the coming
days is of fundamental importance. We propose a simple statistical method for
short-term real-time forecasting of the number of Covid-19 cases and fatalities
in countries that are latecomers -- i.e., countries where cases of the disease
started to appear some time after others. In particular, we propose a penalized
(LASSO) regression with an error correction mechanism to construct a model of a
latecomer in terms of the other countries that were at a similar stage of the
pandemic some days before. By tracking the number of cases and deaths in those
countries, we forecast through an adaptive rolling-window scheme the number of
cases and deaths in the latecomer. We apply this methodology to Brazil, and
show that (so far) it has been performing very well. These forecasts aim to
foster a better short-run management of the health system capacity.
| stat |
Approximate Probabilistic Inference with Composed Flows | We study the problem of probabilistic inference on the joint distribution
defined by a normalizing flow model. Given a pre-trained flow model
$p(\boldsymbol{x})$, we wish to estimate $p(\boldsymbol{x}_2 \mid
\boldsymbol{x}_1)$ for some arbitrary partitioning of the variables
$\boldsymbol{x} = (\boldsymbol{x}_1, \boldsymbol{x}_2)$. We first show that
this task is computationally hard for a large class of flow models. Motivated
by this hardness result, we propose a framework for $\textit{approximate}$
probabilistic inference. Specifically, our method trains a new generative model
with the property that its composition with the given model approximates the
target conditional distribution. By parametrizing this new distribution as
another flow model, we can efficiently train it using variational inference and
also handle conditioning under arbitrary differentiable transformations. We
experimentally demonstrate that our approach outperforms Langevin Dynamics in
terms of sample quality, while requiring much fewer parameters and training
time compared to regular variational inference. We further validate the
flexibility of our method on a variety of inference tasks with applications to
inverse problems.
| stat |
Optimizing Approximate Leave-one-out Cross-validation to Tune
Hyperparameters | For a large class of regularized models, leave-one-out cross-validation can
be efficiently estimated with an approximate leave-one-out formula (ALO). We
consider the problem of adjusting hyperparameters so as to optimize ALO. We
derive efficient formulas to compute the gradient and hessian of ALO and show
how to apply a second-order optimizer to find hyperparameters. We demonstrate
the usefulness of the proposed approach by finding hyperparameters for
regularized logistic regression and ridge regression on various real-world data
sets.
| stat |
The Kendall Interaction Filter for Variable Interaction Screening in
Ultra High Dimensional Classification Problems | Accounting for important interaction effects can improve prediction of many
statistical learning models. Identification of relevant interactions, however,
is a challenging issue owing to their ultrahigh-dimensional nature. Interaction
screening strategies can alleviate such issues. However, due to heavier tail
distribution and complex dependence structure of interaction effects,
innovative robust and/or model-free methods for screening interactions are
required to better scale analysis of complex and high-throughput data. In this
work, we develop a new model-free interaction screening method, termed Kendall
Interaction Filter (KIF), for the classification in high-dimensional settings.
The KIF method suggests a weighted-sum measure, which compares the overall to
the within-cluster Kendall's $\tau$ of pairs of predictors, to select
interactive couples of features. The proposed KIF measure captures relevant
interactions for the clusters response-variable, handles continuous,
categorical or a mixture of continuous-categorical features, and is invariant
under monotonic transformations. We show that the KIF measure enjoys the sure
screening property in the high-dimensional setting under mild conditions,
without imposing sub-exponential moment assumptions on the features'
distributions. We illustrate the favorable behavior of the proposed methodology
compared to the methods in the same category using simulation studies, and we
conduct real data analyses to demonstrate its utility.
| stat |
Missing data analysis and imputation via latent Gaussian Markov random
fields | In this paper we recast the problem of missing values in the covariates of a
regression model as a latent Gaussian Markov random field (GMRF) model in a
fully Bayesian framework. Our proposed approach is based on the definition of
the covariate imputation sub-model as a latent effect with a GMRF structure. We
show how this formulation works for continuous covariates and provide some
insight on how this could be extended to categorical covariates.
The resulting Bayesian hierarchical model naturally fits within the
integrated nested Laplace approximation (INLA) framework, which we use for
model fitting. Hence, our work fills an important gap in the INLA methodology
as it allows to treat models with missing values in the covariates.
As in any other fully Bayesian framework, by relying on INLA for model
fitting it is possible to formulate a joint model for the data, the imputed
covariates and their missingness mechanism. In this way, we are able to tackle
the more general problem of assessing the missingness mechanism by conducting a
sensitivity analysis on the different alternatives to model the non-observed
covariates.
Finally, we illustrate the proposed approach with two examples on modeling
health risk factors and disease mapping. Here, we rely on two different
imputation mechanisms based on a typical multiple linear regression and a
spatial model, respectively. Given the speed of model fitting with INLA we are
able to fit joint models in a short time, and to easily conduct sensitivity
analyses.
| stat |
Projected Robust PCA with Application to Smooth Image Recovery | Most high-dimensional matrix recovery problems are studied under the
assumption that the target matrix has certain intrinsic structures. For image
data related matrix recovery problems, approximate low-rankness and smoothness
are the two most commonly imposed structures. For approximately low-rank matrix
recovery, the robust principal component analysis (PCA) is well-studied and
proved to be effective. For smooth matrix problem, 2d fused Lasso and other
total variation based approaches have played a fundamental role. Although both
low-rankness and smoothness are key assumptions for image data analysis, the
two lines of research, however, have very limited interaction. Motivated by
taking advantage of both features, we in this paper develop a framework named
projected robust PCA (PRPCA), under which the low-rank matrices are projected
onto a space of smooth matrices. Consequently, a large class of image matrices
can be decomposed as a low-rank and smooth component plus a sparse component. A
key advantage of this decomposition is that the dimension of the core low-rank
component can be significantly reduced. Consequently, our framework is able to
address a problematic bottleneck of many low-rank matrix problems: singular
value decomposition (SVD) on large matrices. Theoretically, we provide explicit
statistical recovery guarantees of PRPCA and include classical robust PCA as a
special case.
| stat |
Minimax Estimation of Distances on a Surface and Minimax Manifold
Learning in the Isometric-to-Convex Setting | We start by considering the problem of estimating intrinsic distances on a
smooth surface. We show that sharper estimates can be obtained via a
reconstruction of the surface, and discuss the use of the tangential Delaunay
complex for that purpose. We further show that the resulting approximation rate
is in fact optimal in an information-theoretic (minimax) sense. We then turn to
manifold learning and argue that a variant of Isomap where the distances are
instead computed on a reconstructed surface is minimax optimal for the problem
of isometric manifold embedding.
| stat |
Empirical Likelihood Weighted Estimation of Average Treatment Effects | There has been growing attention on how to effectively and objectively use
covariate information when the primary goal is to estimate the average
treatment effect (ATE) in randomized clinical trials (RCTs). In this paper, we
propose an effective weighting approach to extract covariate information based
on the empirical likelihood (EL) method. The resulting two-sample empirical
likelihood weighted (ELW) estimator includes two classes of weights, which are
obtained from a constrained empirical likelihood estimation procedure, where
the covariate information is effectively incorporated into the form of general
estimating equations. Furthermore, this ELW approach separates the estimation
of ATE from the analysis of the covariate-outcome relationship, which implies
that our approach maintains objectivity. In theory, we show that the proposed
ELW estimator is semiparametric efficient. We extend our estimator to tackle
the scenarios where the outcomes are missing at random (MAR), and prove the
double robustness and multiple robustness properties of our estimator.
Furthermore, we derive the semiparametric efficiency bound of all regular and
asymptotically linear semiparametric ATE estimators under MAR mechanism and
prove that our proposed estimator attains this bound. We conduct simulations to
make comparisons with other existing estimators, which confirm the efficiency
and multiple robustness property of our proposed ELW estimator. An application
to the AIDS Clinical Trials Group Protocol 175 (ACTG 175) data is conducted.
| stat |
Adversarial Networks and Autoencoders: The Primal-Dual Relationship and
Generalization Bounds | Since the introduction of Generative Adversarial Networks (GANs) and
Variational Autoencoders (VAE), the literature on generative modelling has
witnessed an overwhelming resurgence. The impressive, yet elusive empirical
performance of GANs has lead to the rise of many GAN-VAE hybrids, with the
hopes of GAN level performance and additional benefits of VAE, such as an
encoder for feature reduction, which is not offered by GANs. Recently, the
Wasserstein Autoencoder (WAE) was proposed, achieving performance similar to
that of GANs, yet it is still unclear whether the two are fundamentally
different or can be further improved into a unified model. In this work, we
study the $f$-GAN and WAE models and make two main discoveries. First, we find
that the $f$-GAN and WAE objectives partake in a primal-dual relationship and
are equivalent under some assumptions, which then allows us to explicate the
success of WAE. Second, the equivalence result allows us to, for the first
time, prove generalization bounds for Autoencoder models, which is a pertinent
problem when it comes to theoretical analyses of generative models.
Furthermore, we show that the WAE objective is related to other statistical
quantities such as the $f$-divergence and in particular, upper bounded by the
Wasserstein distance, which then allows us to tap into existing efficient
(regularized) optimal transport solvers. Our findings thus present the first
primal-dual relationship between GANs and Autoencoder models, comment on
generalization abilities and make a step towards unifying these models.
| stat |
Torus Graphs for Multivariate Phase Coupling Analysis | Angular measurements are often modeled as circular random variables, where
there are natural circular analogues of moments, including correlation. Because
a product of circles is a torus, a d-dimensional vector of circular random
variables lies on a d-dimensional torus. For such vectors we present here a
class of graphical models, which we call torus graphs, based on the full
exponential family with pairwise interactions. The topological distinction
between a torus and Euclidean space has several important consequences.
Our development was motivated by the problem of identifying phase coupling
among oscillatory signals recorded from multiple electrodes in the brain:
oscillatory phases across electrodes might tend to advance or recede together,
indicating coordination across brain areas. The data analyzed here consisted of
24 phase angles measured repeatedly across 840 experimental trials
(replications) during a memory task, where the electrodes were in 4 distinct
brain regions, all known to be active while memories are being stored or
retrieved. In realistic numerical simulations, we found that a standard
pairwise assessment, known as phase locking value, is unable to describe
multivariate phase interactions, but that torus graphs can accurately identify
conditional associations. Torus graphs generalize several more restrictive
approaches that have appeared in various scientific literatures, and produced
intuitive results in the data we analyzed. Torus graphs thus unify multivariate
analysis of circular data and present fertile territory for future research.
| stat |
Time Series Alignment with Global Invariances | In this work we address the problem of comparing time series while taking
into account both feature space transformation and temporal variability. The
proposed framework combines a latent global transformation of the feature space
with the widely used Dynamic Time Warping (DTW). The latent global
transformation captures the feature invariance while the DTW (or its smooth
counterpart soft-DTW) deals with the temporal shifts. We cast the problem as a
joint optimization over the global transformation and the temporal alignments.
The versatility of our framework allows for several variants depending on the
invariance class at stake. Among our contributions we define a differentiable
loss for time series and present two algorithms for the computation of time
series barycenters under our new geometry. We illustrate the interest of our
approach on both simulated and real world data.
| stat |
Gradient-only line searches to automatically determine learning rates
for a variety of stochastic training algorithms | Gradient-only and probabilistic line searches have recently reintroduced the
ability to adaptively determine learning rates in dynamic mini-batch
sub-sampled neural network training. However, stochastic line searches are
still in their infancy and thus call for an ongoing investigation. We study the
application of the Gradient-Only Line Search that is Inexact (GOLS-I) to
automatically determine the learning rate schedule for a selection of popular
neural network training algorithms, including NAG, Adagrad, Adadelta, Adam and
LBFGS, with numerous shallow, deep and convolutional neural network
architectures trained on different datasets with various loss functions. We
find that GOLS-I's learning rate schedules are competitive with manually tuned
learning rates, over seven optimization algorithms, three types of neural
network architecture, 23 datasets and two loss functions. We demonstrate that
algorithms, which include dominant momentum characteristics, are not well
suited to be used with GOLS-I. However, we find GOLS-I to be effective in
automatically determining learning rate schedules over 15 orders of magnitude,
for most popular neural network training algorithms, effectively removing the
need to tune the sensitive hyperparameters of learning rate schedules in neural
network training.
| stat |
On Linear Identifiability of Learned Representations | Identifiability is a desirable property of a statistical model: it implies
that the true model parameters may be estimated to any desired precision, given
sufficient computational resources and data. We study identifiability in the
context of representation learning: discovering nonlinear data representations
that are optimal with respect to some downstream task. When parameterized as
deep neural networks, such representation functions typically lack
identifiability in parameter space, because they are overparameterized by
design. In this paper, building on recent advances in nonlinear ICA, we aim to
rehabilitate identifiability by showing that a large family of discriminative
models are in fact identifiable in function space, up to a linear
indeterminacy. Many models for representation learning in a wide variety of
domains have been identifiable in this sense, including text, images and audio,
state-of-the-art at time of publication. We derive sufficient conditions for
linear identifiability and provide empirical support for the result on both
simulated and real-world data.
| stat |
CrimAnalyzer: Understanding Crime Patterns in S\~ao Paulo City | S\~ao Paulo is the largest city in South America, with high criminality
rates. The number and type of crimes varies considerably around the city,
assuming different patterns depending on urban and social characteristics. In
this scenario, enabling tools to explore particular locations of the city is
very important for domain experts to understand how urban features as to
mobility, passersby behavior, and urban infrastructures can influence the
quantity and type of crimes. In present work, we present CrimAnalyzer, a
visualization assisted analytic tool that allows users to analyze crime
behavior in specific regions of a city, providing new methodologies to identify
local crime hotspots and their corresponding patterns over time. CrimAnalyzer
has been developed from the demand of experts in criminology and it deals with
three major challenges: i) flexibility to explore local regions and understand
their crime patterns, ii) Identification of not only prevalent hotspots in
terms of number of crimes but also hotspots where crimes are frequent but not
in large amount, and iii) understand the dynamic of crime patterns over time.
The effectiveness and usefulness of the proposed system are demonstrated by
qualitative/quantitative comparisons as well as case studies involving real
data and run by domain experts.
| stat |
Integration of AI and mechanistic modeling in generative adversarial
networks for stochastic inverse problems | Stochastic inverse problems (SIP) address the behavior of a set of objects of
the same kind but with variable properties, such as a population of cells.
Using a population of mechanistic models from a single parametric family, SIP
explains population variability by transferring real-world observations into
the latent space of model parameters. Previous research in SIP focused on
solving the parameter inference problem for a single population using Markov
chain Monte Carlo methods. Here we extend SIP to address multiple related
populations simultaneously. Specifically, we simulate control and treatment
populations in experimental protocols by discovering two related latent spaces
of model parameters. Instead of taking a Bayesian approach, our two-population
SIP is reformulated as the constrained-optimization problem of finding
distributions of model parameters. To minimize the divergence between
distributions of experimental observations and model outputs, we developed
novel deep learning models based on generative adversarial networks (GANs)
which have the structure of our underlying constrained-optimization problem.
The flexibility of GANs allowed us to build computationally scalable solutions
and tackle complex model input parameter inference scenarios, which appear
routinely in physics, biophysics, economics and other areas, and which can not
be handled with existing methods. Specifically, we demonstrate two scenarios of
parameter inference over a control population and a treatment population whose
treatment either selectively affects only a subset of model parameters with
some uncertainty or has a deterministic effect on all model parameters.
| stat |
Depth based inference on conditional distribution with infinite
dimensional data | We develop inference and testing procedures for conditional dispersion and
skewness in a nonparametric regression setup based on statistical depth
functions. The methods developed can be applied in situations, where the
response is multivariate and the covariate is a random element in a metric
space. This includes regression with functional covariate as a special case. We
construct measures of the center, the spread and the skewness of the
conditional distribution of the response given the covariate using depth based
nonparametric regression procedures. We establish the asymptotic consistency of
those measures and develop a test for heteroscedasticity and a test for
conditional skewness. We present level and power study for the tests in several
simulated models. The usefulness of the methodology is also demonstrated in a
real dataset. In that dataset, our responses are the nutritional contents of
different meat samples measured by their protein, fat and moisture contents,
and the functional covariate is the absorbance spectra of the meat samples.
| stat |
Bootstrap Seasonal Unit Root Test under Periodic Variation | Both seasonal unit roots and periodic variation can be prevalent in seasonal
data. When testing seasonal unit roots under periodic variation, the validity
of the existing methods, such as the HEGY test, remains unknown. This paper
analyzes the behavior of the augmented HEGY test and the unaugmented HEGY test
under periodic variation. It turns out that the asymptotic null distributions
of the HEGY statistics testing the single roots at $1$ or $-1$ when there is
periodic variation are identical to the asymptotic null distributions when
there is no periodic variation. On the other hand, the asymptotic null
distributions of the statistics testing any coexistence of roots at $1$, $-1$,
$i$, or $-i$ when there is periodic variation are non-standard and are
different from the asymptotic null distributions when there is no periodic
variation. Therefore, when periodic variation exists, HEGY tests are not
directly applicable to the joint tests for any concurrence of seasonal unit
roots. As a remedy, bootstrap is proposed; in particular, the augmented HEGY
test with seasonal independent and identically distributed (iid) bootstrap and
the unaugmented HEGY test with seasonal block bootstrap are implemented. The
consistency of these bootstrap procedures is established. The finite-sample
behavior of these bootstrap tests is illustrated via simulation and prevails
over their competitors'. Finally, these bootstrap tests are applied to detect
the seasonal unit roots in various economic time series.
| stat |
Parameter estimation and model selection for water sorption in a wood
fibre material | The sorption curve is an essential feature for the modelling of heat and mass
transfer in porous building materials. Several models have been proposed in the
literature to represent the amount of moisture content in the material
according to the water activity (or capillary pressure) level. These models are
based on analytical expressions and few parameters that need to be estimated by
inverse analysis. This article investigates the reliability of eight models
through the accuracy of the estimated parameters. For this, experimental data
for a wood fibre material are generated with special attention to the stop
criterion to capture long time kinetic constants. Among five sets of
measurements, the best estimate is computed. The reliability of the models is
then discussed. After proving the theoretical identifiability of the unknown
parameters for each model, the primary identifiability is analysed. It
evaluates whether the parameters influence on the model output is sufficient to
proceed the parameter estimation with accuracy. For this, a continuous
derivative-based approach is adopted. Seven models have a low primary
identifiability for at least one parameter. Indeed, when estimating the unknown
parameters using the experimental observations, the parameters with low primary
identifiability exhibit large uncertainties. Finally, an Approximation Bayesian
Computation algorithm is used to simultaneously select the best model and
estimate the parameters that best represent the experimental data. The
thermodynamic and \textsc{Feng}--\textsc{Xing} models, together with a proposed
model in this work, were the best ones selected by this algorithm.
| stat |
For2For: Learning to forecast from forecasts | This paper presents a time series forecasting framework which combines
standard forecasting methods and a machine learning model. The inputs to the
machine learning model are not lagged values or regular time series features,
but instead forecasts produced by standard methods. The machine learning model
can be either a convolutional neural network model or a recurrent neural
network model. The intuition behind this approach is that forecasts of a time
series are themselves good features characterizing the series, especially when
the modelling purpose is forecasting. It can also be viewed as a weighted
ensemble method. Tested on the M4 competition dataset, this approach
outperforms all submissions for quarterly series, and is more accurate than all
but the winning algorithm for monthly series.
| stat |
A hybrid landmark Aalen-Johansen estimator for transition probabilities
in partially non-Markov multi-state models | Multi-state models are increasingly being used to model complex
epidemiological and clinical outcomes over time. It is common to assume that
the models are Markov, but the assumption can often be unrealistic. The Markov
assumption is seldomly checked and violations can lead to biased estimation for
many parameters of interest. As argued by Datta and Satten (2001), the
Aalen-Johansen estimator of occupation probabilities is consistent also in the
non-Markov case. Putter and Spitoni (2018) exploit this fact to construct a
consistent estimator of state transition probabilities, the landmark
Aalen-Johansen estimator, which does not rely on the Markov assumption. A
disadvantage of landmarking is data reduction, leading to a loss of power. This
is problematic for less traveled transitions, and undesirable when such
transitions indeed exhibit Markov behaviour. Using a framework of partially
non-Markov multi-state models we suggest a hybrid landmark Aalen-Johansen
estimator for transition probabilities. The proposed estimator is a compromise
between regular Aalen-Johansen and landmark estimation, using transition
specific landmarking, and can drastically improve statistical power. The
methods are compared in a simulation study and in a real data application
modelling individual transitions between states of sick leave, disability,
education, work and unemployment. In the application, a birth cohort of 184951
Norwegian men are followed for 14 years from the year they turn 21, using data
from national registries.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.