title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Robust estimation of the mean with bounded relative standard deviation | Many randomized approximation algorithms operate by giving a procedure for
simulating a random variable $X$ which has mean $\mu$ equal to the target
answer, and a relative standard deviation bounded above by a known constant
$c$. Examples of this type of algorithm includes methods for approximating the
number of satisfying assignments to 2-SAT or DNF, the volume of a convex body,
and the partition function of a Gibbs distribution. Because the answer is
usually exponentially large in the problem input size, it is typical to require
an estimate $\hat \mu$ satisfy $\mathbb{P}(|\hat \mu/\mu - 1| > \epsilon) \leq
\delta$, where $\epsilon$ and $\delta$ are user specified nonnegative
parameters. The current best algorithm uses $\lceil
2c^2\epsilon^{-2}(1+\epsilon)^2 \ln(2/\delta) \rceil$ samples to achieve such
an estimate. By modifying the algorithm in order to balance the tails, it is
possible to improve this result to $\lceil 2(c^2\epsilon^{-2} +
1)/(1-\epsilon^2)\ln(2/\delta) \rceil$ samples. Aside from the theoretical
improvement, we also consider how to best implement this algorithm in practice.
Numerical experiments show the behavior of the estimator on distributions where
the relative standard deviation is unknown or infinite.
| stat |
Outlier-Robust Optimal Transport | Optimal transport (OT) provides a way of measuring distances between
distributions that depends on the geometry of the sample space. In light of
recent advances in solving the OT problem, OT distances are widely used as loss
functions in minimum distance estimation. Despite its prevalence and
advantages, however, OT is extremely sensitive to outliers. A single
adversarially-picked outlier can increase OT distance arbitrarily. To address
this issue, in this work we propose an outlier-robust OT formulation. Our
formulation is convex but challenging to scale at a first glance. We proceed by
deriving an \emph{equivalent} formulation based on cost truncation that is easy
to incorporate into modern stochastic algorithms for regularized OT. We
demonstrate our model applied to mean estimation under the Huber contamination
model in simulation as well as outlier detection on real data.
| stat |
Generalized Score Distribution | A class of discrete probability distributions contains distributions with
limited support, i.e. possible argument values are limited to a set of numbers
(typically consecutive). Examples of such data are results from subjective
experiments utilizing the Absolute Category Rating (ACR) technique, where
possible answers (argument values) are $\{1, 2, \cdots, 5\}$ or typical Likert
scale $\{-3, -2, \cdots, 3\}$. An interesting subclass of those distributions
are distributions limited to two parameters: describing the mean value and the
spread of the answers, and having no more than one change in the probability
monotonicity. In this paper we propose a general distribution passing those
limitations called Generalized Score Distribution (GSD). The proposed GSD
covers all spreads of the answers, from very small, given by the Bernoulli
distribution, to the maximum given by a Beta Binomial distribution. We also
show that GSD correctly describes subjective experiments scores from video
quality evaluations with probability of 99.7\%. A Google Collaboratory website
with implementation of the GSD estimation, simulation, and visualization is
provided.
| stat |
Probabilistic symmetries and invariant neural networks | Treating neural network inputs and outputs as random variables, we
characterize the structure of neural networks that can be used to model data
that are invariant or equivariant under the action of a compact group. Much
recent research has been devoted to encoding invariance under symmetry
transformations into neural network architectures, in an effort to improve the
performance of deep neural networks in data-scarce, non-i.i.d., or unsupervised
settings. By considering group invariance from the perspective of probabilistic
symmetry, we establish a link between functional and probabilistic symmetry,
and obtain generative functional representations of probability distributions
that are invariant or equivariant under the action of a compact group. Our
representations completely characterize the structure of neural networks that
can be used to model such distributions and yield a general program for
constructing invariant stochastic or deterministic neural networks. We
demonstrate that examples from the recent literature are special cases, and
develop the details of the general program for exchangeable sequences and
arrays.
| stat |
Testing goodness-of-fit and conditional independence with approximate
co-sufficient sampling | Goodness-of-fit (GoF) testing is ubiquitous in statistics, with direct ties
to model selection, confidence interval construction, conditional independence
testing, and multiple testing, just to name a few applications. While testing
the GoF of a simple (point) null hypothesis provides an analyst great
flexibility in the choice of test statistic while still ensuring validity, most
GoF tests for composite null hypotheses are far more constrained, as the test
statistic must have a tractable distribution over the entire null model space.
A notable exception is co-sufficient sampling (CSS): resampling the data
conditional on a sufficient statistic for the null model guarantees valid GoF
testing using any test statistic the analyst chooses. But CSS testing requires
the null model to have a compact (in an information-theoretic sense) sufficient
statistic, which only holds for a very limited class of models; even for a null
model as simple as logistic regression, CSS testing is powerless. In this
paper, we leverage the concept of approximate sufficiency to generalize CSS
testing to essentially any parametric model with an asymptotically-efficient
estimator; we call our extension "approximate CSS" (aCSS) testing. We quantify
the finite-sample Type I error inflation of aCSS testing and show that it is
vanishing under standard maximum likelihood asymptotics, for any choice of test
statistic. We apply our proposed procedure both theoretically and in simulation
to a number of models of interest to demonstrate its finite-sample Type I error
and power.
| stat |
Approximate Bayesian inference in spatial environments | Model-based approaches bear great promise for decision making of agents
interacting with the physical world. In the context of spatial environments,
different types of problems such as localisation, mapping, navigation or
autonomous exploration are typically adressed with specialised methods, often
relying on detailed knowledge of the system at hand. We express these tasks as
probabilistic inference and planning under the umbrella of deep sequential
generative models. Using the frameworks of variational inference and neural
networks, our method inherits favourable properties such as flexibility,
scalability and the ability to learn from data. The method performs comparably
to specialised state-of-the-art methodology in two distinct simulated
environments.
| stat |
Benign Overfitting and Noisy Features | Modern machine learning often operates in the regime where the number of
parameters is much higher than the number of data points, with zero training
loss and yet good generalization, thereby contradicting the classical
bias-variance trade-off. This \textit{benign overfitting} phenomenon has
recently been characterized using so called \textit{double descent} curves
where the risk undergoes another descent (in addition to the classical U-shaped
learning curve when the number of parameters is small) as we increase the
number of parameters beyond a certain threshold. In this paper, we examine the
conditions under which \textit{Benign Overfitting} occurs in the random feature
(RF) models, i.e. in a two-layer neural network with fixed first layer weights.
We adopt a new view of random feature and show that \textit{benign overfitting}
arises due to the noise which resides in such features (the noise may already
be present in the data and propagate to the features or it may be added by the
user to the features directly) and plays an important implicit regularization
role in the phenomenon.
| stat |
Statistical Assessment of Replicability via Bayesian Model Criticism | Assessment of replicability is critical to ensure the quality and rigor of
scientific research. In this paper, we discuss inference and modeling
principles for replicability assessment. Targeting distinct application
scenarios, we propose two types of Bayesian model criticism approaches to
identify potentially irreproducible results in scientific experiments. They are
motivated by established Bayesian prior and posterior predictive model-checking
procedures and generalize many existing replicability assessment methods.
Finally, we discuss the statistical properties of the proposed replicability
assessment approaches and illustrate their usages by simulations and examples
of real data analysis, including the data from the Reproducibility Project:
Psychology and a systematic review of impacts of pre-existing cardiovascular
disease on COVID-19 outcomes.
| stat |
A P\'olya-Gamma Sampler for a Generalized Logistic Regression | In this paper we introduce a novel Bayesian data augmentation approach for
estimating the parameters of the generalised logistic regression model. We
propose a P\'olya-Gamma sampler algorithm that allows us to sample from the
exact posterior distribution, rather than relying on approximations. A
simulation study illustrates the flexibility and accuracy of the proposed
approach to capture heavy and light tails in binary response data of different
dimensions. The methodology is applied to two different real datasets, where we
demonstrate that the P\'olya-Gamma sampler provides more precise estimates than
the empirical likelihood method, outperforming approximate approaches.
| stat |
Adaptive deep density approximation for Fokker-Planck equations | In this paper we present a novel adaptive deep density approximation strategy
based on KRnet (ADDA-KR) for solving the steady-state Fokker-Planck equation.
It is known that this equation typically has high-dimensional spatial variables
posed on unbounded domains, which limit the application of traditional grid
based numerical methods. With the Knothe-Rosenblatt rearrangement, our newly
proposed flow-based generative model, called KRnet, provides a family of
probability density functions to serve as effective solution candidates of the
Fokker-Planck equation, which have weaker dependence on dimensionality than
traditional computational approaches. To result in effective stochastic
collocation points for training KRnet, we develop an adaptive sampling
procedure, where samples are generated iteratively using KRnet at each
iteration. In addition, we give a detailed discussion of KRnet and show that it
can efficiently estimate general high-dimensional density functions. We present
a general mathematical framework of ADDA-KR, validate its accuracy and
demonstrate its efficiency with numerical experiments.
| stat |
Modeling from Features: a Mean-field Framework for Over-parameterized
Deep Neural Networks | This paper proposes a new mean-field framework for over-parameterized deep
neural networks (DNNs), which can be used to analyze neural network training.
In this framework, a DNN is represented by probability measures and functions
over its features (that is, the function values of the hidden units over the
training data) in the continuous limit, instead of the neural network
parameters as most existing studies have done. This new representation
overcomes the degenerate situation where all the hidden units essentially have
only one meaningful hidden unit in each middle layer, and further leads to a
simpler representation of DNNs, for which the training objective can be
reformulated as a convex optimization problem via suitable re-parameterization.
Moreover, we construct a non-linear dynamics called neural feature flow, which
captures the evolution of an over-parameterized DNN trained by Gradient
Descent. We illustrate the framework via the standard DNN and the Residual
Network (Res-Net) architectures. Furthermore, we show, for Res-Net, when the
neural feature flow process converges, it reaches a global minimal solution
under suitable conditions. Our analysis leads to the first global convergence
proof for over-parameterized neural network training with more than $3$ layers
in the mean-field regime.
| stat |
Learning Adjustment Sets from Observational and Limited Experimental
Data | Estimating causal effects from observational data is not always possible due
to confounding. Identifying a set of appropriate covariates (adjustment set)
and adjusting for their influence can remove confounding bias; however, such a
set is typically not identifiable from observational data alone. Experimental
data do not have confounding bias, but are typically limited in sample size and
can therefore yield imprecise estimates. Furthermore, experimental data often
include a limited set of covariates, and therefore provide limited insight into
the causal structure of the underlying system. In this work we introduce a
method that combines large observational and limited experimental data to
identify adjustment sets and improve the estimation of causal effects. The
method identifies an adjustment set (if possible) by calculating the marginal
likelihood for the experimental data given observationally-derived prior
probabilities of potential adjustmen sets. In this way, the method can make
inferences that are not possible using only the conditional dependencies and
independencies in all the observational and experimental data. We show that the
method successfully identifies adjustment sets and improves causal effect
estimation in simulated data, and it can sometimes make additional inferences
when compared to state-of-the-art methods for combining experimental and
observational data.
| stat |
Exact Bayesian inference in spatio-temporal Cox processes driven by
multivariate Gaussian processes | In this paper we present a novel inference methodology to perform Bayesian
inference for spatiotemporal Cox processes where the intensity function depends
on a multivariate Gaussian process. Dynamic Gaussian processes are introduced
to allow for evolution of the intensity function over discrete time. The
novelty of the method lies on the fact that no discretisation error is involved
despite the non-tractability of the likelihood function and infinite
dimensionality of the problem. The method is based on a Markov chain Monte
Carlo algorithm that samples from the joint posterior distribution of the
parameters and latent variables of the model. The models are defined in a
general and flexible way but they are amenable to direct sampling from the
relevant distributions, due to careful characterisation of its components. The
models also allow for the inclusion of regression covariates and/or temporal
components to explain the variability of the intensity function. These
components may be subject to relevant interaction with space and/or time. Real
and simulated examples illustrate the methodology, followed by concluding
remarks.
| stat |
Gradient tree boosting with random output projections for multi-label
classification and multi-output regression | In many applications of supervised learning, multiple classification or
regression outputs have to be predicted jointly. We consider several extensions
of gradient boosting to address such problems. We first propose a
straightforward adaptation of gradient boosting exploiting multiple output
regression trees as base learners. We then argue that this method is only
expected to be optimal when the outputs are fully correlated, as it forces the
partitioning induced by the tree base learners to be shared by all outputs. We
then propose a novel extension of gradient tree boosting to specifically
address this issue. At each iteration of this new method, a regression tree
structure is grown to fit a single random projection of the current residuals
and the predictions of this tree are fitted linearly to the current residuals
of all the outputs, independently. Because of this linear fit, the method can
adapt automatically to any output correlation structure. Extensive experiments
are conducted with this method, as well as other algorithmic variants, on
several artificial and real problems. Randomly projecting the output space is
shown to provide a better adaptation to different output correlation patterns
and is therefore competitive with the best of the other methods in most
settings. Thanks to model sharing, the convergence speed is also improved,
reducing the computing times (or the complexity of the model) to reach a
specific accuracy.
| stat |
Prediction and Evaluation in College Hockey using the
Bradley-Terry-Zermelo Model | We describe the application of the Bradley-Terry model to NCAA Division I
Men's Ice Hockey. A Bayesian construction gives a joint posterior probability
distribution for the log-strength parameters, given a set of game results and a
choice of prior distribution. For several suitable choices of prior, it is
straightforward to find the maximum a posteriori point (MAP) and a Hessian
matrix, allowing a Gaussian approximation to be constructed. Posterior
predictive probabilities can be estimated by 1) setting the log-strengths to
their MAP values, 2) using the Gaussian approximation for analytical or Monte
Carlo integration, or 3) applying importance sampling to re-weight the results
of a Monte Carlo simulation. We define a method to evaluate any models which
generate predicted probabilities for future outcomes, using the Bayes factor
given the actual outcomes, and apply it to NCAA tournament results. Finally, we
describe an on-line tool which currently estimates probabilities of future
results using MAP evaluation and describe how it can be refined using the
Gaussian approximation or importance sampling.
| stat |
Running on empty: Recharge dynamics from animal movement data | Vital rates such as survival and recruitment have always been important in
the study of population and community ecology. At the individual level,
physiological processes such as energetics are critical in understanding
biomechanics and movement ecology and also scale up to influence food webs and
trophic cascades. Although vital rates and population-level characteristics are
tied with individual-level animal movement, most statistical models for
telemetry data are not equipped to provide inference about these relationships
because they lack the explicit, mechanistic connection to physiological
dynamics. We present a framework for modeling telemetry data that explicitly
includes an aggregated physiological process associated with decision making
and movement in heterogeneous environments. Our framework accommodates a wide
range of movement and physiological process specifications. We illustrate a
specific model formulation in continuous-time to provide direct inference about
gains and losses associated with physiological processes based on movement. Our
approach can also be extended to accommodate auxiliary data when available. We
demonstrate our model to infer mountain lion (in Colorado, USA) and African
buffalo (in Kruger National Park, South Africa) recharge dynamics.
| stat |
To BAN or not to BAN: Bayesian Attention Networks for Reliable Hate
Speech Detection | Hate speech is an important problem in the management of user-generated
content. To remove offensive content or ban misbehaving users, content
moderators need reliable hate speech detectors. Recently, deep neural networks
based on the transformer architecture, such as the (multilingual) BERT model,
achieve superior performance in many natural language classification tasks,
including hate speech detection. So far, these methods have not been able to
quantify their output in terms of reliability. We propose a Bayesian method
using Monte Carlo dropout within the attention layers of the transformer models
to provide well-calibrated reliability estimates. We evaluate and visualize the
results of the proposed approach on hate speech detection problems in several
languages. Additionally, we test if affective dimensions can enhance the
information extracted by the BERT model in hate speech classification. Our
experiments show that Monte Carlo dropout provides a viable mechanism for
reliability estimation in transformer networks. Used within the BERT model, it
ofers state-of-the-art classification performance and can detect less trusted
predictions. Also, it was observed that affective dimensions extracted using
sentic computing methods can provide insights toward interpretation of emotions
involved in hate speech. Our approach not only improves the classification
performance of the state-of-the-art multilingual BERT model but the computed
reliability scores also significantly reduce the workload in an inspection of
ofending cases and reannotation campaigns. The provided visualization helps to
understand the borderline outcomes.
| stat |
ICU Disparnumerophobia and Triskaidekaphobia: The 'Irrational Care
Unit'? | Whilst evidence-based medicine is the cornerstone of modern practice, it is
likely that clinicians are influenced by cultural biases. This work set out to
look for evidence of number preference in invasive mechanical ventilatory
therapy as a concrete example of subconscious treatment bias. A retrospective
observational intensive care electronic medical record database search and
analysis was carried out in adult general, specialist neurosciences and
paediatric intensive care units within a tertiary referral hospital. All
admitted, invasively mechanically ventilated patients between October 2014 and
August 2015 were included. Set positive end-expiratory pressure (PEEP),
respiratory rate (RR) and inspiratory pressure (Pinsp) settings were extracted.
Statistical analysis using conventional testing and a novel Monte Carlo method
were used to look for evidence of two culturally prevalent superstitions:
Odd/even preference and aversion to the number 13. Patients spent significantly
longer with odd choices for PEEP ($OR=0.16$, $p<2\times10^{-16}$), RR
($OR=0.31$, $p<2\times10^{-16}$) and Pinsp (OR=0.48, $p=2.9\times10^{-7}$). An
aversion to the number 13 was detected for choices of RR ($p=0.00024$) and
Pinsp ($p=3.9\times10^{-5}$). However a PEEP of 13 was more prevalent than
expected by chance ($p=0.00028$). These findings suggest superstitious
preferences in intensive care therapy do exist and practitioners should be
alert to guard against other, less obvious but perhaps more clinically
significant decision-making biases. The methodology described may be useful for
detecting statistically significant number preferences in other domains.
| stat |
Variational Inference with Continuously-Indexed Normalizing Flows | Continuously-indexed flows (CIFs) have recently achieved improvements over
baseline normalizing flows in a variety of density estimation tasks. In this
paper, we adapt CIFs to the task of variational inference (VI) through the
framework of auxiliary VI, and demonstrate that the advantages of CIFs over
baseline flows can also translate to the VI setting for both sampling from
posteriors with complicated topology and performing maximum likelihood
estimation in latent-variable models.
| stat |
Bayesian Estimation of Sparse Spiked Covariance Matrices in High
Dimensions | We propose a Bayesian methodology for estimating spiked covariance matrices
with jointly sparse structure in high dimensions. The spiked covariance matrix
is reparametrized in terms of the latent factor model, where the loading matrix
is equipped with a novel matrix spike-and-slab LASSO prior, which is a
continuous shrinkage prior for modeling jointly sparse matrices. We establish
the rate-optimal posterior contraction for the covariance matrix with respect
to the operator norm as well as that for the principal subspace with respect to
the projection operator norm loss. We also study the posterior contraction rate
of the principal subspace with respect to the two-to-infinity norm loss, a
novel loss function measuring the distance between subspaces that is able to
capture element-wise eigenvector perturbations. We show that the posterior
contraction rate with respect to the two-to-infinity norm loss is tighter than
that with respect to the routinely used projection operator norm loss under
certain low-rank and bounded coherence conditions. In addition, a point
estimator for the principal subspace is proposed with the rate-optimal risk
bound with respect to the projection operator norm loss. These results are
based on a collection of concentration and large deviation inequalities for the
matrix spike-and-slab LASSO prior. The numerical performance of the proposed
methodology is assessed through synthetic examples and the analysis of a
real-world face data example.
| stat |
Accounting for Location Measurement Error in Imaging Data with
Application to Atomic Resolution Images of Crystalline Materials | Scientists use imaging to identify objects of interest and infer properties
of these objects. The locations of these objects are often measured with error,
which when ignored leads to biased parameter estimates and inflated variance.
Current measurement error methods require an estimate or knowledge of the
measurement error variance to correct these estimates, which may not be
available. Instead, we create a spatial Bayesian hierarchical model that treats
the locations as parameters, it using the image itself to incorporate
positional uncertainty. We lower the computational burden by approximating the
likelihood using a non-contiguous block design around the object locations. We
apply this model in a materials science setting to study the relationship
between the chemistry and displacement of hundreds of atom columns in crystal
structures directly imaged via scanning transmission electron microscopy.
Greater knowledge of this relationship can lead to engineering materials with
improved properties of interest. We find strong evidence of a negative
relationship between atom column displacement and the intensity of neighboring
atom columns, which is related to the local chemistry. A simulation study shows
our method corrects the bias in the parameter of interest and drastically
improves coverage in high noise scenarios compared to non-measurement error
models.
| stat |
Multi-level conformal clustering: A distribution-free technique for
clustering and anomaly detection | In this work we present a clustering technique called \textit{multi-level
conformal clustering (MLCC)}. The technique is hierarchical in nature because
it can be performed at multiple significance levels which yields greater
insight into the data than performing it at just one level. We describe the
theoretical underpinnings of MLCC, compare and contrast it with the
hierarchical clustering algorithm, and then apply it to real world datasets to
assess its performance. There are several advantages to using MLCC over more
classical clustering techniques: Once a significance level has been set, MLCC
is able to automatically select the number of clusters. Furthermore, thanks to
the conformal prediction framework the resulting clustering model has a clear
statistical meaning without any assumptions about the distribution of the data.
This statistical robustness also allows us to perform clustering and anomaly
detection simultaneously. Moreover, due to the flexibility of the conformal
prediction framework, our algorithm can be used on top of many other machine
learning algorithms.
| stat |
Sequential testing for structural stability in approximate factor models | We develop a monitoring procedure to detect changes in a large approximate
factor model. Letting $r$ be the number of common factors, we base our
statistics on the fact that the $\left( r+1\right) $-th eigenvalue of the
sample covariance matrix is bounded under the null of no change, whereas it
becomes spiked under changes. Given that sample eigenvalues cannot be estimated
consistently under the null, we randomise the test statistic, obtaining a
sequence of \textit{i.i.d} statistics, which are used for the monitoring
scheme. Numerical evidence shows a very small probability of false detections,
and tight detection times of change-points.
| stat |
Incorporating Expert Prior Knowledge into Experimental Design via
Posterior Sampling | Scientific experiments are usually expensive due to complex experimental
preparation and processing. Experimental design is therefore involved with the
task of finding the optimal experimental input that results in the desirable
output by using as few experiments as possible. Experimenters can often acquire
the knowledge about the location of the global optimum. However, they do not
know how to exploit this knowledge to accelerate experimental design. In this
paper, we adopt the technique of Bayesian optimization for experimental design
since Bayesian optimization has established itself as an efficient tool for
optimizing expensive black-box functions. Again, it is unknown how to
incorporate the expert prior knowledge about the global optimum into Bayesian
optimization process. To address it, we represent the expert knowledge about
the global optimum via placing a prior distribution on it and we then derive
its posterior distribution. An efficient Bayesian optimization approach has
been proposed via posterior sampling on the posterior distribution of the
global optimum. We theoretically analyze the convergence of the proposed
algorithm and discuss the robustness of incorporating expert prior. We evaluate
the efficiency of our algorithm by optimizing synthetic functions and tuning
hyperparameters of classifiers along with a real-world experiment on the
synthesis of short polymer fiber. The results clearly demonstrate the
advantages of our proposed method.
| stat |
A Higher-Order Correct Fast Moving-Average Bootstrap for Dependent Data | We develop and implement a novel fast bootstrap for dependent data. Our
scheme is based on the i.i.d. resampling of the smoothed moment indicators. We
characterize the class of parametric and semi-parametric estimation problems
for which the method is valid. We show the asymptotic refinements of the
proposed procedure, proving that it is higher-order correct under mild
assumptions on the time series, the estimating functions, and the smoothing
kernel. We illustrate the applicability and the advantages of our procedure for
Generalized Empirical Likelihood estimation. As a by-product, our fast
bootstrap provides higher-order correct asymptotic confidence distributions.
Monte Carlo simulations on an autoregressive conditional duration model provide
numerical evidence that the novel bootstrap yields higher-order accurate
confidence intervals. A real-data application on dynamics of trading volume of
stocks illustrates the advantage of our method over the routinely-applied
first-order asymptotic theory, when the underlying distribution of the test
statistic is skewed or fat-tailed.
| stat |
Spatial disease mapping using Directed Acyclic Graph Auto-Regressive
(DAGAR) models | Hierarchical models for regionally aggregated disease incidence data commonly
involve region specific latent random effects that are modeled jointly as
having a multivariate Gaussian distribution. The covariance or precision matrix
incorporates the spatial dependence between the regions. Common choices for the
precision matrix include the widely used ICAR model, which is singular, and its
nonsingular extension which lacks interpretability. We propose a new parametric
model for the precision matrix based on a directed acyclic graph (DAG)
representation of the spatial dependence. Our model guarantees positive
definiteness and, hence, in addition to being a valid prior for regional
spatially correlated random effects, can also directly model the outcome from
dependent data like images and networks. Theoretical results establish a link
between the parameters in our model and the variance and covariances of the
random effects. Substantive simulation studies demonstrate that the improved
interpretability of our model reaps benefits in terms of accurately recovering
the latent spatial random effects as well as for inference on the spatial
covariance parameters. Under modest spatial correlation, our model far
outperforms the CAR models, while the performances are similar when the spatial
correlation is strong. We also assess sensitivity to the choice of the ordering
in the DAG construction using theoretical and empirical results which testify
to the robustness of our model. We also present a large-scale public health
application demonstrating the competitive performance of the model.
| stat |
Predictive Learning on Hidden Tree-Structured Ising Models | We provide high-probability sample complexity guarantees for exact structure
recovery and accurate predictive learning using noise-corrupted samples from an
acyclic (tree-shaped) graphical model. The hidden variables follow a
tree-structured Ising model distribution, whereas the observable variables are
generated by a binary symmetric channel taking the hidden variables as its
input (flipping each bit independently with some constant probability $q\in
[0,1/2)$). In the absence of noise, predictive learning on Ising models was
recently studied by Bresler and Karzand (2020); this paper quantifies how noise
in the hidden model impacts the tasks of structure recovery and marginal
distribution estimation by proving upper and lower bounds on the sample
complexity. Our results generalize state-of-the-art bounds reported in prior
work, and they exactly recover the noiseless case ($q=0$). In fact, for any
tree with $p$ vertices and probability of incorrect recovery $\delta>0$, the
sufficient number of samples remains logarithmic as in the noiseless case,
i.e., $\mathcal{O}(\log(p/\delta))$, while the dependence on $q$ is
$\mathcal{O}\big( 1/(1-2q)^{4} \big)$, for both aforementioned tasks. We also
present a new equivalent of Isserlis' Theorem for sign-valued tree-structured
distributions, yielding a new low-complexity algorithm for higher-order moment
estimation.
| stat |
UMAP: Uniform Manifold Approximation and Projection for Dimension
Reduction | UMAP (Uniform Manifold Approximation and Projection) is a novel manifold
learning technique for dimension reduction. UMAP is constructed from a
theoretical framework based in Riemannian geometry and algebraic topology. The
result is a practical scalable algorithm that applies to real world data. The
UMAP algorithm is competitive with t-SNE for visualization quality, and
arguably preserves more of the global structure with superior run time
performance. Furthermore, UMAP has no computational restrictions on embedding
dimension, making it viable as a general purpose dimension reduction technique
for machine learning.
| stat |
Scalable Modeling of Spatiotemporal Data using the Variational
Autoencoder: an Application in Glaucoma | As big spatial data becomes increasingly prevalent, classical spatiotemporal
(ST) methods often do not scale well. While methods have been developed to
account for high-dimensional spatial objects, the setting where there are
exceedingly large samples of spatial observations has had less attention. The
variational autoencoder (VAE), an unsupervised generative model based on deep
learning and approximate Bayesian inference, fills this void using a latent
variable specification that is inferred jointly across the large number of
samples. In this manuscript, we compare the performance of the VAE with a more
classical ST method when analyzing longitudinal visual fields from a large
cohort of patients in a prospective glaucoma study. Through simulation and a
case study, we demonstrate that the VAE is a scalable method for analyzing ST
data, when the goal is to obtain accurate predictions. R code to implement the
VAE can be found on GitHub: https://github.com/berchuck/vaeST.
| stat |
Avoiding Latent Variable Collapse With Generative Skip Models | Variational autoencoders learn distributions of high-dimensional data. They
model data with a deep latent-variable model and then fit the model by
maximizing a lower bound of the log marginal likelihood. VAEs can capture
complex distributions, but they can also suffer from an issue known as "latent
variable collapse," especially if the likelihood model is powerful.
Specifically, the lower bound involves an approximate posterior of the latent
variables; this posterior "collapses" when it is set equal to the prior, i.e.,
when the approximate posterior is independent of the data. While VAEs learn
good generative models, latent variable collapse prevents them from learning
useful representations. In this paper, we propose a simple new way to avoid
latent variable collapse by including skip connections in our generative model;
these connections enforce strong links between the latent variables and the
likelihood function. We study generative skip models both theoretically and
empirically. Theoretically, we prove that skip models increase the mutual
information between the observations and the inferred latent variables.
Empirically, we study images (MNIST and Omniglot) and text (Yahoo). Compared to
existing VAE architectures, we show that generative skip models maintain
similar predictive performance but lead to less collapse and provide more
meaningful representations of the data.
| stat |
Symmetry-Aware Actor-Critic for 3D Molecular Design | Automating molecular design using deep reinforcement learning (RL) has the
potential to greatly accelerate the search for novel materials. Despite recent
progress on leveraging graph representations to design molecules, such methods
are fundamentally limited by the lack of three-dimensional (3D) information. In
light of this, we propose a novel actor-critic architecture for 3D molecular
design that can generate molecular structures unattainable with previous
approaches. This is achieved by exploiting the symmetries of the design process
through a rotationally covariant state-action representation based on a
spherical harmonics series expansion. We demonstrate the benefits of our
approach on several 3D molecular design tasks, where we find that building in
such symmetries significantly improves generalization and the quality of
generated molecules.
| stat |
Coupled conditional backward sampling particle filter | The conditional particle filter (CPF) is a promising algorithm for general
hidden Markov model smoothing. Empirical evidence suggests that the variant of
CPF with backward sampling (CBPF) performs well even with long time series.
Previous theoretical results have not been able to demonstrate the improvement
brought by backward sampling, whereas we provide rates showing that CBPF can
remain effective with a fixed number of particles independent of the time
horizon. Our result is based on analysis of a new coupling of two CBPFs, the
coupled conditional backward sampling particle filter (CCBPF). We show that
CCBPF has good stability properties in the sense that with fixed number of
particles, the coupling time in terms of iterations increases only linearly
with respect to the time horizon under a general (strong mixing) condition. The
CCBPF is useful not only as a theoretical tool, but also as a practical method
that allows for unbiased estimation of smoothing expectations, following the
recent developments by Jacob et al. (to appear). Unbiased estimation has many
advantages, such as enabling the construction of asymptotically exact
confidence intervals and straightforward parallelisation.
| stat |
diproperm: An R Package for the DiProPerm Test | High-dimensional low sample size (HDLSS) data sets emerge frequently in many
biomedical applications. A common task for analyzing HDLSS data is to assign
data to the correct class using a classifier. Classifiers which use two labels
and a linear combination of features are known as binary linear classifiers.
The direction-projection-permutation (DiProPerm) test was developed for testing
the difference of two high-dimensional distributions induced by a binary linear
classifier. This paper discusses the key components of the DiProPerm test,
introduces the diproperm R package, and demonstrates the package on a
real-world data set.
| stat |
Diabetes Mellitus Forecasting Using Population Health Data in Ontario,
Canada | Leveraging health administrative data (HAD) datasets for predicting the risk
of chronic diseases including diabetes has gained a lot of attention in the
machine learning community recently. In this paper, we use the largest health
records datasets of patients in Ontario,Canada. Provided by the Institute of
Clinical Evaluative Sciences (ICES), this database is age, gender and
ethnicity-diverse. The datasets include demographics, lab measurements,drug
benefits, healthcare system interactions, ambulatory and hospitalizations
records. We perform one of the first large-scale machine learning studies with
this data to study the task of predicting diabetes in a range of 1-10 years
ahead, which requires no additional screening of individuals.In the best setup,
we reach a test AUC of 80.3 with a single-model trained on an observation
window of 5 years with a one-year buffer using all datasets. A subset of top 15
features alone (out of a total of 963) could provide a test AUC of 79.1. In
this paper, we provide extensive machine learning model performance and feature
contribution analysis, which enables us to narrow down to the most important
features useful for diabetes forecasting. Examples include chronic conditions
such as asthma and hypertension, lab results, diagnostic codes in insurance
claims, age and geographical information.
| stat |
Post-Processing of MCMC | Markov chain Monte Carlo (MCMC) is the engine of modern Bayesian statistics,
being used to approximate the posterior and derived quantities of interest.
Despite this, the issue of how the output from a Markov chain is post-processed
and reported is often overlooked. Convergence diagnostics can be used to
control bias via burn-in removal, but these do not account for (common)
situations where a limited computational budget engenders a bias-variance
trade-off. The aim of this article is to review state-of-the-art techniques for
post-processing Markov chain output. Our review covers methods based on
discrepancy minimisation, which directly address the bias-variance trade-off,
as well as general-purpose control variate methods for approximating expected
quantities of interest.
| stat |
An Alternating Manifold Proximal Gradient Method for Sparse PCA and
Sparse CCA | Sparse principal component analysis (PCA) and sparse canonical correlation
analysis (CCA) are two essential techniques from high-dimensional statistics
and machine learning for analyzing large-scale data. Both problems can be
formulated as an optimization problem with nonsmooth objective and nonconvex
constraints. Since non-smoothness and nonconvexity bring numerical
difficulties, most algorithms suggested in the literature either solve some
relaxations or are heuristic and lack convergence guarantees. In this paper, we
propose a new alternating manifold proximal gradient method to solve these two
high-dimensional problems and provide a unified convergence analysis. Numerical
experiment results are reported to demonstrate the advantages of our algorithm.
| stat |
Spatial deformation for non-stationary extremal dependence | Modelling the extremal dependence structure of spatial data is considerably
easier if that structure is stationary. However, for data observed over large
or complicated domains, non-stationarity will often prevail. Current methods
for modelling non-stationarity in extremal dependence rely on models that are
either computationally difficult to fit or require prior knowledge of
covariates. Sampson and Guttorp (1992) proposed a simple technique for handling
non-stationarity in spatial dependence by smoothly mapping the sampling
locations of the process from the original geographical space to a latent space
where stationarity can be reasonably assumed. We present an extension of this
method to a spatial extremes framework by considering least squares
minimisation of pairwise theoretical and empirical extremal dependence
measures. Along with some practical advice on applying these deformations, we
provide a detailed simulation study in which we propose three spatial processes
with varying degrees of non-stationarity in their extremal and central
dependence structures. The methodology is applied to Australian summer
temperature extremes and UK precipitation to illustrate its efficacy compared
to a naive modelling approach.
| stat |
Bayesian log-Gaussian Cox process regression: applications to
meta-analysis of neuroimaging working memory studies | Working memory (WM) was one of the first cognitive processes studied with
functional magnetic resonance imaging. With now over 20 years of studies on WM,
each study with tiny sample sizes, there is a need for meta-analysis to
identify the brain regions that are consistently activated by WM tasks, and to
understand the interstudy variation in those activations. However, current
methods in the field cannot fully account for the spatial nature of
neuroimaging meta-analysis data or the heterogeneity observed among WM studies.
In this work, we propose a fully Bayesian random-effects metaregression model
based on log-Gaussian Cox processes, which can be used for meta-analysis of
neuroimaging studies. An efficient Markov chain Monte Carlo scheme for
posterior simulations is presented which makes use of some recent advances in
parallel computing using graphics processing units. Application of the proposed
model to a real data set provides valuable insights regarding the function of
the WM.
| stat |
The Weight Function in the Subtree Kernel is Decisive | Tree data are ubiquitous because they model a large variety of situations,
e.g., the architecture of plants, the secondary structure of RNA, or the
hierarchy of XML files. Nevertheless, the analysis of these non-Euclidean data
is difficult per se. In this paper, we focus on the subtree kernel that is a
convolution kernel for tree data introduced by Vishwanathan and Smola in the
early 2000's. More precisely, we investigate the influence of the weight
function from a theoretical perspective and in real data applications. We
establish on a 2-classes stochastic model that the performance of the subtree
kernel is improved when the weight of leaves vanishes, which motivates the
definition of a new weight function, learned from the data and not fixed by the
user as usually done. To this end, we define a unified framework for computing
the subtree kernel from ordered or unordered trees, that is particularly
suitable for tuning parameters. We show through eight real data classification
problems the great efficiency of our approach, in particular for small
datasets, which also states the high importance of the weight function.
Finally, a visualization tool of the significant features is derived.
| stat |
Wasserstein Measure Coresets | The proliferation of large data sets and Bayesian inference techniques
motivates demand for better data sparsification. Coresets provide a principled
way of summarizing a large dataset via a smaller one that is guaranteed to
match the performance of the full data set on specific problems. Classical
coresets, however, neglect the underlying data distribution, which is often
continuous. We address this oversight by introducing Wasserstein measure
coresets, an extension of coresets which by definition takes into account
generalization. Our formulation of the problem, which essentially consists in
minimizing the Wasserstein distance, is solvable via stochastic gradient
descent. This yields an algorithm which simply requires sample access to the
data distribution and is able to handle large data streams in an online manner.
We validate our construction for inference and clustering.
| stat |
A multi-resolution approximation via linear projection for large spatial
datasets | Recent technical advances in collecting spatial data have been increasing the
demand for methods to analyze large spatial datasets. The statistical analysis
for these types of datasets can provide useful knowledge in various fields.
However, conventional spatial statistical methods, such as maximum likelihood
estimation and kriging, are impractically time-consuming for large spatial
datasets due to the necessary matrix inversions. To cope with this problem, we
propose a multi-resolution approximation via linear projection ($M$-RA-lp). The
$M$-RA-lp conducts a linear projection approach on each subregion whenever a
spatial domain is subdivided, which leads to an approximated covariance
function capturing both the large- and small-scale spatial variations.
Moreover, we elicit the algorithms for fast computation of the log-likelihood
function and predictive distribution with the approximated covariance function
obtained by the $M$-RA-lp. Simulation studies and a real data analysis for air
dose rates demonstrate that our proposed $M$-RA-lp works well relative to the
related existing methods.
| stat |
Assessing racial inequality in COVID-19 testing with Bayesian threshold
tests | There are racial disparities in the COVID-19 test positivity rate, suggesting
that minorities may be under-tested. Here, drawing on the literature on
statistically assessing racial disparities in policing, we 1) illuminate a
statistical flaw, known as infra-marginality, in using the positivity rate as a
metric for assessing racial disparities in under-testing; 2) develop a new type
of Bayesian threshold test to measure disparities in COVID-19 testing and 3)
apply the test to measure racial disparities in testing thresholds in a
real-world COVID-19 dataset.
| stat |
Simulating Execution Time of Tensor Programs using Graph Neural Networks | Optimizing the execution time of tensor program, e.g., a convolution,
involves finding its optimal configuration. Searching the configuration space
exhaustively is typically infeasible in practice. In line with recent research
using TVM, we propose to learn a surrogate model to overcome this issue. The
model is trained on an acyclic graph called an abstract syntax tree, and
utilizes a graph convolutional network to exploit structure in the graph. We
claim that a learnable graph-based data processing is a strong competitor to
heuristic-based feature extraction. We present a new dataset of graphs
corresponding to configurations and their execution time for various tensor
programs. We provide baselines for a runtime prediction task.
| stat |
Point Pattern Processes and Models | In recent years there has been a substantial increase in the availability of
datasets which contain information about the location and timing of an event or
group of events and the application of methods to analyse spatio-temporal
datasets spans many disciplines. This chapter defines and provides an overview
of tools for analysing spatial and temporal point patterns and processes, where
discrete events occur at random across space or over time respectively. It also
introduces the concept of spatial-temporal point patterns and methods of
analysis for data where events occur in both space and time. We discuss models,
methods and tools for analysing point processes.
| stat |
Neural Gaussian Mirror for Controlled Feature Selection in Neural
Networks | Deep neural networks (DNNs) have become increasingly popular and achieved
outstanding performance in predictive tasks. However, the DNN framework itself
cannot inform the user which features are more or less relevant for making the
prediction, which limits its applicability in many scientific fields. We
introduce neural Gaussian mirrors (NGMs), in which mirrored features are
created, via a structured perturbation based on a kernel-based conditional
dependence measure, to help evaluate feature importance. We design two
modifications of the DNN architecture for incorporating mirrored features and
providing mirror statistics to measure feature importance. As shown in
simulated and real data examples, the proposed method controls the feature
selection error rate at a predefined level and maintains a high selection power
even with the presence of highly correlated features.
| stat |
COVID-19 Data Analysis and Forecasting: Algeria and the World | The novel coronavirus disease 2019 COVID-19 has been leading the world into a
prominent crisis. As of May 19, 2020, the virus had spread to 215 countries
with more than 4,622,001 confirmed cases and 311,916 reported deaths worldwide,
including Algeria with 7201 cases and 555 deaths. Analyze and forecast COVID-19
cases and deaths growth could be useful in many ways, governments could
estimate medical equipment and take appropriate policy responses, and experts
could approximate the peak and the end of the disease. In this work, we first
train a time series Prophet model to analyze and forecast the number of
COVID-19 cases and deaths in Algeria based on the previously reported numbers.
Then, to better understand the spread and the properties of the COVID-19, we
include external factors that may contribute to accelerate/slow the spread of
the virus, construct a dataset from reliable sources, and conduct a large-scale
data analysis considering 82 countries worldwide. The evaluation results show
that the time series Prophet model accurately predicts the number of cases and
deaths in Algeria with low RMSE scores of 218.87 and 4.79 respectively, while
the forecast suggests that the total number of cases and deaths are expected to
increase in the coming weeks. Moreover, the worldwide data-driven analysis
reveals several correlations between the increase/decrease in the number of
cases and deaths and external factors that may contribute to accelerate/slow
the spread of the virus such as geographic, climatic, health, economic, and
demographic factors.
| stat |
Principled Selection of Baseline Covariates to Account for Censoring in
Randomized Trials with a Survival Endpoint | The analysis of randomized trials with time-to-event endpoints is nearly
always plagued by the problem of censoring. As the censoring mechanism is
usually unknown, analyses typically employ the assumption of non-informative
censoring. While this assumption usually becomes more plausible as more
baseline covariates are being adjusted for, such adjustment also raises
concerns. Pre-specification of which covariates will be adjusted for (and how)
is difficult, thus prompting the use of data-driven variable selection
procedures, which may impede valid inferences to be drawn. The adjustment for
covariates moreover adds concerns about model misspecification, and the fact
that each change in adjustment set, also changes the censoring assumption and
the treatment effect estimand. In this paper, we discuss these concerns and
propose a simple variable selection strategy that aims to produce a valid test
of the null in large samples. The proposal can be implemented using
off-the-shelf software for (penalized) Cox regression, and is empirically found
to work well in simulation studies and real data analyses.
| stat |
Filtering for Aggregate Hidden Markov Models with Continuous
Observations | We consider a class of filtering problems for large populations where each
individual is modeled by the same hidden Markov model (HMM). In this paper, we
focus on aggregate inference problems in HMMs with discrete state space and
continuous observation space. The continuous observations are aggregated in a
way such that the individuals are indistinguishable from measurements. We
propose an aggregate inference algorithm called continuous observation
collective forward-backward algorithm. It extends the recently proposed
collective forward-backward algorithm for aggregate inference in HMMs with
discrete observations to the case of continuous observations. The efficacy of
this algorithm is illustrated through several numerical experiments.
| stat |
Bayesian Spatial Homogeneity Pursuit Regression for Count Value Data | Spatial regression models are ubiquitous in many different areas such as
environmental science, geoscience, and public health. Exploring relationships
between response variables and covariates with complex spatial patterns is a
very important work. In this paper, we propose a novel spatially clustered
coefficients regression model for count value data based on nonparametric
Bayesian methods. Our proposed method detects the spatial homogeneity of the
Poisson regression coefficients. A Markov random field constraint mixture of
finite mixtures prior provides a consistent estimator of the number of the
clusters of regression coefficients with the geographically neighborhood
information. The theoretical properties of our proposed method are established.
An efficient Markov chain Monte Carlo algorithm is developed by using
multivariate log gamma distribution as a base distribution. Extensive
simulation studies are carried out to examine empirical performance of the
proposed method. Additionally, we analyze Georgia premature deaths data as an
illustration of the effectiveness of our approach.
| stat |
Vecchia approximations of Gaussian-process predictions | Gaussian processes (GPs) are highly flexible function estimators used for
geospatial analysis, nonparametric regression, and machine learning, but they
are computationally infeasible for large datasets. Vecchia approximations of
GPs have been used to enable fast evaluation of the likelihood for parameter
inference. Here, we study Vecchia approximations of spatial predictions at
observed and unobserved locations, including obtaining joint predictive
distributions at large sets of locations. We consider a general Vecchia
framework for GP predictions, which contains some novel and some existing
special cases. We study the accuracy and computational properties of these
approaches theoretically and numerically, proving that our new methods exhibit
linear computational complexity in the total number of spatial locations. We
show that certain choices within the framework can have a strong effect on
uncertainty quantification and computational cost, which leads to specific
recommendations on which methods are most suitable for various settings. We
also apply our methods to a satellite dataset of chlorophyll fluorescence,
showing that the new methods are faster or more accurate than existing methods,
and reduce unrealistic artifacts in prediction maps.
| stat |
Learning non-Gaussian graphical models via Hessian scores and triangular
transport | Undirected probabilistic graphical models represent the conditional
dependencies, or Markov properties, of a collection of random variables.
Knowing the sparsity of such a graphical model is valuable for modeling
multivariate distributions and for efficiently performing inference. While the
problem of learning graph structure from data has been studied extensively for
certain parametric families of distributions, most existing methods fail to
consistently recover the graph structure for non-Gaussian data. Here we propose
an algorithm for learning the Markov structure of continuous and non-Gaussian
distributions. To characterize conditional independence, we introduce a score
based on integrated Hessian information from the joint log-density, and we
prove that this score upper bounds the conditional mutual information for a
general class of distributions. To compute the score, our algorithm SING
estimates the density using a deterministic coupling, induced by a triangular
transport map, and iteratively exploits sparse structure in the map to reveal
sparsity in the graph. For certain non-Gaussian datasets, we show that our
algorithm recovers the graph structure even with a biased approximation to the
density. Among other examples, we apply sing to learn the dependencies between
the states of a chaotic dynamical system with local interactions.
| stat |
High-Dimensional Sparse Linear Bandits | Stochastic linear bandits with high-dimensional sparse features are a
practical model for a variety of domains, including personalized medicine and
online advertising. We derive a novel $\Omega(n^{2/3})$ dimension-free minimax
regret lower bound for sparse linear bandits in the data-poor regime where the
horizon is smaller than the ambient dimension and where the feature vectors
admit a well-conditioned exploration distribution. This is complemented by a
nearly matching upper bound for an explore-then-commit algorithm showing that
that $\Theta(n^{2/3})$ is the optimal rate in the data-poor regime. The results
complement existing bounds for the data-rich regime and provide another example
where carefully balancing the trade-off between information and regret is
necessary. Finally, we prove a dimension-free $O(\sqrt{n})$ regret upper bound
under an additional assumption on the magnitude of the signal for relevant
features.
| stat |
Local Regularization of Noisy Point Clouds: Improved Global Geometric
Estimates and Data Analysis | Several data analysis techniques employ similarity relationships between data
points to uncover the intrinsic dimension and geometric structure of the
underlying data-generating mechanism. In this paper we work under the model
assumption that the data is made of random perturbations of feature vectors
lying on a low-dimensional manifold. We study two questions: how to define the
similarity relationship over noisy data points, and what is the resulting
impact of the choice of similarity in the extraction of global geometric
information from the underlying manifold. We provide concrete mathematical
evidence that using a local regularization of the noisy data to define the
similarity improves the approximation of the hidden Euclidean distance between
unperturbed points. Furthermore, graph-based objects constructed with the
locally regularized similarity function satisfy better error bounds in their
recovery of global geometric ones. Our theory is supported by numerical
experiments that demonstrate that the gain in geometric understanding
facilitated by local regularization translates into a gain in classification
accuracy in simulated and real data.
| stat |
Score matching for compositional distributions | Compositional data and multivariate count data with known totals are
challenging to analyse due to the non-negativity and sum-to-one constraints on
the sample space. It is often the case that many of the compositional
components are highly right-skewed, with large numbers of zeros. A major
limitation of currently available estimators for compositional models is that
they either cannot handle many zeros in the data or are not computationally
feasible in moderate to high dimensions. We derive a new set of novel score
matching estimators applicable to distributions on a Riemannian manifold with
boundary, of which the standard simplex is a special case. The score matching
method is applied to estimate the parameters in a new flexible truncation model
for compositional data and we show that the estimators are scalable and
available in closed form. Through extensive simulation studies, the scoring
methodology is demonstrated to work well for estimating the parameters in the
new truncation model and also for the Dirichlet distribution. We apply the new
model and estimators to real microbiome compositional data and show that the
model provides a good fit to the data.
| stat |
Predictive Risk Analysis in Collective Risk Model: Choices between
Historical Frequency and Aggregate Severity | Typical risk classification procedure in insurance is consists of a priori
risk classification determined by observable risk characteristics, and a
posteriori risk classification where the premium is adjusted to reflect the
policyholder's claim history. While using the full claim history data is
optimal in a posteriori risk classification procedure, i.e. giving premium
estimators with the minimal variances, some insurance sectors, however, only
use partial information of the claim history for determining the appropriate
premium to charge. Classical examples include that auto insurances premium are
determined by the claim frequency data and workers' compensation insurances are
based on the aggregate severity. The motivation for such practice is to have a
simplified and efficient posteriori risk classification procedure which is
customized to the involved insurance policy. This paper compares the relative
efficiency of the two simplified posteriori risk classifications, i.e. based on
frequency versus severity, and provides the mathematical framework to assist
practitioners in choosing the most appropriate practice.
| stat |
Functional Autoregressive Processes in Reproducing Kernel Hilbert Spaces | We study the estimation and prediction of functional autoregressive~(FAR)
processes, a statistical tool for modeling functional time series data. Due to
the infinite-dimensional nature of FAR processes, the existing literature
addresses its inference via dimension reduction and theoretical results therein
require the (unrealistic) assumption of fully observed functional time series.
We propose an alternative inference framework based on Reproducing Kernel
Hilbert Spaces~(RKHS). Specifically, a nuclear norm regularization method is
proposed for estimating the transition operators of the FAR process directly
from discrete samples of the functional time series. We derive a representer
theorem for the FAR process, which enables infinite-dimensional inference
without dimension reduction. Sharp theoretical guarantees are established under
the (more realistic) assumption that we only have finite discrete samples of
the FAR process. Extensive numerical experiments and a real data application of
energy consumption prediction are further conducted to illustrate the promising
performance of the proposed approach compared to the state-of-the-art methods
in the literature.
| stat |
Central Quantile Subspace | Quantile regression (QR) is becoming increasingly popular due to its
relevance in many scientific investigations. There is a great amount of work
about linear and nonlinear QR models. Specifically, nonparametric estimation of
the conditional quantiles received particular attention, due to its model
flexibility. However, nonparametric QR techniques are limited in the number of
covariates. Dimension reduction offers a solution to this problem by
considering low-dimensional smoothing without specifying any parametric or
nonparametric regression relation. Existing dimension reduction techniques
focus on the entire conditional distribution. We, on the other hand, turn our
attention to dimension reduction techniques for conditional quantiles and
introduce a new method for reducing the dimension of the predictor X. The
novelty of this paper is threefold. We start by considering a single index
quantile regression model, which assumes that the conditional quantile depends
on X through a single linear combination of the predictors, then extend to a
multi index quantile regression model, and finally, generalize the proposed
methodology to any statistical functional of the conditional distribution. The
performance of the methodology is demonstrated through simulation examples and
a real data application. Our results suggest that this method has a good finite
sample performance and often outperforms existing methods.
| stat |
Estimating the Standard Error of Cross-Validation-Based Estimators of
Classification Rules Performance | First, we analyze the variance of the Cross Validation (CV)-based estimators
used for estimating the performance of classification rules. Second, we propose
a novel estimator to estimate this variance using the Influence Function (IF)
approach that had been used previously very successfully to estimate the
variance of the bootstrap-based estimators. The motivation for this research is
that, as the best of our knowledge, the literature lacks a rigorous method for
estimating the variance of the CV-based estimators. What is available is a set
of ad-hoc procedures that have no mathematical foundation since they ignore the
covariance structure among dependent random variables. The conducted
experiments show that the IF proposed method has small RMS error with some
bias. However, surprisingly, the ad-hoc methods still work better than the
IF-based method. Unfortunately, this is due to the lack of enough smoothness if
compared to the bootstrap estimator. This opens the research for three points:
(1) more comprehensive simulation study to clarify when the IF method win or
loose; (2) more mathematical analysis to figure out why the ad-hoc methods work
well; and (3) more mathematical treatment to figure out the connection between
the appropriate amount of "smoothness" and decreasing the bias of the IF
method.
| stat |
Predicting the Number of Future Events | This paper describes prediction methods for the number of future events from
a population of units associated with an on-going time-to-event process.
Examples include the prediction of warranty returns and the prediction of the
number of future product failures that could cause serious threats to property
or life. Important decisions such as whether a product recall should be
mandated are often based on such predictions. Data, generally right-censored
(and sometimes left truncated and right-censored), are used to estimate the
parameters of a time-to-event distribution. This distribution can then be used
to predict the number of events over future periods of time. Such predictions
are sometimes called within-sample predictions and differ from other prediction
problems considered in most of the prediction literature. This paper shows that
the plug-in (also known as estimative or naive) prediction method is not
asymptotically correct (i.e., for large amounts of data, the coverage
probability always fails to converge to the nominal confidence level). However,
a commonly used prediction calibration method is shown to be asymptotically
correct for within-sample predictions, and two alternative
predictive-distributionbased methods that perform better than the calibration
method are presented and justified.
| stat |
Estimating Multiple Precision Matrices with Cluster Fusion
Regularization | We propose a penalized likelihood framework for estimating multiple precision
matrices from different classes. Most existing methods either incorporate no
information on relationships between the precision matrices, or require this
information be known a priori. The framework proposed in this article allows
for simultaneous estimation of the precision matrices and relationships between
the precision matrices, jointly. Sparse and non-sparse estimators are proposed,
both of which require solving a non-convex optimization problem. To compute our
proposed estimators, we use an iterative algorithm which alternates between a
convex optimization problem solved by blockwise coordinate descent and a
k-means clustering problem. Blockwise updates for computing the sparse
estimator require solving an elastic net penalized precision matrix estimation
problem, which we solve using a proximal gradient descent algorithm. We prove
that this subalgorithm has a linear rate of convergence. In simulation studies
and two real data applications, we show that our method can outperform
competitors that ignore relevant relationships between precision matrices and
performs similarly to methods which use prior information often uknown in
practice.
| stat |
Long-Term Implications of the Revenue Transfer Methodology in the
Affordable Care Act | The Affordable Care Act introduced a revenue transfer formula that requires
insurance plans with generally healthier enrollees to pay funds into a revenue
transfer pool for to reimburse plans with generally less healthy enrollees. For
a given plan, the issue arises of whether the plan will be a payer into or a
receiver from the pool in a chosen future year. To examine that issue, we
analyze data from The Actuary Magazine on transfer payments for 2014-2015, and
we infer strong evidence of a statistical relationship between year-to-year
transfer payments. We also apply to the data a Markov transition model to study
annual changes in the payer-receiver statuses of insurance plans. We estimate
that the limiting conditional probability that an insurance plan will pay into
the pool, given that the plan had paid into the pool in 2014, is 55.6 percent.
Further, that limiting probability is attained quickly because the conditional
probability that an insurance plan will pay into the pool in 2024, given that
the plan had paid into the pool in 2014, is estimated to be 55.7 percent. We
also find the revenue transfer system to have the disturbing feature that once
a plan enters the "state" of paying into the pool then it will stay in that
state for an average period of 4.87 years; moreover, once a plan has received
funds from the pool then it will stay in that state for an average period of
3.89 years.
| stat |
SURF: A Simple, Universal, Robust, Fast Distribution Learning Algorithm | Sample- and computationally-efficient distribution estimation is a
fundamental tenet in statistics and machine learning. We present SURF, an
algorithm for approximating distributions by piecewise polynomials. SURF is:
simple, replacing prior complex optimization techniques by straight-forward
{empirical probability} approximation of each potential polynomial piece
{through simple empirical-probability interpolation}, and using plain
divide-and-conquer to merge the pieces; universal, as well-known
polynomial-approximation results imply that it accurately approximates a large
class of common distributions; robust to distribution mis-specification as for
any degree $d \le 8$, it estimates any distribution to an $\ell_1$ distance $<
3$ times that of the nearest degree-$d$ piecewise polynomial, improving known
factor upper bounds of 3 for single polynomials and 15 for polynomials with
arbitrarily many pieces; fast, using optimal sample complexity, running in near
sample-linear time, and if given sorted samples it may be parallelized to run
in sub-linear time. In experiments, SURF outperforms state-of-the art
algorithms.
| stat |
Transferability of Operational Status Classification Models Among
Different Wind Turbine Typesq | A detailed understanding of wind turbine performance status classification
can improve operations and maintenance in the wind energy industry. Due to
different engineering properties of wind turbines, the standard supervised
learning models used for classification do not generalize across data sets
obtained from different wind sites. We propose two methods to deal with the
transferability of the trained models: first, data normalization in the form of
power curve alignment, and second, a robust method based on convolutional
neural networks and feature-space extension. We demonstrate the success of our
methods on real-world data sets with industrial applications.
| stat |
Supervised PCA: A Multiobjective Approach | Methods for supervised principal component analysis (SPCA) aim to incorporate
label information into principal component analysis (PCA), so that the
extracted features are more useful for a prediction task of interest. Prior
work on SPCA has focused primarily on optimizing prediction error, and has
neglected the value of maximizing variance explained by the extracted features.
We propose a new method for SPCA that addresses both of these objectives
jointly, and demonstrate empirically that our approach dominates existing
approaches, i.e., outperforms them with respect to both prediction error and
variation explained. Our approach accommodates arbitrary supervised learning
losses and, through a statistical reformulation, provides a novel low-rank
extension of generalized linear models.
| stat |
Revisiting Identifying Assumptions for Population Size Estimation | The problem of estimating the size of a population based on a subset of
individuals observed across multiple data sources is often referred to as
capture-recapture or multiple-systems estimation. This is fundamentally a
missing data problem, where the number of unobserved individuals represents the
missing data. As with any missing data problem, multiple-systems estimation
requires users to make an untestable identifying assumption in order to
estimate the population size from the observed data. Approaches to
multiple-systems estimation often do not emphasize the role of the identifying
assumption during model specification, which makes it difficult to decouple the
specification of the model for the observed data from the identifying
assumption. We present a re-framing of the multiple-systems estimation problem
that decouples the specification of the observed-data model from the
identifying assumptions, and discuss how log-linear models and the associated
no-highest-order interaction assumption fit into this framing. We present an
approach to computation in the Bayesian setting which takes advantage of
existing software and facilitates various sensitivity analyses. We demonstrate
our approach in a case study of estimating the number of civilian casualties in
the Kosovo war. Code used to produce this manuscript is available at
https://github.com/aleshing/revisiting-identifying-assumptions.
| stat |
Identifying Precipitation Regimes in China Using Model-Based Clustering
of Spatial Functional Data | The identification of precipitation regimes is important for many purposes
such as agricultural planning, water resource management, and return period
estimation. Since precipitation and other related meteorological data typically
exhibit spatial dependency and different characteristics at different time
scales, clustering such data presents unique challenges. In this paper, we
develop a flexible model-based approach to cluster multi-scale spatial
functional data to address such problems. The underlying clustering model is a
functional linear model , and the cluster memberships are assumed to be a
realization from a Markov random field with geographic covariates. The
methodology is applied to a precipitation data from China to identify
precipitation regimes.
| stat |
The Promises and Pitfalls of Deep Kernel Learning | Deep kernel learning and related techniques promise to combine the
representational power of neural networks with the reliable uncertainty
estimates of Gaussian processes. One crucial aspect of these models is an
expectation that, because they are treated as Gaussian process models optimized
using the marginal likelihood, they are protected from overfitting. However, we
identify pathological behavior, including overfitting, on a simple toy example.
We explore this pathology, explaining its origins and considering how it
applies to real datasets. Through careful experimentation on UCI datasets,
CIFAR-10, and the UTKFace dataset, we find that the overfitting from
overparameterized deep kernel learning, in which the model is "somewhat
Bayesian", can in certain scenarios be worse than that from not being Bayesian
at all. However, we find that a fully Bayesian treatment of deep kernel
learning can rectify this overfitting and obtain the desired performance
improvements over standard neural networks and Gaussian processes.
| stat |
Bayesian Networks for Max-linear Models | We study Bayesian networks based on max-linear structural equations as
introduced in Gissibl and Kl\"uppelberg [16] and provide a summary of their
independence properties. In particular we emphasize that distributions for such
networks are generally not faithful to the independence model determined by
their associated directed acyclic graph. In addition, we consider some of the
basic issues of estimation and discuss generalized maximum likelihood
estimation of the coefficients, using the concept of a generalized likelihood
ratio for non-dominated families as introduced by Kiefer and Wolfowitz [21].
Finally we argue that the structure of a minimal network asymptotically can be
identified completely from observational data.
| stat |
Personality Traits and Drug Consumption. A Story Told by Data | This is a preprint version of the first book from the series: "Stories told
by data". In this book a story is told about the psychological traits
associated with drug consumption. The book includes:
- A review of published works on the psychological profiles of drug users.
- Analysis of a new original database with information on 1885 respondents
and usage of 18 drugs. (Database is available online.)
- An introductory description of the data mining and machine learning methods
used for the analysis of this dataset.
- The demonstration that the personality traits (five factor model,
impulsivity, and sensation seeking), together with simple demographic data,
give the possibility of predicting the risk of consumption of individual drugs
with sensitivity and specificity above 70% for most drugs.
- The analysis of correlations of use of different substances and the
description of the groups of drugs with correlated use (correlation pleiades).
- Proof of significant differences of personality profiles for users of
different drugs. This is explicitly proved for benzodiazepines, ecstasy, and
heroin.
- Tables of personality profiles for users and non-users of 18 substances.
The book is aimed at advanced undergraduates or first-year PhD students, as
well as researchers and practitioners. No previous knowledge of machine
learning, advanced data mining concepts or modern psychology of personality is
assumed. For more detailed introduction into statistical methods we recommend
several undergraduate textbooks. Familiarity with basic statistics and some
experience in the use of probabilities would be helpful as well as some basic
technical understanding of psychology.
| stat |
Enabling Cost-Effective Population Health Monitoring By Exploiting
Spatiotemporal Correlation: An Empirical Study | Because of its important role in health policy-shaping, population health
monitoring (PHM) is considered a fundamental block for public health services.
However, traditional public health data collection approaches, such as
clinic-visit-based data integration or health surveys, could be very costly and
time-consuming. To address this challenge, this paper proposes a cost-effective
approach called Compressive Population Health (CPH), where a subset of a given
area is selected in terms of regions within the area for data collection in the
traditional way, while leveraging inherent spatial correlations of neighboring
regions to perform data inference for the rest of the area. By alternating
selected regions longitudinally, this approach can validate and correct
previously assessed spatial correlations. To verify whether the idea of CPH is
feasible, we conduct an in-depth study based on spatiotemporal morbidity rates
of chronic diseases in more than 500 regions around London for over ten years.
We introduce our CPH approach and present three extensive analytical studies.
The first confirms that significant spatiotemporal correlations do exist. In
the second study, by deploying multiple state-of-the-art data recovery
algorithms, we verify that these spatiotemporal correlations can be leveraged
to do data inference accurately using only a small number of samples. Finally,
we compare different methods for region selection for traditional data
collection and show how such methods can further reduce the overall cost while
maintaining high PHM quality.
| stat |
Non-Homogeneous Poisson Process Intensity Modeling and Estimation using
Measure Transport | Non-homogeneous Poisson processes are used in a wide range of scientific
disciplines, ranging from the environmental sciences to the health sciences.
Often, the central object of interest in a point process is the underlying
intensity function. Here, we present a general model for the intensity function
of a non-homogeneous Poisson process using measure transport. The model is
built from a flexible bijective mapping that maps from the underlying intensity
function of interest to a simpler reference intensity function. We enforce
bijectivity by modeling the map as a composition of multiple simple bijective
maps, and show that the model exhibits an important approximation property.
Estimation of the flexible mapping is accomplished within an optimization
framework, wherein computations are efficiently done using recent technological
advances in deep learning and a graphics processing unit. Although we find that
intensity function estimates obtained with our method are not necessarily
superior to those obtained using conventional methods, the modeling
representation brings with it other advantages such as facilitated point
process simulation and uncertainty quantification. Modeling point processes in
higher dimensions is also facilitated using our approach. We illustrate the use
of our model on both simulated data, and a real data set containing the
locations of seismic events near Fiji since 1964.
| stat |
Adversarial Robustness through Local Linearization | Adversarial training is an effective methodology for training deep neural
networks that are robust against adversarial, norm-bounded perturbations.
However, the computational cost of adversarial training grows prohibitively as
the size of the model and number of input dimensions increase. Further,
training against less expensive and therefore weaker adversaries produces
models that are robust against weak attacks but break down under attacks that
are stronger. This is often attributed to the phenomenon of gradient
obfuscation; such models have a highly non-linear loss surface in the vicinity
of training examples, making it hard for gradient-based attacks to succeed even
though adversarial examples still exist. In this work, we introduce a novel
regularizer that encourages the loss to behave linearly in the vicinity of the
training data, thereby penalizing gradient obfuscation while encouraging
robustness. We show via extensive experiments on CIFAR-10 and ImageNet, that
models trained with our regularizer avoid gradient obfuscation and can be
trained significantly faster than adversarial training. Using this regularizer,
we exceed current state of the art and achieve 47% adversarial accuracy for
ImageNet with l-infinity adversarial perturbations of radius 4/255 under an
untargeted, strong, white-box attack. Additionally, we match state of the art
results for CIFAR-10 at 8/255.
| stat |
Modern strategies for time series regression | This paper discusses several modern approaches to regression analysis
involving time series data where some of the predictor variables are also
indexed by time. We discuss classical statistical approaches as well as methods
that have been proposed recently in the machine learning literature. The
approaches are compared and contrasted, and it will be seen that there are
advantages and disadvantages to most currently available approaches. There is
ample room for methodological developments in this area. The work is motivated
by an application involving the prediction of water levels as a function of
rainfall and other climate variables in an aquifer in eastern Australia.
| stat |
Bayesian Active Learning for Structured Output Design | In this paper, we propose an active learning method for an inverse problem
that aims to find an input that achieves a desired structured-output. The
proposed method provides new acquisition functions for minimizing the error
between the desired structured-output and the prediction of a Gaussian process
model, by effectively incorporating the correlation between multiple outputs of
the underlying multi-valued black box output functions. The effectiveness of
the proposed method is verified by applying it to two synthetic shape search
problem and real data. In the real data experiment, we tackle the input
parameter search which achieves the desired crystal growth rate in silicon
carbide (SiC) crystal growth modeling, that is a problem of materials
informatics.
| stat |
Distribution-Free Multisample Test Based on Optimal Matching with
Applications to Single Cell Genomics | In this paper we propose a nonparametric graphical test based on optimal
matching, for assessing the equality of multiple unknown multivariate
probability distributions. Our procedure pools the data from the different
classes to create a graph based on the minimum non-bipartite matching, and then
utilizes the number of edges connecting data points from different classes to
examine the closeness between the distributions. The proposed test is exactly
distribution-free (the null distribution does not depend on the distribution of
the data) and can be efficiently applied to multivariate as well as
non-Euclidean data, whenever the inter-point distances are well-defined. We
show that the test is universally consistent, and prove a distributional limit
theorem for the test statistic under general alternatives. Through simulation
studies, we demonstrate its superior performance against other common and
well-known multisample tests. In scenarios where our test suggests
distributional differences across classes, we also propose an approach for
identifying which class or group contributes to this overall difference. The
method is applied to single cell transcriptomics data obtained from the
peripheral blood, cancer tissue, and tumor-adjacent normal tissue of human
subjects with hepatocellular carcinoma and non-small-cell lung cancer. Our
method unveils patterns in how biochemical metabolic pathways are altered
across immune cells in a cancer setting, depending on the tissue location. All
of the methods described herein are implemented in the R package multicross.
| stat |
A Relationship Between SIR Model and Generalized Logistic Distribution
with Applications to SARS and COVID-19 | This paper shows that the generalized logistic distribution model is derived
from the well-known compartment model, consisting of susceptible, infected and
recovered compartments, abbreviated as the SIR model, under certain conditions.
In the SIR model, there are uncertainties in predicting the final values for
the number of infected population and the infectious parameter. However, by
utilizing the information obtained from the generalized logistic distribution
model, we can perform the SIR numerical computation more stably and more
accurately. Applications to severe acute respiratory syndrome (SARS) and
Coronavirus disease 2019 (COVID-19) using this combined method are also
introduced.
| stat |
Decontamination of Mutual Contamination Models | Many machine learning problems can be characterized by mutual contamination
models. In these problems, one observes several random samples from different
convex combinations of a set of unknown base distributions and the goal is to
infer these base distributions. This paper considers the general setting where
the base distributions are defined on arbitrary probability spaces. We examine
three popular machine learning problems that arise in this general setting:
multiclass classification with label noise, demixing of mixed membership
models, and classification with partial labels. In each case, we give
sufficient conditions for identifiability and present algorithms for the
infinite and finite sample settings, with associated performance guarantees.
| stat |
Regularized Estimation in High-Dimensional Vector Auto-Regressive Models
using Spatio-Temporal Information | A Vector Auto-Regressive (VAR) model is commonly used to model multivariate
time series, and there are many penalized methods to handle high
dimensionality. However in terms of spatio-temporal data, most methods do not
take the spatial and temporal structure of the data into consideration, which
may lead to unreliable network detection and inaccurate forecasts. This paper
proposes a data-driven weighted l1 regularized approach for spatio-temporal VAR
model. Extensive simulation studies are carried out to compare the proposed
method with four existing methods of high-dimensional VAR model, demonstrating
improvements of our method over others in parameter estimation, network
detection and out-of-sample forecasts. We also apply our method on a traffic
data set to evaluate its performance in real application. In addition, we
explore the theoretical properties of l1 regularized estimation of VAR model
under the weakly sparse scenario, in which the exact sparsity can be viewed as
a special case. To the best of our knowledge, this direction has not been
considered yet in the literature. For general stationary VAR process, we derive
the non-asymptotic upper bounds on l1 regularized estimation errors under the
weakly sparse scenario, provide the conditions of estimation consistency, and
further simplify these conditions for a special VAR(1) case.
| stat |
Likelihood Inference for Possibly Non-Stationary Processes via Adaptive
Overdifferencing | We make a simple observation that facilitates valid likelihood-based
inference for the parameters of the popular ARFIMA or FARIMA model without
requiring stationarity by allowing the upper bound $\bar{d}$ for the memory
parameter $d$ to exceed $0.5$. We observe that estimating the parameters of a
single non-stationary ARFIMA model is equivalent to estimating the parameters
of a sequence of stationary ARFIMA models. This enables improved inference
because many standard methods perform poorly when estimates are close to the
boundary of the parameter space. It also allows us to leverage the wealth of
likelihood approximations that have been introduced for estimating the
parameters of a stationary process. We explore how estimation of the memory
parameter $d$ depends on the upper bound $\bar{d}$ and introduce adaptive
procedures for choosing $\bar{d}$. Via simulations, we examine the performance
of our adaptive procedures for estimating the memory parameter when the true
value is as large as $2.5$. Our adaptive procedures estimate the memory
parameter well, can be used to obtain confidence intervals for the memory
parameter that achieve nominal coverage rates, and perform favorably relative
to existing alternatives.
| stat |
Predicting phenotypes from microarrays using amplified, initially
marginal, eigenvector regression | Motivation: The discovery of relationships between gene expression
measurements and phenotypic responses is hampered by both computational and
statistical impediments. Conventional statistical methods are less than ideal
because they either fail to select relevant genes, predict poorly, ignore the
unknown interaction structure between genes, or are computationally
intractable. Thus, the creation of new methods which can handle many expression
measurements on relatively small numbers of patients while also uncovering
gene-gene relationships and predicting well is desirable.
Results: We develop a new technique for using the marginal relationship
between gene expression measurements and patient survival outcomes to identify
a small subset of genes which appear highly relevant for predicting survival,
produce a low-dimensional embedding based on this small subset, and amplify
this embedding with information from the remaining genes. We motivate our
methodology by using gene expression measurements to predict survival time for
patients with diffuse large B-cell lymphoma, illustrate the behavior of our
methodology on carefully constructed synthetic examples, and test it on a
number of other gene expression datasets. Our technique is computationally
tractable, generally outperforms other methods, is extensible to other
phenotypes, and also identifies different genes (relative to existing methods)
for possible future study.
Key words: regression; principal components; matrix sketching;
preconditioning
Availability: All of the code and data are available at
https://github.com/dajmcdon/aimer/.
| stat |
An integer-valued time series model for multivariate surveillance | In recent days different types of surveillance data are becoming available
for public health reasons. In most cases several variables are monitored and
events of different types are reported. As the amount of surveillance data
increases, statistical methods that can effectively address multivariate
surveillance scenarios are demanded. Even though research activity in this
field is increasing rapidly in recent years, only a few approaches have
simultaneously addressed the integer-valued property of the data and its
correlation (both time correlation and cross correlation) structure. In this
paper, we suggest a multivariate integer-valued autoregressive model that
allows for both serial and cross correlation between the series and can easily
accommodate overdispersion and covariate information. Moreover, its structure
implies a natural decomposition into an endemic and an epidemic component, a
common distinction in dynamic models for infectious disease counts. Detection
of disease outbreaks is achieved through the comparison of surveillance data
with one-step-ahead predictions obtained after fitting the suggested model to a
set of clean historical data. The performance of the suggested model is
illustrated on a trivariate series of syndromic surveillance data collected
during Athens 2004 Olympic Games.
| stat |
Interpretable, predictive spatio-temporal models via enhanced Pairwise
Directions Estimation | This article concerns the predictive modeling for spatio-temporal data as
well as model interpretation using data information in space and time.
Intrinsically, we develop a novel approach based on dimension reduction for
such data in order to capture nonlinear mean structures without requiring a
prespecified parametric model. In addition to prediction as a common interest,
this approach focuses more on the exploration of geometric information in the
data. The method of Pairwise Directions Estimation (PDE) is incorporated in our
approach to implement the data-driven function searching of spatial structures
and temporal patterns, useful in exploring data trends. The benefit of using
geometrical information from the method of PDE is highlighted. We further
enhance PDE, referring to it as PDE+, by using resolution adaptive fixed rank
kriging to estimate the random effects not explained in the mean structures.
Our proposal can not only produce more accurate and explainable prediction, but
also increase the computation efficiency for model building. Several simulation
examples are conducted and comparisons are made with four existing methods. The
results demonstrate that the proposed PDE+ method is very useful for exploring
and interpreting the patterns of trend for spatio-temporal data. Illustrative
applications to two real datasets are also presented.
| stat |
Coupled Graphs and Tensor Factorization for Recommender Systems and
Community Detection | Joint analysis of data from multiple information repositories facilitates
uncovering the underlying structure in heterogeneous datasets. Single and
coupled matrix-tensor factorization (CMTF) has been widely used in this context
for imputation-based recommendation from ratings, social network, and other
user-item data. When this side information is in the form of item-item
correlation matrices or graphs, existing CMTF algorithms may fall short.
Alleviating current limitations, we introduce a novel model coined coupled
graph-tensor factorization (CGTF) that judiciously accounts for graph-related
side information. The CGTF model has the potential to overcome practical
challenges, such as missing slabs from the tensor and/or missing rows/columns
from the correlation matrices. A novel alternating direction method of
multipliers (ADMM) is also developed that recovers the nonnegative factors of
CGTF. Our algorithm enjoys closed-form updates that result in reduced
computational complexity and allow for convergence claims. A novel direction is
further explored by employing the interpretable factors to detect graph
communities having the tensor as side information. The resulting community
detection approach is successful even when some links in the graphs are
missing. Results with real data sets corroborate the merits of the proposed
methods relative to state-of-the-art competing factorization techniques in
providing recommendations and detecting communities.
| stat |
Personalized Treatment Selection using Causal Heterogeneity | Randomized experimentation (also known as A/B testing or bucket testing) is
widely used in the internet industry to measure the metric impact obtained by
different treatment variants. A/B tests identify the treatment variant showing
the best performance, which then becomes the chosen or selected treatment for
the entire population. However, the effect of a given treatment can differ
across experimental units and a personalized approach for treatment selection
can greatly improve upon the usual global selection strategy. In this work, we
develop a framework for personalization through (i) estimation of heterogeneous
treatment effect at either a cohort or member-level, followed by (ii) selection
of optimal treatment variants for cohorts (or members) obtained through
(deterministic or stochastic) constrained optimization.
We perform a two-fold evaluation of our proposed methods. First, a simulation
analysis is conducted to study the effect of personalized treatment selection
under carefully controlled settings. This simulation illustrates the
differences between the proposed methods and the suitability of each with
increasing uncertainty. We also demonstrate the effectiveness of the method
through a real-life example related to serving notifications at Linkedin. The
solution significantly outperformed both heuristic solutions and the global
treatment selection baseline leading to a sizable win on top-line metrics like
member visits.
| stat |
Bayesian Protein Sequence and Structure Alignment | The structure of a protein is crucial in determining its functionality, and
is much more conserved than sequence during evolution. A key task in structural
biology is to compare protein structures in order to determine evolutionary
relationships, estimate the function of newly-discovered structures, and
predict unknown structures. We propose a Bayesian method for protein structure
alignment, with the prior on alignments based on functions which penalise
``gaps'' in the aligned sequences. We show how a broad class of penalty
functions fits into this framework, and how the resulting posterior
distribution can be efficiently sampled. A commonly-used gap penalty function
is shown to be a special case, and we propose a new penalty function which
alleviates an undesirable feature of the commonly-used penalty. We illustrate
our method on benchmark data sets, and find it competes well with popular tools
from computational biology. Our method has the benefit of being able to
potentially explore multiple competing alignments and quantify their merits
probabilistically. The framework naturally allows for further information such
as amino acid sequence to be included, and could be adapted to other situations
such as flexible proteins or domain swaps.
| stat |
Bayesian Learning of Conditional Kernel Mean Embeddings for Automatic
Likelihood-Free Inference | In likelihood-free settings where likelihood evaluations are intractable,
approximate Bayesian computation (ABC) addresses the formidable inference task
to discover plausible parameters of simulation programs that explain the
observations. However, they demand large quantities of simulation calls.
Critically, hyperparameters that determine measures of simulation discrepancy
crucially balance inference accuracy and sample efficiency, yet are difficult
to tune. In this paper, we present kernel embedding likelihood-free inference
(KELFI), a holistic framework that automatically learns model hyperparameters
to improve inference accuracy given limited simulation budget. By leveraging
likelihood smoothness with conditional mean embeddings, we nonparametrically
approximate likelihoods and posteriors as surrogate densities and sample from
closed-form posterior mean embeddings, whose hyperparameters are learned under
its approximate marginal likelihood. Our modular framework demonstrates
improved accuracy and efficiency on challenging inference problems in ecology.
| stat |
NLMEModeling: A Wolfram Mathematica Package for Nonlinear Mixed Effects
Modeling of Dynamical Systems | Nonlinear mixed effects modeling is a powerful tool when analyzing data from
several entities in an experiment. In this paper, we present NLMEModeling, a
package for mixed effects modeling in Wolfram Mathematica. NLMEModeling
supports mixed effects modeling of dynamical systems where the underlying
dynamics are described by either ordinary or stochastic differential equations
combined with a flexible observation error model. Moreover, NLMEModeling is a
user-friendly package with functionality for model validation, visual
predictive checks and simulation capabilities. The package is freely available
and provides a flexible add-on to Wolfram Mathematica.
| stat |
A Factor Stochastic Volatility Model with Markov-Switching Panic Regimes | The use of factor stochastic volatility models requires choosing the number
of latent factors used to describe the dynamics of the financial returns
process; however, empirical evidence suggests that the number and makeup of
pertinent factors is time-varying and economically situational. We present a
novel factor stochastic volatility model that allows for random subsets of
assets to have their members experience non-market-wide panics. These
participating assets will experience an increase in their variances and
within-group covariances. We also give an estimation algorithm for this model
that takes advantage of recent results on Particle Markov chain Monte Carlo
techniques.
| stat |
Longitudinal Variational Autoencoder | Longitudinal datasets measured repeatedly over time from individual subjects,
arise in many biomedical, psychological, social, and other studies. A common
approach to analyse high-dimensional data that contains missing values is to
learn a low-dimensional representation using variational autoencoders (VAEs).
However, standard VAEs assume that the learnt representations are i.i.d., and
fail to capture the correlations between the data samples. We propose the
Longitudinal VAE (L-VAE), that uses a multi-output additive Gaussian process
(GP) prior to extend the VAE's capability to learn structured low-dimensional
representations imposed by auxiliary covariate information, and derive a new KL
divergence upper bound for such GPs. Our approach can simultaneously
accommodate both time-varying shared and random effects, produce structured
low-dimensional representations, disentangle effects of individual covariates
or their interactions, and achieve highly accurate predictive performance. We
compare our model against previous methods on synthetic as well as clinical
datasets, and demonstrate the state-of-the-art performance in data imputation,
reconstruction, and long-term prediction tasks.
| stat |
Structural Equation Models as Computation Graphs | Structural equation modeling (SEM) is a popular tool in the social and
behavioural sciences, where it is being applied to ever more complex data
types. The high-dimensional data produced by modern sensors, brain images, or
(epi)genetic measurements require variable selection using parameter
penalization; experimental models combining disparate data sources benefit from
regularization to obtain a stable result; and genomic SEM or network models
lead to alternative objective functions. With each proposed extension,
researchers currently have to completely reformulate SEM and its optimization
algorithm -- a challenging and time-consuming task.
In this paper, we consider each SEM as a computation graph, a flexible method
of specifying objective functions borrowed from the field of deep learning.
When combined with state-of-the-art optimizers, our computation graph approach
can extend SEM without the need for bespoke software development. We show that
both existing and novel SEM improvements follow naturally from our approach. To
demonstrate, we discuss least absolute deviation estimation and penalized
regression models. We also introduce spike-and-slab SEM, which may perform
better when shrinkage of large factor loadings is not desired. By applying
computation graphs to SEM, we hope to greatly accelerate the process of
developing SEM techniques, paving the way for new applications. We provide an
accompanying R package tensorsem.
| stat |
The Normal-Generalised Gamma-Pareto process: A novel pure-jump L\'evy
process with flexible tail and jump-activity properties | Pure-jump L\'evy processes are popular classes of stochastic processes which
have found many applications in finance, statistics or machine learning. In
this paper, we propose a novel family of self-decomposable L\'evy processes
where one can control separately the tail behavior and the jump activity of the
process, via two different parameters. Crucially, we show that one can sample
exactly increments of this process, at any time scale; this allows the
implementation of likelihood-free Markov chain Monte Carlo algorithms for
(asymptotically) exact posterior inference. We use this novel process in
L\'evy-based stochastic volatility models to predict the returns of stock
market data, and show that the proposed class of models leads to superior
predictive performances compared to classical alternatives.
| stat |
A New Approach to Determine the Coefficient of Skewness and An
Alternative Form of Boxplot | To solve the problems in measuring coefficient of skewness related to extreme
value, irregular distance from the middle point and distance between two
consecutive numbers, "Rank skewness" a new measure of the coefficient of
skewness has been proposed in this paper. Comparing with other measures of the
coefficient of skewness, proposed measure of the coefficient of skewness
performs better specially for skewed distribution. An alternative of five point
summary boxplot, a four point summary graph has also been proposed which is
simpler than the traditional boxplot. It is based on all observation and give
better result than the five point summary.
| stat |
Multi-resolution Spatial Regression for Aggregated Data with an
Application to Crop Yield Prediction | We develop a new methodology for spatial regression of aggregated outputs on
multi-resolution covariates. Such problems often occur with spatial data, for
example in crop yield prediction, where the output is spatially-aggregated over
an area and the covariates may be observed at multiple resolutions. Building
upon previous work on aggregated output regression, we propose a regression
framework to synthesise the effects of the covariates at different resolutions
on the output and provide uncertainty estimation. We show that, for a crop
yield prediction problem, our approach is more scalable, via variational
inference, than existing multi-resolution regression models. We also show that
our framework yields good predictive performance, compared to existing
multi-resolution crop yield models, whilst being able to provide estimation of
the underlying spatial effects.
| stat |
Contributed Discussion of "A Bayesian Conjugate Gradient Method" | We would like to congratulate the authors of "A Bayesian Conjugate Gradient
Method" on their insightful paper, and welcome this publication which we firmly
believe will become a fundamental contribution to the growing field of
probabilistic numerical methods and in particular the sub-field of Bayesian
numerical methods. In this short piece, which will be published as a comment
alongside the main paper, we first initiate a discussion on the choice of
priors for solving linear systems, then propose an extension of the Bayesian
conjugate gradient (BayesCG) algorithm for solving several related linear
systems simultaneously.
| stat |
On the minimax optimality and superiority of deep neural network
learning over sparse parameter spaces | Deep learning has been applied to various tasks in the field of machine
learning and has shown superiority to other common procedures such as kernel
methods. To provide a better theoretical understanding of the reasons for its
success, we discuss the performance of deep learning and other methods on a
nonparametric regression problem with a Gaussian noise. Whereas existing
theoretical studies of deep learning have been based mainly on mathematical
theories of well-known function classes such as H\"{o}lder and Besov classes,
we focus on function classes with discontinuity and sparsity, which are those
naturally assumed in practice. To highlight the effectiveness of deep learning,
we compare deep learning with a class of linear estimators representative of a
class of shallow estimators. It is shown that the minimax risk of a linear
estimator on the convex hull of a target function class does not differ from
that of the original target function class. This results in the suboptimality
of linear methods over a simple but non-convex function class, on which deep
learning can attain nearly the minimax-optimal rate. In addition to this
extreme case, we consider function classes with sparse wavelet coefficients. On
these function classes, deep learning also attains the minimax rate up to log
factors of the sample size, and linear methods are still suboptimal if the
assumed sparsity is strong. We also point out that the parameter sharing of
deep neural networks can remarkably reduce the complexity of the model in our
setting.
| stat |
Fast and Robust Comparison of Probability Measures in Heterogeneous
Spaces | Comparing two probability measures supported on heterogeneous spaces is an
increasingly important problem in machine learning. Such problems arise when
comparing for instance two populations of biological cells, each described with
its own set of features, or when looking at families of word embeddings trained
across different corpora/languages. For such settings, the Gromov Wasserstein
(GW) distance is often presented as the gold standard. GW is intuitive, as it
quantifies whether one measure can be isomorphically mapped to the other.
However, its exact computation is intractable, and most algorithms that claim
to approximate it remain expensive. Building on \cite{memoli-2011}, who
proposed to represent each point in each distribution as the 1D distribution of
its distances to all other points, we introduce in this paper the Anchor Energy
(AE) and Anchor Wasserstein (AW) distances, which are respectively the energy
and Wasserstein distances instantiated on such representations. Our main
contribution is to propose a sweep line algorithm to compute AE \emph{exactly}
in log-quadratic time, where a naive implementation would be cubic. This is
quasi-linear w.r.t. the description of the problem itself. Our second
contribution is the proposal of robust variants of AE and AW that uses rank
statistics rather than the original distances. We show that AE and AW perform
well in various experimental settings at a fraction of the computational cost
of popular GW approximations. Code is available at
\url{https://github.com/joisino/anchor-energy}.
| stat |
A Second look at Exponential and Cosine Step Sizes: Simplicity,
Convergence, and Performance | Stochastic Gradient Descent (SGD) is a popular tool in training large-scale
machine learning models. Its performance, however, is highly variable,
depending crucially on the choice of the step sizes. Accordingly, a variety of
strategies for tuning the step sizes have been proposed. Yet, most of them lack
a theoretical guarantee, whereas those backed by theories often do not shine in
practice. In this paper, we study two heuristic step size schedules whose power
has been repeatedly confirmed in practice: the exponential and the cosine step
sizes. For the first time, we provide theoretical support for them: we prove
their (almost) optimal convergence rates for stochastic optimization of smooth
non-convex functions. Furthermore, if in addition, the Polyak-\L{}ojasiewicz
(PL) condition holds, they both automatically adapt to the level of noise, with
a rate interpolating between a linear rate for the noiseless case and a
sub-linear one for the noisy case. Finally, we conduct a fair and comprehensive
empirical evaluation of real-world datasets with deep learning architectures.
Results show that, even if only requiring at most two hyperparameters to tune,
they best or match the performance of various finely-tuned state-of-the-art
strategies.
| stat |
Bayesian Optimization with Unknown Search Space | Applying Bayesian optimization in problems wherein the search space is
unknown is challenging. To address this problem, we propose a systematic volume
expansion strategy for the Bayesian optimization. We devise a strategy to
guarantee that in iterative expansions of the search space, our method can find
a point whose function value within epsilon of the objective function maximum.
Without the need to specify any parameters, our algorithm automatically
triggers a minimal expansion required iteratively. We derive analytic
expressions for when to trigger the expansion and by how much to expand. We
also provide theoretical analysis to show that our method achieves
epsilon-accuracy after a finite number of iterations. We demonstrate our method
on both benchmark test functions and machine learning hyper-parameter tuning
tasks and demonstrate that our method outperforms baselines.
| stat |
The Wasserstein Impact Measure (WIM): a generally applicable, practical
tool for quantifying prior impact in Bayesian statistics | The prior distribution is a crucial building block in Bayesian analysis, and
its choice will impact the subsequent inference. It is therefore important to
have a convenient way to quantify this impact, as such a measure of prior
impact will help us to choose between two or more priors in a given situation.
A recently proposed approach consists in determining the Wasserstein distance
between posteriors resulting from two distinct priors, revealing how close or
distant they are. In particular, if one prior is the uniform/flat prior, this
distance leads to a genuine measure of prior impact for the other prior. While
highly appealing and successful from a theoretical viewpoint, this proposal
suffers from practical limitations: it requires prior distributions to be
nested, posterior distributions should not be of a too complex form, in most
considered settings the exact distance was not computed but sharp upper and
lower bounds were proposed, and the proposal so far is restricted to scalar
parameter settings. In this paper, we overcome all these limitations by
introducing a practical version of this theoretical approach, namely the
Wasserstein Impact Measure (WIM). In three simulated scenarios, we will compare
the WIM to the theoretical Wasserstein approach, as well as to two competitor
prior impact measures from the literature. We finally illustrate the
versatility of the WIM by applying it on two datasets.
| stat |
MaxSkew and MultiSkew: Two R Packages for Detecting, Measuring and
Removing Multivariate Skewness | Skewness plays a relevant role in several multivariate statistical
techniques. Sometimes it is used to recover data features, as in cluster
analysis. In other circumstances, skewness impairs the performances of
statistical methods, as in the Hotelling's one-sample test. In both cases,
there is the need to check the symmetry of the underlying distribution, either
by visual inspection or by formal testing. The R packages MaxSkew and MultiSkew
address these issues by measuring, testing and removing skewness from
multivariate data. Skewness is assessed by the third multivariate cumulant and
its functions. The hypothesis of symmetry is tested either nonparametrically,
with the bootstrap, or parametrically, under the normality assumption. Skewness
is removed or at least alleviated by projecting the data onto appropriate
linear subspaces. Usages of MaxSkew and MultiSkew are illustrated with the Iris
dataset.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.