Statistics Theory
New submissions
[ showing up to 2000 entries per page: fewer  more ]
New submissions for Wed, 15 Jul 20
 [1] arXiv:2007.06735 [pdf, other]

Title: On the Current and Longest Head RunAuthors: Dennis KohComments: 17 pages, 8 figuresSubjects: Statistics Theory (math.ST); Probability (math.PR)
In this paper, the open problem of finding a closed analytical expression for the distribution function of the length of the longest pure head run in coin tosses of a possibly biased coin is solved by studying the closely related Markov chain of current head runs. Based on this, inequaltities and an asymptotic expression for the centered length of the longest head run can be derived. Moreover, formulae for parameters like expected value and variance solely by means of the distribution function are given. The corresponding results for the length of the longest whatever run in tosses of fair coins are also included, heuristics are discussed as well.
Crosslists for Wed, 15 Jul 20
 [2] arXiv:2007.06697 (crosslist from math.PR) [pdf, other]

Title: Species tree estimation under joint modeling of coalescence and duplication: sample complexity of quartet methodsComments: 35 pages, 1 figureSubjects: Probability (math.PR); Statistics Theory (math.ST); Populations and Evolution (qbio.PE)
We consider species tree estimation under a standard stochastic model of gene tree evolution that incorporates incomplete lineage sorting (as modeled by a coalescent process) and gene duplication and loss (as modeled by a branching process). Through a probabilistic analysis of the model, we derive sample complexity bounds for widely used quartetbased inference methods that highlight the effect of the duplication and loss rates in both subcritical and supercritical regimes.
 [3] arXiv:2007.06715 (crosslist from math.DS) [pdf, other]

Title: Dynamics of coordinate ascent variational inference: A case study in 2D Ising modelsSubjects: Dynamical Systems (math.DS); Statistics Theory (math.ST)
Variational algorithms have gained prominence over the past two decades as a scalable computational environment for Bayesian inference. In this article, we explore tools from the dynamical systems literature to study convergence of coordinate ascent algorithms for mean field variational inference. Focusing on the Ising model defined on two nodes, we fully characterize the dynamics of the sequential coordinate ascent algorithm and its parallel version. We observe that in the regime where the objective function is convex, both the algorithms are stable and exhibit convergence to the unique fixed point. Our analyses reveal interesting {\em discordances} between these two versions of the algorithm in the region when the objective function is nonconvex. In fact, the parallel version exhibits a periodic oscillatory behavior which is absent in the sequential version. Drawing intuition from the Markov chain Monte Carlo literature, we {\em empirically} show that a parameter expansion of the Ising model, popularly called as the EdwardSokal coupling, leads to an enlargement of the regime of convergence to the global optima.
 [4] arXiv:2007.06799 (crosslist from stat.ML) [pdf, ps, other]

Title: A Decentralized Approach to Bayesian LearningComments: 52 pages, 29 figuresSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
Motivated by decentralized approaches to machine learning, we propose a collaborative Bayesian learning algorithm taking the form of decentralized Langevin dynamics in a nonconvex setting. Our analysis show that the initial KLdivergence between the Markov Chain and the target posterior distribution is exponentially decreasing while the error contributions to the overall KLdivergence from the additive noise is decreasing in polynomial time. We further show that the polynomialterm experiences speedup with number of agents and provide sufficient conditions on the timevarying stepsizes to guarantee convergence to the desired distribution. The performance of the proposed algorithm is evaluated on a wide variety of machine learning tasks. The empirical results show that the performance of individual agents with locally available data is on par with the centralized setting with considerable improvement in the convergence rate.
 [5] arXiv:2007.06827 (crosslist from stat.ML) [pdf, other]

Title: Early stopping and polynomial smoothing in regression with reproducing kernelsSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
In this paper we study the problem of early stopping for iterative learning algorithms in reproducing kernel Hilbert space (RKHS) in the nonparametric regression framework. In particular, we work with gradient descent and (iterative) kernel ridge regression algorithms. We present a datadriven rule to perform early stopping without a validation set that is based on the socalled minimum discrepancy principle. This method enjoys only one assumption on the regression function: it belongs to a reproducing kernel Hilbert space (RKHS). The proposed rule is proved to be minimax optimal over different types of kernel spaces, including finite rank and Sobolev smoothness classes. The proof is derived from the fixedpoint analysis of the localized Rademacher complexities, which is a standard technique for obtaining optimal rates in the nonparametric regression literature. In addition to that, we present simulations results on artificial datasets that show comparable performance of the designed rule with respect to other stopping rules such as the one determined by Vfold crossvalidation.
 [6] arXiv:2007.07065 (crosslist from econ.EM) [pdf, other]

Title: A More Robust tTestAuthors: Ulrich K. MuellerSubjects: Econometrics (econ.EM); Statistics Theory (math.ST)
Standard inference about a scalar parameter estimated via GMM amounts to applying a ttest to a particular set of observations. If the number of observations is not very large, then moderately heavy tails can lead to poor behavior of the ttest. This is a particular problem under clustering, since the number of observations then corresponds to the number of clusters, and heterogeneity in cluster sizes induces a form of heavy tails. This paper combines extreme value theory for the smallest and largest observations with a normal approximation for the average of the remaining observations to construct a more robust alternative to the ttest. The new test is found to control size much more successfully in small samples compared to existing methods. Analytical results in the canonical inference for the mean problem demonstrate that the new test provides a refinement over the full sample ttest under more than two but less than three moments, while the bootstrapped ttest does not.
Replacements for Wed, 15 Jul 20
 [7] arXiv:1909.03540 (replaced) [pdf, other]

Title: Inference In Highdimensional SingleIndex Models Under Symmetric DesignsSubjects: Statistics Theory (math.ST); Other Statistics (stat.OT)
 [8] arXiv:1911.10604 (replaced) [pdf, other]

Title: Optimal Permutation Recovery in Permuted Monotone Matrix ModelJournalref: Journal of the American Statistical Association, 2020Subjects: Statistics Theory (math.ST); Methodology (stat.ME)
 [9] arXiv:2001.11201 (replaced) [pdf, other]

Title: FiniteTime Analysis of RoundRobin KullbackLeibler Upper Confidence Bounds for Optimal Adaptive Allocation with Multiple Plays and Markovian RewardsAuthors: Vrettos MoulosComments: 31 pages, simulation results addedSubjects: Statistics Theory (math.ST); Machine Learning (cs.LG); Machine Learning (stat.ML)
 [10] arXiv:2002.08422 (replaced) [pdf, other]

Title: On conditional versus marginal bias in multiarmed banditsComments: 18 pagesSubjects: Statistics Theory (math.ST); Machine Learning (stat.ML)
 [11] arXiv:2007.01958 (replaced) [pdf, other]

Title: Twosample Testing for Large, Sparse HighDimensional Multinomials under Rare/Weak PerturbationsSubjects: Statistics Theory (math.ST); Computation (stat.CO)
 [12] arXiv:1910.09485 (replaced) [pdf, other]

Title: Counterexamples for optimal scaling of MetropolisHastings chains with rough target densitiesComments: 44 pages, 3 figuresSubjects: Probability (math.PR); Statistics Theory (math.ST)
[ showing up to 2000 entries per page: fewer  more ]
Disable MathJax (What is MathJax?)
Links to: arXiv, form interface, find, stat, recent, 2007, contact, help (Access key information)