Stochastics and Statistics Seminar

Abstract: Many existing literature on bandits and reinforcement learning assume a linear reward/value function, but what happens if the reward is non-linear? Two curious phenomena arise for non-linear bandits: first, in addition to the "learning phase" with a standard \Theta(\sqrt(T)) regret, there is an "initialization phase" with a fixed cost determined by the reward function; second, achieving the smallest cost of the initialization phase requires new learning algorithms other than traditional ones such as UCB. For a special family of…

ABSTRACT Recent advances in data collection have made sports an attractive testing ground for new analyses and algorithms, and a fascinating controlled microcosm in which to explore social interactions. In this talk I will describe two studies in this arena: one related to public health and the pandemic and one related to decision-making in basketball.  In the first, I will discuss what can be learned from the natural experiments that were (fortuitously) run in America football stadiums. During the 2020…

Abstract: Motivated by applications to single-particle cryo-electron microscopy, we study a problem of group orbit estimation where samples of an unknown signal are observed under uniform random rotations from a rotational group. In high-noise settings, we show that geometric properties of the log-likelihood function are closely related to algebraic properties of the invariant algebra of the group action. Eigenvalues of the Fisher information matrix are stratified according to a sequence of transcendence degrees in this invariant algebra, and critical points…

Abstract: I will present an algorithm which efficiently samples from the Sherrington-Kirkpatrick (SK) measure with no external field at high temperature. The approach is based on the stochastic localization process of Eldan, together with a subroutine for computing the mean vectors of a family of SK measures tilted by an appropriate external field. This approach is general and can potentially be applied to other discrete or continuous non-log-concave problems. We show that the algorithm outputs a sample within vanishing rescaled Wasserstein…

Abstract:  Phylogenetic trees are mathematical objects of great importance used to model hierarchical data and evolutionary relationships with applications in many fields including evolutionary biology and genetic epidemiology. Bayesian phylogenetic inference usually explore the posterior distribution of trees via Markov Chain Monte Carlo methods, however assessing uncertainty and summarizing distributions remains challenging for these types of structures. In this talk I will first introduce a distance metric on the space of unlabeled ranked tree shapes and genealogies. I will then…

Abstract: We consider the problem of sequential probability assignment in the Gaussian setting, where one aims to predict (or equivalently compress) a sequence of real-valued observations almost as well as the best Gaussian distribution with mean constrained to a general domain. First, in the case of a convex constraint set K, we express the hardness of the prediction problem (the minimax regret) in terms of the intrinsic volumes of K. We then establish a comparison inequality for the minimax regret…

Abstract: Contemporary generative models used in the context of unsupervised learning have primarily been designed around the construction of a map between two probability distributions that transform samples from the first into samples from the second. Advances in this domain have been governed by the introduction of algorithms or inductive biases that make learning this map, and the Jacobian of the associated change of variables, more tractable. The challenge is to choose what structure to impose on the transport to…

Abstract: Energy-based models are a recent class of probabilistic generative models wherein the distribution being learned is parametrized up to a constant of proportionality (i.e. a partition function). Fitting such models using maximum likelihood (i.e. finding the parameters which maximize the probability of the observed data) is computationally challenging, as evaluating the partition function involves a high dimensional integral. Thus, newer incarnations of this paradigm instead train other losses which obviate the need to evaluate partition functions. Prominent examples include score matching (in which we fit…

Abstract: The Paley graph is a classical number-theoretic construction of a graph that is believed to behave "pseudorandomly" in many regards. Accurately bounding the clique number of the Paley graph is a long-standing open problem in number theory, with applications to several other questions about the statistics of finite fields. I will present recent results studying the application of convex optimization and spectral graph theory to this problem, which involve understanding the extent to which the Paley graph is "spectrally…

Abstract: Sampling from high-dimensional probability distributions is a fundamental and challenging problem encountered throughout science and engineering. One of the most popular approaches to tackle such problems is the Markov chain Monte Carlo (MCMC) paradigm. While MCMC algorithms are often simple to implement and widely used in practice, analyzing the rate of convergence to stationarity, i.e. the "mixing time", remains a challenging problem in many settings. I will describe a new technique based on pairwise correlations called "spectral independence", which has been…