Select a calendar:
Filter November Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for November 30, 2017
-
CS Colloquium: Andrew Miller (Harvard) - Advances in Monte Carlo Variational Inference
Thu, Nov 30, 2017 @ 02:00 PM - 03:20 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Andrew Miller, Harvard
Talk Title: Advances in Monte Carlo Variational Inference
Abstract: This lecture satisfies requirements for CSCI 591: Research Colloquium.
Probabilistic modeling is a natural framework for reasoning about noisy data. Well-constructed probabilistic models that combine prior knowledge with data can uncover latent structure, make predictions, and support scientific discovery. However, specifying a model and actually applying a model to data are two distinct challenges. In this talk, I will illustrate and address these challenges by presenting new models and inference methods. Monte Carlo variational inference (MCVI) is an optimization-based class of approximate inference algorithms applicable to a wide range of probabilistic models. I will present work that improves MCVI by increasing the expressiveness of approximations and the robustness of optimization. I will also present new probabilistic models developed for a variety of applied problems.
Biography: Andy Miller is a PhD candidate in computer science at Harvard University, studying statistical machine learning. He develops probabilistic models and inference methods for complex, high-dimensional data in applications ranging from astronomy to health care to sports analytics. He is currently in the final year of his program, advised by Ryan Adams (Princeton and Google Brain), Finale Doshi-Velez (Harvard), and Luke Bornn (Simon Fraser).
Host: Fei Sha
Location: Seeley G. Mudd Building (SGM) - 123
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
CS Colloquium: Zi Wang (MIT) - Bayesian Optimization and How to Scale it Up
Thu, Nov 30, 2017 @ 03:30 PM - 04:50 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Zi Wang, MIT
Talk Title: Bayesian Optimization and How to Scale it Up
Abstract: This lecture satisfies requirements for CSCI 591: Research Colloquium.
In recent years, Bayesian optimization (BO) has become a popular and effective approach to optimize an expensive black-box function with assumptions usually expressed by a Gaussian process prior. Successful applications include tuning hyper-parameters for neural networks, optimizing control parameters for robots, and designing biological experiments. Despite these successes, BO has been limited to small-scale and low-dimensional problems due to computational challenges with Gaussian processes and statistical challenges in high-dimensional settings. In this talk, I will present our recent work on scaling up BO from several aspects. First, I will introduce Max-value Entropy Search, a new BO strategy that improves sample-efficiency and obtains the first regret bound for a variant of the entropy search methods. Building on the new acquisition function, we extend our approach to high dimensions by learning the additive structures of the kernel. And finally, we propose a scalable high-dimensional BO approach that gives previously impossible results of scaling up BO to tens of thousands of observations within minutes of computation. We also show some interesting new findings on how evolutionary algorithms and BO are related, and establish novel connections among several well-known BO methods including entropy search, GP-UCB, and probability of improvement.
Biography: Zi Wang is a Ph.D. student at the MIT Computer Science and Artificial Intelligence Laboratory, advised by Stefanie Jegelka, Leslie Kaelbling and Tomás Lozano-Pérez. She received her S.M. in Electrical Engineering and Computer Science from MIT in Feb. 2016, and B.Eng. in Computer Science and Technology from Tsinghua University in Jul. 2014. Her research interests lie broadly in machine learning and artificial intelligence, currently with applications to robotics problems.
Host: Fei Sha
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Computer Science Department
