January 18, 2024 10:34

Abstract

Date, Time, and Venue:
January 25, 2024: 2:00 pm – 3:00 pm (JST)
Venue: Online and the Open Space at the RIKEN AIP Nihonbashi office*
The Open Space is only available to RIKEN AIP researchers
Language: English

Title: Bayesian Optimization over High-Dimensional Combinatorial Spaces

Speaker: Dr. Seungjin Choi (Director of the Machine Learning Lab, Intellicode)

Abstract: Bayesian optimization has proven to be a highly effective and sample-efficient approach for finding global optima of an expensive-to-evaluate black-box function. Its applications span a wide range, including hyperparameter optimization, automated machine learning, protein/DNA sequence design, material design. The standard Bayesian optimization leverages Gaussian process regression for constructing a surrogate model to estimate a landscape of the black-box objective, assuming decision variables are continuous in Euclidean space. In this talk, we address two critical issues in Bayesian optimization: (1) Bayesian optimization over high-dimensional spaces; (2) Bayesian optimization over combinatorial spaces. We will provide an overview of some of existing methods on these topics before delving into our own solution – combinatorial Bayesian optimization with random projection. Our presentation will introduce the algorithm and its regret analysis.

Short bio: Seungjin Choi is currently a director of machine learning lab in Intellicode, Korea. Before he took a research director position, he spent about 20 years as a professor of computer science in POSTECH, Korea. He also held advisory professor in Shinhan Card Big Data Center, Samsung Research, Samsung Advanced Institute of Technology. His current research interests include Bayesian optimization, uncertainty quantification, conformal inference.


All participants are required to agree with the AIP Seminar Series Code of Conduct.
Please see the URL below.
https://aip.riken.jp/event-list/termsofparticipation/?lang=en

RIKEN AIP will expect adherence to this code throughout the event. We expect cooperation from all participants to help ensure a safe environment for everybody.


More Information

Date January 25, 2024 (Thu) 14:00 - 15:00
URL https://c5dc59ed978213830355fc8978.doorkeeper.jp/events/168586

Related Laboratories

last updated on February 14, 2024 14:33Laboratory