Saved in:
Bibliographic Details
Main Authors: Rajpal, Mohit, Tran, Lac Gia, Zhang, Yehong, Low, Bryan Kian Hsiang
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2308.00629
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929584227745792
author Rajpal, Mohit
Tran, Lac Gia
Zhang, Yehong
Low, Bryan Kian Hsiang
author_facet Rajpal, Mohit
Tran, Lac Gia
Zhang, Yehong
Low, Bryan Kian Hsiang
contents Many approaches for optimizing decision making models rely on gradient based methods requiring informative feedback from the environment. However, in the case where such feedback is sparse or uninformative, such approaches may result in poor performance. Derivative-free approaches such as Bayesian Optimization mitigate the dependency on the quality of gradient feedback, but are known to scale poorly in the high-dimension setting of complex decision making models. This problem is exacerbated if the model requires interactions between several agents cooperating to accomplish a shared goal. To address the dimensionality challenge, we propose a compact multi-layered architecture modeling the dynamics of agent interactions through the concept of role. We introduce Dependency Structure Search Bayesian Optimization to efficiently optimize the multi-layered architecture parameterized by a large number of parameters, and show an improved regret bound. Our approach shows strong empirical results under malformed or sparse reward.
format Preprint
id arxiv_https___arxiv_org_abs_2308_00629
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Dependency Structure Search Bayesian Optimization for Decision Making Models
Rajpal, Mohit
Tran, Lac Gia
Zhang, Yehong
Low, Bryan Kian Hsiang
Machine Learning
Artificial Intelligence
Many approaches for optimizing decision making models rely on gradient based methods requiring informative feedback from the environment. However, in the case where such feedback is sparse or uninformative, such approaches may result in poor performance. Derivative-free approaches such as Bayesian Optimization mitigate the dependency on the quality of gradient feedback, but are known to scale poorly in the high-dimension setting of complex decision making models. This problem is exacerbated if the model requires interactions between several agents cooperating to accomplish a shared goal. To address the dimensionality challenge, we propose a compact multi-layered architecture modeling the dynamics of agent interactions through the concept of role. We introduce Dependency Structure Search Bayesian Optimization to efficiently optimize the multi-layered architecture parameterized by a large number of parameters, and show an improved regret bound. Our approach shows strong empirical results under malformed or sparse reward.
title Dependency Structure Search Bayesian Optimization for Decision Making Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2308.00629