Bayesian Inverse Transition Learning: Learning Dynamics From Near-Optimal Trajectories

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Benac, Leo, Sharma, Abhishek, Parbhoo, Sonali, Doshi-Velez, Finale
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911627294539776
author Benac, Leo
Sharma, Abhishek
Parbhoo, Sonali
Doshi-Velez, Finale
author_facet Benac, Leo
Sharma, Abhishek
Parbhoo, Sonali
Doshi-Velez, Finale
contents We consider the problem of estimating the transition dynamics $T^*$ from near-optimal expert trajectories in the context of offline model-based reinforcement learning. We develop a novel constraint-based method, Inverse Transition Learning, that treats the limited coverage of the expert trajectories as a \emph{feature}: we use the fact that the expert is near-optimal to inform our estimate of $T^*$. We integrate our constraints into a Bayesian approach. Across both synthetic environments and real healthcare scenarios like Intensive Care Unit (ICU) patient management in hypotension, we demonstrate not only significant improvements in decision-making, but that our posterior can inform when transfer will be successful.
format Preprint
id arxiv_https___arxiv_org_abs_2411_05174
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Bayesian Inverse Transition Learning: Learning Dynamics From Near-Optimal Trajectories
Benac, Leo
Sharma, Abhishek
Parbhoo, Sonali
Doshi-Velez, Finale
Machine Learning
Artificial Intelligence
We consider the problem of estimating the transition dynamics $T^*$ from near-optimal expert trajectories in the context of offline model-based reinforcement learning. We develop a novel constraint-based method, Inverse Transition Learning, that treats the limited coverage of the expert trajectories as a \emph{feature}: we use the fact that the expert is near-optimal to inform our estimate of $T^*$. We integrate our constraints into a Bayesian approach. Across both synthetic environments and real healthcare scenarios like Intensive Care Unit (ICU) patient management in hypotension, we demonstrate not only significant improvements in decision-making, but that our posterior can inform when transfer will be successful.
title Bayesian Inverse Transition Learning: Learning Dynamics From Near-Optimal Trajectories
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2411.05174