Saved in:
Bibliographic Details
Main Authors: Palattuparambil, Ajsal Shereef, Karimpanal, Thommen George, Rana, Santu
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.20016
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915501532250112
author Palattuparambil, Ajsal Shereef
Karimpanal, Thommen George
Rana, Santu
author_facet Palattuparambil, Ajsal Shereef
Karimpanal, Thommen George
Rana, Santu
contents Deep reinforcement learning (RL) policies, although optimal in terms of task rewards, may not align with the personal preferences of human users. To ensure this alignment, a naive solution would be to retrain the agent using a reward function that encodes the user's specific preferences. However, such a reward function is typically not readily available, and as such, retraining the agent from scratch can be prohibitively expensive. We propose a more practical approach - to adapt the already trained policy to user-specific needs with the help of human feedback. To this end, we infer the user's intent through trajectory-level feedback and combine it with the trained task policy via a theoretically grounded dynamic policy fusion approach. As our approach collects human feedback on the very same trajectories used to learn the task policy, it does not require any additional interactions with the environment, making it a zero-shot approach. We empirically demonstrate in a number of environments that our proposed dynamic policy fusion approach consistently achieves the intended task while simultaneously adhering to user-specific needs.
format Preprint
id arxiv_https___arxiv_org_abs_2409_20016
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Dynamic Policy Fusion for User Alignment Without Re-Interaction
Palattuparambil, Ajsal Shereef
Karimpanal, Thommen George
Rana, Santu
Artificial Intelligence
Machine Learning
Deep reinforcement learning (RL) policies, although optimal in terms of task rewards, may not align with the personal preferences of human users. To ensure this alignment, a naive solution would be to retrain the agent using a reward function that encodes the user's specific preferences. However, such a reward function is typically not readily available, and as such, retraining the agent from scratch can be prohibitively expensive. We propose a more practical approach - to adapt the already trained policy to user-specific needs with the help of human feedback. To this end, we infer the user's intent through trajectory-level feedback and combine it with the trained task policy via a theoretically grounded dynamic policy fusion approach. As our approach collects human feedback on the very same trajectories used to learn the task policy, it does not require any additional interactions with the environment, making it a zero-shot approach. We empirically demonstrate in a number of environments that our proposed dynamic policy fusion approach consistently achieves the intended task while simultaneously adhering to user-specific needs.
title Dynamic Policy Fusion for User Alignment Without Re-Interaction
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2409.20016