Saved in:
Bibliographic Details
Main Authors: Zollicoffer, Geigh, Eaton, Kenneth, Balloch, Jonathan, Kim, Julia, Zhou, Wei, Wright, Robert, Riedl, Mark O.
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2310.08731
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913862537707520
author Zollicoffer, Geigh
Eaton, Kenneth
Balloch, Jonathan
Kim, Julia
Zhou, Wei
Wright, Robert
Riedl, Mark O.
author_facet Zollicoffer, Geigh
Eaton, Kenneth
Balloch, Jonathan
Kim, Julia
Zhou, Wei
Wright, Robert
Riedl, Mark O.
contents Reinforcement learning (RL) using world models has found significant recent successes. However, when a sudden change to world mechanics or properties occurs then agent performance and reliability can dramatically decline. We refer to the sudden change in visual properties or state transitions as novelties. Implementing novelty detection within generated world model frameworks is a crucial task for protecting the agent when deployed. In this paper, we propose straightforward bounding approaches to incorporate novelty detection into world model RL agents, by utilizing the misalignment of the world model's hallucinated states and the true observed states as an anomaly score. We provide effective approaches to detecting novelties in a distribution of transitions learned by an agent in a world model. Finally, we show the advantage of our work in a novel environment compared to traditional machine learning novelty detection methods as well as currently accepted RL focused novelty detection algorithms.
format Preprint
id arxiv_https___arxiv_org_abs_2310_08731
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Novelty Detection in Reinforcement Learning with World Models
Zollicoffer, Geigh
Eaton, Kenneth
Balloch, Jonathan
Kim, Julia
Zhou, Wei
Wright, Robert
Riedl, Mark O.
Artificial Intelligence
Machine Learning
Systems and Control
Reinforcement learning (RL) using world models has found significant recent successes. However, when a sudden change to world mechanics or properties occurs then agent performance and reliability can dramatically decline. We refer to the sudden change in visual properties or state transitions as novelties. Implementing novelty detection within generated world model frameworks is a crucial task for protecting the agent when deployed. In this paper, we propose straightforward bounding approaches to incorporate novelty detection into world model RL agents, by utilizing the misalignment of the world model's hallucinated states and the true observed states as an anomaly score. We provide effective approaches to detecting novelties in a distribution of transitions learned by an agent in a world model. Finally, we show the advantage of our work in a novel environment compared to traditional machine learning novelty detection methods as well as currently accepted RL focused novelty detection algorithms.
title Novelty Detection in Reinforcement Learning with World Models
topic Artificial Intelligence
Machine Learning
Systems and Control
url https://arxiv.org/abs/2310.08731