• search hit 11 of 38
Back to Result List

Uncertainty-based out-of-distribution classification in deep reinforcement learning

  • Robustness to out-of-distribution (OOD) data is an important goal in building reliable machine learning systems. As a first step towards a solution, we consider the problem of detecting such data in a value-based deep reinforcement learning (RL) setting. Modelling this problem as a one-class classification problem, we propose a framework for uncertainty-based OOD classification: UBOOD. It is based on the effect that an agent’s epistemic uncertainty is reduced for situations encountered during training (in-distribution), and thus lower than for unencountered (OOD) situations. Being agnostic towards the approach used for estimating epistemic uncertainty, combinations with different uncertainty estimation methods, e.g. approximate Bayesian inference methods or ensembling techniques are possible. Evaluation shows that the framework produces reliable classification results when combined with ensemble-based estimators, while the combination with concrete dropout-based estimators fails to r eliably detect OOD situations.

Download full text files

Export metadata

Additional Services

Search Google Scholar
Metadaten
Author:Andreas Sedlmeier, Thomas Gabor, Thomy Phan, Lenz BelznerORCiD, Claudia Linnhoff-Popien
Language:English
Document Type:Conference Paper
Conference:12th International Conference on Agents and Artificial Intelligence, Valletta (Malta), 22.-24.02.2020
Year of first Publication:2020
published in (English):Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART
Editor(s):Ana Rocha, Luc Steels, Jaap van den Herik
Publisher:SciTePress
Place of publication:Setúbal
ISBN:978-989-758-395-7
ISSN:2184-433X
First Page:522
Last Page:529
Review:peer-review
Version:published
Tag:Deep Reinforcement Learning; Out-of-Distribution Classification; Uncertainty in AI
Related Identifier:https://doi.org/10.5220/0008949905220529
Licence (German):License Logo Creative Commons BY-NC-ND 4.0
Release Date:2022/03/15