Abstract
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 reliably detect OOD situations.
Dokumententyp: | Zeitschriftenartikel |
---|---|
Fakultät: | Mathematik, Informatik und Statistik > Informatik |
Themengebiete: | 000 Informatik, Informationswissenschaft, allgemeine Werke > 004 Informatik |
Sprache: | Englisch |
Dokumenten ID: | 89079 |
Datum der Veröffentlichung auf Open Access LMU: | 25. Jan. 2022, 09:28 |
Letzte Änderungen: | 25. Jan. 2022, 09:28 |