SENTINEL: Taming Uncertainty with Ensemble-based Distributional Reinforcement Learning
Paper in proceeding, 2022

In this paper, we consider risk-sensitive sequential decision-making in Reinforcement Learning (RL).
Our contributions are two-fold. First, we introduce a novel and coherent quantification of risk, namely composite risk, which quantifies the joint effect of aleatory and epistemic risk during the learning process.
Existing works considered either aleatory or epistemic risk individually, or as an additive combination.
We prove that the additive formulation is a particular case of the composite risk when the epistemic risk measure is replaced with expectation.
Thus, the composite risk is more sensitive to both aleatory and epistemic uncertainty than the individual and additive formulations.
We also propose an algorithm, SENTINEL-K, based on ensemble bootstrapping and distributional RL for representing epistemic and aleatory uncertainty respectively. The ensemble of K learners uses Follow The Regularised Leader (FTRL) to aggregate the return distributions and obtain the composite risk.
We experimentally verify that SENTINEL-K estimates the return distribution better, and while used with composite risk estimates, demonstrates higher risk-sensitive performance than state-of-the-art risk-sensitive and distributional RL algorithms.

Epistemic uncertainty

Reinforcement Learning

Ensemble methods

Author

Hannes Eriksson

Zenseact AB

Chalmers, Computer Science and Engineering (Chalmers), Data Science

Debabrota Basu

Cent Lille CRIStAL

Inria Lille Nord Europe

Mina Alibeigi

Zenseact AB

Christos Dimitrakakis

Chalmers, Computer Science and Engineering (Chalmers), Data Science

University of Oslo

University of Neuchatel

Proceedings of Machine Learning Research

2640-3498 (ISSN)

Vol. 180 631-640
9781713863298 (ISBN)

38th Conference on Uncertainty in Artificial Intelligence
Eindhoven, Netherlands,

Subject Categories

Other Computer and Information Science

Computer Vision and Robotics (Autonomous Systems)

More information

Latest update

1/23/2023