Robust reinforcement learning: A review of foundations and recent advances

J Moos, K Hansel, H Abdulsamad, S Stark… - Machine Learning and …, 2022‏ - mdpi.com
Reinforcement learning (RL) has become a highly successful framework for learning in
Markov decision processes (MDP). Due to the adoption of RL in realistic and complex …

Tilted empirical risk minimization

T Li, A Beirami, M Sanjabi, V Smith - arxiv preprint arxiv:2007.01162, 2020‏ - arxiv.org
Empirical risk minimization (ERM) is typically designed to perform well on the average loss,
which can result in estimators that are sensitive to outliers, generalize poorly, or treat …

Conditionally elicitable dynamic risk measures for deep reinforcement learning

A Coache, S Jaimungal, Á Cartea - SIAM Journal on Financial Mathematics, 2023‏ - SIAM
We propose a novel framework to solve risk-sensitive reinforcement learning problems
where the agent optimizes time-consistent dynamic spectral risk measures. Based on the …

Reinforcement learning with dynamic convex risk measures

A Coache, S Jaimungal - Mathematical Finance, 2024‏ - Wiley Online Library
We develop an approach for solving time‐consistent risk‐sensitive stochastic optimization
problems using model‐free reinforcement learning (RL). Specifically, we assume agents …

Bridging distributional and risk-sensitive reinforcement learning with provable regret bounds

H Liang, ZQ Luo - Journal of Machine Learning Research, 2024‏ - jmlr.org
We study the regret guarantee for risk-sensitive reinforcement learning (RSRL) via
distributional reinforcement learning (DRL) methods. In particular, we consider finite …

Risk-sensitive reinforcement learning with exponential criteria

E Noorani, C Mavridis, J Baras - arxiv preprint arxiv:2212.09010, 2022‏ - arxiv.org
While reinforcement learning has shown experimental success in a number of applications,
it is known to be sensitive to noise and perturbations in the parameters of the system …

[PDF][PDF] Risk-aware transfer in reinforcement learning using successor features

M Gimelfarb, A Barreto, S Sanner… - Advances in Neural …, 2021‏ - proceedings.neurips.cc
Sample efficiency and risk-awareness are central to the development of practical
reinforcement learning (RL) for complex decision-making. The former can be addressed by …

Cascaded gaps: Towards logarithmic regret for risk-sensitive reinforcement learning

Y Fei, R Xu - International Conference on Machine Learning, 2022‏ - proceedings.mlr.press
In this paper, we study gap-dependent regret guarantees for risk-sensitive reinforcement
learning based on the entropic risk measure. We propose a novel definition of sub-optimality …

Policy gradient bayesian robust optimization for imitation learning

Z Javed, DS Brown, S Sharma, J Zhu… - International …, 2021‏ - proceedings.mlr.press
The difficulty in specifying rewards for many real-world problems has led to an increased
focus on learning rewards from human feedback, such as demonstrations. However, there …

Risk-sensitive reinforce: A monte carlo policy gradient algorithm for exponential performance criteria

E Noorani, JS Baras - 2021 60th IEEE Conference on Decision …, 2021‏ - ieeexplore.ieee.org
Risk is an inherent component of any decision making process under uncertain conditions,
and failure to consider risk may lead to significant performance degradation. We present a …