Identifiability and generalizability in constrained inverse reinforcement learning

A Schlaginhaufen… - … Conference on Machine …, 2023 - proceedings.mlr.press
Two main challenges in Reinforcement Learning (RL) are designing appropriate reward
functions and ensuring the safety of the learned policy. To address these challenges, we …

Understanding expertise through demonstrations: A maximum likelihood framework for offline inverse reinforcement learning

S Zeng, C Li, A Garcia, M Hong - arXiv preprint arXiv:2302.07457, 2023 - arxiv.org
Offline inverse reinforcement learning (Offline IRL) aims to recover the structure of rewards
and environment dynamics that underlie observed actions in a fixed, finite set of …

When demonstrations meet generative world models: A maximum likelihood framework for offline inverse reinforcement learning

S Zeng, C Li, A Garcia, M Hong - Advances in Neural …, 2024 - proceedings.neurips.cc
Offline inverse reinforcement learning (Offline IRL) aims to recover the structure of rewards
and environment dynamics that underlie observed actions in a fixed, finite set of …

Inverse Reinforcement Learning with Sub-optimal Experts

R Poiani, G Curti, AM Metelli, M Restelli - arXiv preprint arXiv:2401.03857, 2024 - arxiv.org
Inverse Reinforcement Learning (IRL) techniques deal with the problem of deducing a
reward function that explains the behavior of an expert agent who is assumed to act …

Environment Design for Inverse Reinforcement Learning

TK Buening, C Dimitrakakis - arXiv preprint arXiv:2210.14972, 2022 - arxiv.org
The task of learning a reward function from expert demonstrations suffers from high sample
complexity as well as inherent limitations to what can be learned from demonstrations in a …

Towards the Transferability of Rewards Recovered via Regularized Inverse Reinforcement Learning

A Schlaginhaufen, M Kamgarpour - arXiv preprint arXiv:2406.01793, 2024 - arxiv.org
Inverse reinforcement learning (IRL) aims to infer a reward from expert demonstrations,
motivated by the idea that the reward, rather than the policy, is the most succinct and …

Learning true objectives: Linear algebraic characterizations of identifiability in inverse reinforcement learning

ML Shehab, A Aspeel, N Aréchiga… - 6th Annual Learning …, 2024 - proceedings.mlr.press
Inverse reinforcement Learning (IRL) has emerged as a powerful paradigm for extracting
expert skills from observed behavior, with applications ranging from autonomous systems to …

Rethinking Adversarial Inverse Reinforcement Learning: From the Angles of Policy Imitation and Transferable Reward Recovery

Y Zhang, Y Zhou - arXiv preprint arXiv:2403.14593, 2024 - arxiv.org
Adversarial inverse reinforcement learning (AIRL) stands as a cornerstone approach in
imitation learning. This paper rethinks the two different angles of AIRL: policy imitation and …

Necessary and sufficient conditions for inverse reinforcement learning of Bayesian stopping time problems

K Pattanayak, V Krishnamurthy - Journal of Machine Learning Research, 2023 - jmlr.org
This paper presents an inverse reinforcement learning (IRL) framework for Bayesian
stopping time problems. By observing the actions of a Bayesian decision maker, we provide …

Inverse Reinforcement Learning: A Microeconomics-Based Approach

K Pattanayak - 2023 - search.proquest.com
The general theme of this thesis is inverse reinforcement learning (IRL) for cognitive
systems. By observing the end decisions generated from a cognitive system in multiple …