no code implementations • 17 Dec 2021 • Jasmina Gajcin, Rahul Nair, Tejaswini Pedapati, Radu Marinescu, Elizabeth Daly, Ivana Dusparic
In complex tasks where the reward function is not straightforward and consists of a set of objectives, multiple reinforcement learning (RL) policies that perform task adequately, but employ different strategies can be trained by adjusting the impact of individual objectives on reward function.
1 code implementation • 21 Mar 2022 • Jasmina Gajcin, Ivana Dusparic
We propose ReCCoVER, an algorithm which detects causal confusion in agent's reasoning before deployment, by executing its policy in alternative environments where certain correlations between features do not hold.
no code implementations • 21 Oct 2022 • Jasmina Gajcin, Ivana Dusparic
Additionally, we explore the differences between counterfactual explanations in supervised learning and RL and identify the main challenges that prevent the adoption of methods from supervised in reinforcement learning.
no code implementations • 9 Feb 2024 • Jasmina Gajcin, Ivana Dusparic
In this work, we propose ACTER (Actionable Counterfactual Sequences for Explaining Reinforcement Learning Outcomes), an algorithm for generating counterfactual sequences that provides actionable advice on how failure can be avoided.
1 code implementation • 8 Mar 2023 • Jasmina Gajcin, Ivana Dusparic
In this work, we propose RACCER, the first RL-specific approach to generating counterfactual explanations for the behavior of RL agents.
1 code implementation • 30 Aug 2023 • Jasmina Gajcin, James McCarthy, Rahul Nair, Radu Marinescu, Elizabeth Daly, Ivana Dusparic
Our approach allows the user to provide trajectory-level feedback on agent's behavior during training, which can be integrated as a reward shaping signal in the following training iteration.
2 code implementations • 2 Nov 2022 • Tom He, Jasmina Gajcin, Ivana Dusparic
We apply CausalCF to complex robotic tasks and show that it improves the RL agent's robustness using CausalWorld.