{"type":"ScholarlyArticle","title":"A Case for Validation Buffer in Pessimistic Actor-Critic","authors":[{"name":"Michał Nauman","slug":"micha-nauman","given":"Michał","family":"Nauman","affiliation":""},{"name":"Mateusz Ostaszewski","slug":"mateusz-ostaszewski","given":"Mateusz","family":"Ostaszewski","affiliation":""},{"name":"Marek Cygan","slug":"marek-cygan","given":"Marek","family":"Cygan","affiliation":""}],"abstract":"In this paper, we investigate the issue of error accumulation in critic networks updated via pessimistic temporal difference objectives. We show that the critic approximation error can be approximated via a recursive fixed-point model similar to that of the Bellman value. We use such recursive definition to retrieve the conditions under which the pessimistic critic is unbiased. Building on these insights, we propose Validation Pessimism Learning (VPL) algorithm. VPL uses a small validation buffer to adjust the levels of pessimism throughout the agent training, with the pessimism set such that the approximation error of the critic targets is minimized. We investigate the proposed approach on a variety of locomotion and manipulation tasks and report improvements in sample efficiency and performance.","doi":"10.24963/ijcai.2025/665","url":"https://ijcai.org/proceedings/2025/665","pdf":"https://ijcai.org/proceedings/2025/0665.pdf","markdown":"https://ijcai.org/proceedings/2025/665.md","bibtex":"https://ijcai.org/proceedings/2025/bibtex/665","volume":{"slug":"2025","title":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2025","url":"https://ijcai.org/proceedings/2025/","year":2025,"location":"Montreal, Canada","dates":"16-22 August 2025","editors":"James Kwok","isbn":"978-1-956792-06-5","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2025"},"section":"Machine Learning","track":"Main Track","number":665,"pages":{"first":5976,"last":5984},"keywords":["Robotics: ROB: Behavior and control","Planning and Scheduling: PS: Markov decisions processes","Machine Learning: ML: Reinforcement learning","Robotics: ROB: Learning in robotics"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}