悲観的Actor-Criticにおける検証バッファの有効性
A Case for Validation Buffer in Pessimistic Actor-Critic
悲観的TD学習における批評家ネットワークの誤差蓄積を解析し、検証バッファで悲観度を調整するVPLを提案。移動・操作タスクで性能とサンプル効率を改善。
著者: Michal Nauman, Mateusz Ostaszewski, Marek Cygan
分類: cs.LG
原文アブストラクト
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.