最終的割引による時相論理の反事実経験リプレイ
Eventual Discounting Temporal Logic Counterfactual Experience Replay
線形時相論理(LTL)仕様を最大確率で満たす方策を強化学習で見つけるため、最終的割引という価値関数プロキシと、反事実推論によるオフポリシーデータ生成法を提案した。
著者: Cameron Voloshin, Abhinav Verma, Yisong Yue
分類: cs.LG, cs.AI, cs.LO, cs.RO, cs.SY, eess.SY
原文アブストラクト
Linear temporal logic (LTL) offers a simplified way of specifying tasks for policy optimization that may otherwise be difficult to describe with scalar reward functions. However, the standard RL framework can be too myopic to find maximally LTL satisfying policies. This paper makes two contributions. First, we develop a new value-function based proxy, using a technique we call eventual discounting, under which one can find policies that satisfy the LTL specification with highest achievable probability. Second, we develop a new experience replay method for generating off-policy data from on-policy rollouts via counterfactual reasoning on different ways of satisfying the LTL specification. Our experiments, conducted in both discrete and continuous state-action spaces, confirm the effectiveness of our counterfactual experience replay approach.
