arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.16668cs.LGcs.AI
收藏

基于程序的强化学习策略归纳

Program-Based Strategy Induction for Reinforcement Learning

Princeton Neuroscience Institute(普林斯顿神经科学研究所);Princeton University(普林斯顿大学);Department of Psychology(心理学系);Department of Computer Science(计算机科学系)

Carlos G. Correa, Thomas L. Griffiths, Nathaniel D. Daw

AI总结:

本文提出用贝叶斯程序归纳发现可解释的强化学习策略,在赌博机任务中揭示出经典增量学习难以捕捉的非对称学习、自适应随机探索和离散状态切换等策略。

摘要:

典型的学习模型假设对连续变化的决策变量(如期望奖励)进行增量估计。然而,这类模型无法捕捉人类和动物似乎表现出的更具特异性的离散启发式与策略。尽管近期利用循环网络等工具在策略发现方面取得了进展,这些工具推广了经典模型,但所得策略往往难以解释,使得与认知建立联系变得困难。我们使用贝叶斯程序归纳来发现由程序实现的策略,让策略的简洁性与其有效性进行权衡。以赌博机任务为重点,我们发现了经典增量学习难以或意想不到的策略,例如对获得奖励和未获得奖励的试次进行非对称学习、自适应视界依赖的随机探索以及离散状态切换。

英文摘要:

Typical models of learning assume incremental estimation of continuously-varying decision variables like expected rewards. However, this class of models fails to capture more idiosyncratic, discrete heuristics and strategies that people and animals appear to exhibit. Despite recent advances in strategy discovery using tools like recurrent networks that generalize the classic models, the resulting strategies are often onerous to interpret, making connections to cognition difficult to establish. We use Bayesian program induction to discover strategies implemented by programs, letting the simplicity of strategies trade off against their effectiveness. Focusing on bandit tasks, we find strategies that are difficult or unexpected with classical incremental learning, like asymmetric learning from rewarded and unrewarded trials, adaptive horizon-dependent random exploration, and discrete state switching.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑