arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.09754cs.LGcs.AI
收藏

SimBa:用于深度强化学习参数规模扩展的简单性偏置

SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learning

KAIST(韩国科学技术院);Sony AI(索尼人工智能);KRAFTON;Coventry University(科文特大学);UT Austin(得克萨斯大学奥斯汀分校)

Hojoon Lee, Dongyoon Hwang, Donghu Kim, Hyunseung Kim, Jun Jet Tai, Kaushik Subramanian, Peter R. Wurman, Jaegul Choo, Peter Stone, Takuma Seno

AI总结:

提出SimBa架构,通过观测归一化、残差前馈块和层归一化注入简单性偏置来扩展深度强化学习的网络参数,显著提升了多种RL算法的样本效率,并在多个基准上以高计算效率达到或超越现有最优方法。

摘要:

近期CV和NLP的进展很大程度上是由扩大网络参数数量所驱动的,尽管传统理论认为较大的网络容易过拟合。这些大型网络通过引入产生简单性偏置的组件来避免过拟合,引导模型趋向简单且可泛化的解。然而,在深度RL中,网络的设计与扩展研究较少。受此启发,我们提出了SimBa,一种旨在通过注入简单性偏置来扩展深度RL参数的架构。SimBa包含三个组件:(i)利用运行统计量标准化输入的观测归一化层,(ii)提供从输入到输出线性路径的残差前馈块,以及(iii)控制特征幅度的层归一化。通过用SimBa扩展参数,多种深度RL算法(包括off-policy、on-policy和无监督方法)的样本效率均得到持续提升。此外,仅通过将SimBa架构集成到SAC中,它便以高计算效率在DMC、MyoSuite和HumanoidBench上匹配或超越了最先进的深度RL方法。这些结果证明了SimBa在多种RL算法和环境中具有广泛的适用性和有效性。

英文摘要:

Recent advances in CV and NLP have been largely driven by scaling up the number of network parameters, despite traditional theories suggesting that larger networks are prone to overfitting. These large networks avoid overfitting by integrating components that induce a simplicity bias, guiding models toward simple and generalizable solutions. However, in deep RL, designing and scaling up networks have been less explored. Motivated by this opportunity, we present SimBa, an architecture designed to scale up parameters in deep RL by injecting a simplicity bias. SimBa consists of three components: (i) an observation normalization layer that standardizes inputs with running statistics, (ii) a residual feedforward block to provide a linear pathway from the input to output, and (iii) a layer normalization to control feature magnitudes. By scaling up parameters with SimBa, the sample efficiency of various deep RL algorithms-including off-policy, on-policy, and unsupervised methods-is consistently improved. Moreover, solely by integrating SimBa architecture into SAC, it matches or surpasses state-of-the-art deep RL methods with high computational efficiency across DMC, MyoSuite, and HumanoidBench. These results demonstrate SimBa's broad applicability and effectiveness across diverse RL algorithms and environments.

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。

↑