基于可重构阵列架构的强化学习计算引擎-集成技术-中国科学院深圳.PDF

第 7 卷 第6期 集 成 技 术 Vol. 7 No. 6 2018年11月 JOURNAL OF INTEGRATION TECHNOLOGY Nov. 2018 基于可重构阵列架构的强化学习计算引擎 1,2 1 2 梁明兰 王 峥 陈名松 1 (中国科学院深圳先进技术研究院 深圳 518055) 2 (桂林电子科技大学 桂林 54 1004) 摘  要  现有神经网络处理器已广泛应用于计算机视觉、自然语言处理等领域。然而，现有片上加速 方案对控制领域的强化学习算法支持较少，而基于神经网络的强化学习是智能系统决策技术的核心。 该文采用可重构阵列体系结构，通过片上配置、动作与奖励存储的系统设计方案，可实现多种神经网 络算法的灵活部署，并支持强化学习使用模式。基于65 nm CMOS 工艺的逻辑综合结果显示，处理器 2 主频为200 MHz 时，计算模块面积仅需0.32 mm ，计算功率约15.46 mW 。 关键词  人工智能；可重构阵列架构；强化学习；片上自我意识系统 中图分类号  TG 181 文献标志码  A A Reconfigurable Computing Engine for Neural Network-Based Reinforcement Learning LIANG Minglan1,2 WANG Zheng1 CHEN Mingsong2 1( Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China ) 2( Guilin University of Electronic Technology, Guilin 541004, China ) Abstract Current artificial intelligent (AI) engines are usually designed for specific supervised learning algorithms, which have been widely used in computer vision and natural language processing domains etc. However, very few AI engines have been designed to support on-chip reinforcement learning algorithms, which is the foremost algorithm kernel for decision-making subsystem of many autonomous systems. In this work, a coarse-grained reconfigurable array like AI computing engine has been designed for the deployments of both supervised and reinforcement learning through on-chip configurati