Processing Element Architecture Design for Deep Reinforcement Learning with Flexible Block Floating Point Exploiting Signal Statistics

Juyn Da Su, Pei Yun Tsai

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Deep reinforcement learning is a technique that allows the agent to have evolving learning capability for unknown environments and thus has the potential to surpass human expertise. The hardware architecture for DRL supporting on-line Q-learning and on-line training is presented in this paper. Two processing element (PE) arrays are used for handling evaluation network and target network respectively. Through configuration of two modes for PE operations, all required forward and backward computations can be accomplished and the number of processing cycles can be derived. Due to the precision required for on-line Q-learning and training, we propose flexible block floating-point (FBFP) to reduce the overhead of floating-point adders. The FBFP exploits different signal statistics during the learning process. Furthermore, the respective block exponents of gradients are adjusted following the variation of temporal difference (TD) error to reserve resolution. From the simulation results, the FBFP multiplier-and-accumulator (MAC) can reduce 15.8% of complexity compared to FP MAC while good learning performance can be maintained.

Original languageEnglish
Title of host publication2020 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2020 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages82-87
Number of pages6
ISBN (Electronic)9789881476883
StatePublished - 7 Dec 2020
Event2020 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2020 - Virtual, Auckland, New Zealand
Duration: 7 Dec 202010 Dec 2020

Publication series

Name2020 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2020 - Proceedings

Conference

Conference2020 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2020
Country/TerritoryNew Zealand
CityVirtual, Auckland
Period7/12/2010/12/20

Keywords

  • architecture design
  • Block floating-point
  • deep Q network
  • reinforcement learning

Fingerprint

Dive into the research topics of 'Processing Element Architecture Design for Deep Reinforcement Learning with Flexible Block Floating Point Exploiting Signal Statistics'. Together they form a unique fingerprint.

Cite this