A3C-S: Automated Agent Accelerator Co-Search towards Efficient Deep Reinforcement Learning

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Fu, Yonggan, Zhang, Yongan, Li, Chaojian, Yu, Zhongzhi, Lin, Yingyan Celine
Format: Preprint
Veröffentlicht: 2021
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913634661171200
author Fu, Yonggan
Zhang, Yongan
Li, Chaojian
Yu, Zhongzhi
Lin, Yingyan Celine
author_facet Fu, Yonggan
Zhang, Yongan
Li, Chaojian
Yu, Zhongzhi
Lin, Yingyan Celine
contents Driven by the explosive interest in applying deep reinforcement learning (DRL) agents to numerous real-time control and decision-making applications, there has been a growing demand to deploy DRL agents to empower daily-life intelligent devices, while the prohibitive complexity of DRL stands at odds with limited on-device resources. In this work, we propose an Automated Agent Accelerator Co-Search (A3C-S) framework, which to our best knowledge is the first to automatically co-search the optimally matched DRL agents and accelerators that maximize both test scores and hardware efficiency. Extensive experiments consistently validate the superiority of our A3C-S over state-of-the-art techniques.
format Preprint
id arxiv_https___arxiv_org_abs_2106_06577
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle A3C-S: Automated Agent Accelerator Co-Search towards Efficient Deep Reinforcement Learning
Fu, Yonggan
Zhang, Yongan
Li, Chaojian
Yu, Zhongzhi
Lin, Yingyan Celine
Machine Learning
Driven by the explosive interest in applying deep reinforcement learning (DRL) agents to numerous real-time control and decision-making applications, there has been a growing demand to deploy DRL agents to empower daily-life intelligent devices, while the prohibitive complexity of DRL stands at odds with limited on-device resources. In this work, we propose an Automated Agent Accelerator Co-Search (A3C-S) framework, which to our best knowledge is the first to automatically co-search the optimally matched DRL agents and accelerators that maximize both test scores and hardware efficiency. Extensive experiments consistently validate the superiority of our A3C-S over state-of-the-art techniques.
title A3C-S: Automated Agent Accelerator Co-Search towards Efficient Deep Reinforcement Learning
topic Machine Learning
url https://arxiv.org/abs/2106.06577