ERDSE: efficient reinforcement learning based design space exploration method for CNN accelerator on resource limited platform

ERDSE: efficient reinforcement learning based design space exploration method for CNN accelerator on resource limited platform

Convolutional Neural Network (CNN) accelerator design on resource limited platform faces the challenge of lacking efficient design space exploration (DSE) method because of its huge and irregular design space. Numerous parameters belong to accelerator architecture and dataflow mode jointly construct...

Full description

Bibliographic Details
Main Authors: Kaijie Feng, Xiaoya Fan, Jianfeng An, Xiping Wang, Kaiyue Di, Jiangfei Li, Minghao Lu, Chuxi Li
Format: Article
Language:English
Published: Elsevier 2021-06-01
Series:Graphics and Visual Computing
Subjects:
CNN accelerator
Customized computing
Design space exploration
Reinforcement learning
Online Access:http://www.sciencedirect.com/science/article/pii/S2666629421000073

Internet

http://www.sciencedirect.com/science/article/pii/S2666629421000073

Similar Items