This work presents a framework for dynamic energy reduction in hardware accelerators for convolutional neural networks (CNNs). The key idea is based on the early prediction of the features that may be important, with the deactivation of computations related to unimportant features and static bitwidth reduction. The former is applied in late layers of the CNN, while the latter is more effective in the early layers. The procedure includes a methodology for automated threshold tuning to detect feature activation. For various state-of-the-art neural networks, the results show that energy savings of up to about 30% are achievable, after accounting for all implementation overheads, with a small loss in the accuracy.
|Original language||English (US)|
|Number of pages||14|
|Journal||IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems|
|State||Published - Jul 1 2021|
Bibliographical noteFunding Information:
Manuscript received February 9, 2020; revised June 28, 2020; accepted July 31, 2020. Date of publication August 13, 2020; date of current version June 18, 2021. This work was supported in part by the NSF under Award CCF-1525925, Award CCF-1763761, and Award CCF-1525749. A preliminary version of this work appeared in . This article was recommended by Associate Editor H. Li. (Corresponding author: Farhana Sharmin Snigdha.) Farhana Sharmin Snigdha, Susmita Dey Manasi, and Sachin S. Sapatnekar are with the Department of ECE, University of Minnesota, Minneapolis, MN 55455 USA (e-mail: email@example.com; firstname.lastname@example.org; email@example.com).
Dr. Hu has received two conference Best Paper Awards, the IBM Invention Achievement Award, and the Humboldt Research Fellowship. He has served on the editorial boards of the IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS and the ACM Transactions on Design Automation of Electronic Systems.
© 1982-2012 IEEE.
- Convolutional neural network (CNN)
- deep learning
- energy optimization
- low-power design
- neural network