Skipping CNN convolutions through efficient memoization

de Moura RF, Santos PC, de Lima JPC, Alves MA, Beck AC, Carro L (2019)

Publication Type: Conference contribution

Publication year: 2019

Journal

Lecture Notes in Computer Science Springer Verlag

Publisher: Springer Verlag

Book Volume: 11733 LNCS

Pages Range: 65-76

Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Event location: Samos, GRC

ISBN: 9783030275617

DOI: 10.1007/978-3-030-27562-4_5

Abstract

Convolutional Neural Networks (CNNs) have become a de-facto standard for image and video recognition. However, current software and hardware implementations targeting convolutional operations still lack embracing energy budget constraints due to the CNN intensive data processing behavior. This paper proposes a software-based memoization technique to skip entire convolution calculations. We demonstrate that, by grouping output values within proximity-based clusters, it is possible to reduce by hundreds of times the amount of memory necessary to store all the tables. Also, we present a table mapping scheme to index the input set of each convolutional layer to its output value. Our experimental results show that for a YOLOv3-tiny CNN, it is possible to achieve a speedup up to 3.5× while reducing the energy consumption to 22% of the baseline with an accuracy loss of 7.4%.

Involved external institutions

Universidade Federal do Rio Grande do Sul (UFRGS) BR Brazil (BR) Universidade Federal do Paraná (UFPR) BR Brazil (BR)

How to cite

APA:

de Moura, R.F., Santos, P.C., de Lima, J.P.C., Alves, M.A., Beck, A.C., & Carro, L. (2019). Skipping CNN convolutions through efficient memoization. In Dionisios N. Pnevmatikatos, Maxime Pelcat, Matthias Jung (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 65-76). Samos, GRC: Springer Verlag.

MLA:

de Moura, Rafael Fão, et al. "Skipping CNN convolutions through efficient memoization." Proceedings of the 19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019, Samos, GRC Ed. Dionisios N. Pnevmatikatos, Maxime Pelcat, Matthias Jung, Springer Verlag, 2019. 65-76.

BibTeX: Download