Analysis and Solution of CNN Accuracy Reduction over Channel Loop Tiling
Yesung Kang1,a, Yoonho Park2, Sunghoon Kim2, Eunji Kwon2, Taeho Lim3, Sangyun Oh4, Mingyu Woo5 and Seokhyeong Kang2,b
1Future IT Innovation Laboratory
ayeskang@postech.ac.kr
2EE Department, POSTECH, South Korea
bshkang@postech.ac.kr
3SK Hynix Inc., South Korea
4CSE Department, UNIST, South Korea
5ECE Department, UC San Diego, La Jolla, CA, USA
ABSTRACT
Owing to the growth of the size of convolutional neural networks (CNNs), quantization and loop tiling (also called loop breaking) are mandatory to implement CNN on an embedded system. However, channel loop tiling of quantized CNNs induces unexpected errors. We explain why channel loop tiling of quantized CNNs induces the unexpected errors, and how the errors affect the accuracy of state-of-the-art CNNs. We also propose a method to recover accuracy under channel tiling by compressing and decompressing the most-significant bits of partial sums. Using the proposed method, we can recover accuracy by 12.3% with only 1% circuit area overhead and an additional 2% of power consumption.
