FT-CNN: Algorithm-Based Fault Tolerance for Convolutional Neural Networks

• Published in: IEEE transactions on parallel and distributed systems Vol. 32; no. 7; pp. 1677 - 1689
• Main Authors: Zhao, Kai, Di, Sheng, Li, Sihuan, Liang, Xin, Zhai, Yujia, Chen, Jieyang, Ouyang, Kaiming, Cappello, Franck, Chen, Zizhong
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.07.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithm-based fault tolerance; Algorithms; Artificial neural networks; Convolution; deep learning; Error correcting codes; Error correction; Error correction codes; Fault tolerance; Fault tolerant systems; High temperature; high-performance computing; Inference; Kernel; Mathematical model; Multiplication; Neural networks; reliability; Run time (computers); Runtime; Safety critical; silent data corruption; Soft errors; Workflow
• ISSN: 1045-9219; 1558-2183
• DOI: 10.1109/TPDS.2020.3043449

Convolutional neural networks (CNNs) are becoming more and more important for solving challenging and critical problems in many fields. CNN inference applications have been deployed in safety-critical systems, which may suffer from soft errors caused by high-energy particles, high temperature, or abnormal voltage. Of critical importance is ensuring the stability of the CNN inference process against soft errors. Traditional fault tolerance methods are not suitable for CNN inference because error-correcting code is unable to protect computational components, instruction duplication techniques incur high overhead, and existing algorithm-based fault tolerance (ABFT) techniques cannot protect all convolution implementations. In this article, we focus on how to protect the CNN inference process against soft errors as efficiently as possible, with the following three contributions. (1) We propose several systematic ABFT schemes based on checksum techniques and analyze their fault protection ability and runtime thoroughly. Unlike traditional ABFT based on matrix-matrix multiplication, our schemes support any convolution implementations. (2) We design a novel workflow integrating all the proposed schemes to obtain a high detection/correction ability with limited total runtime overhead. (3) We perform our evaluation using ImageNet with well-known CNN models including AlexNet, VGG-19, ResNet-18, and YOLOv2. Experimental results demonstrate that our implementation can handle soft errors with very limited runtime overhead (4%<inline-formula><tex-math notation="LaTeX">\sim</tex-math> <mml:math><mml:mo>∼</mml:mo></mml:math><inline-graphic xlink:href="zhao-ieq1-3043449.gif"/> </inline-formula>8% in both error-free and error-injected situations).