Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference

• Published in: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition pp. 2704 - 2713
• Main Authors: Jacob, Benoit, Kligys, Skirmantas, Chen, Bo, Zhu, Menglong, Tang, Matthew, Howard, Andrew, Adam, Hartwig, Kalenichenko, Dmitry
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2018
• Subjects: Arrays; Computational modeling; Hardware; Neural networks; Quantization (signal); Training
• ISSN: 1063-6919
• DOI: 10.1109/CVPR.2018.00286

The rising popularity of intelligent mobile devices and the daunting computational cost of deep learning-based models call for efficient and accurate on-device inference schemes. We propose a quantization scheme that allows inference to be carried out using integer-only arithmetic, which can be implemented more efficiently than floating point inference on commonly available integer-only hardware. We also co-design a training procedure to preserve end-to-end model accuracy post quantization. As a result, the proposed quantization scheme improves the tradeoff between accuracy and on-device latency. The improvements are significant even on MobileNets, a model family known for run-time efficiency, and are demonstrated in ImageNet classification and COCO detection on popular CPUs.