TBN: Convolutional Neural Network with Ternary Inputs and Binary Weights

Diwen Wan, Fumin Shen, Li Liu, Fan Zhu, Jie Qin, Ling Shao, Heng Tao Shen ; Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 315-332


Despite the remarkable success of Convolutional Neural Networks (CNNs) on generalized visual tasks, high computational and memory costs restrict their comprehensive applications on consumer electronics (e.g., portable or smart wearable devices). Recent advancements in binarized networks have demonstrated progress on reducing computational and memory costs, however, they suffer from significant performance degradation comparing to their full-precision counterparts. Thus, a highly-economical yet effective CNN that is authentically applicable to consumer electronics is at urgent need. In this work, we propose a Ternary-Binary Network (TBN), which provides an efficient approximation to standard CNNs. Based on an accelerated ternary-binary matrix multiplication, TBN replaces the arithmetical operations in standard CNNs with efficient XOR, AND and bitcount operations, and thus provides an optimal tradeoff between memory, efficiency and performance. TBN demonstrates its consistent effectiveness when applied to various CNN architectures (e.g., AlexNet and ResNet) on multiple datasets of different scales, and provides ~32x memory savings and 40x faster convolutional operations. Meanwhile, TBN can outperform XNOR-Network by up to 5.5% (top-1 accuracy) on the ImageNet classification task, and up to 4.4% (mAP score) on the PASCAL VOC object detection task.

Related Material

author = {Wan, Diwen and Shen, Fumin and Liu, Li and Zhu, Fan and Qin, Jie and Shao, Ling and Shen, Heng Tao},
title = {TBN: Convolutional Neural Network with Ternary Inputs and Binary Weights},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
month = {September},
year = {2018}