Adaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers. Recent emerged quantization technique has been applied to inference of deep neural networks for fast and efficient execution. However, directly applying quantization in training can cause significant accuracy loss, thus remaining an open challenge.
@article{arxiv.1911.00361,
title = {Adaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers},
author = {Xishan Zhang and Shaoli Liu and Rui Zhang and Chang Liu and Di Huang and Shiyi Zhou and Jiaming Guo and Yu Kang and Qi Guo and Zidong Du and Yunji Chen},
journal= {arXiv preprint arXiv:1911.00361},
year = {2020}
}
Comments
We would like to withdraw the manuscript because it lacks of comparisons. The main contribution is not well verified by experiments