Training binary neural networks without floating point precision

Federico Fontana

Training binary neural networks without floating point precision

Machine Learning 2023-11-01 v1 Artificial Intelligence Neural and Evolutionary Computing

Authors: Federico Fontana

Abstract

The main goal of this work is to improve the efficiency of training binary neural networks, which are low latency and low energy networks. The main contribution of this work is the proposal of two solutions comprised of topology changes and strategy training that allow the network to achieve near the state-of-the-art performance and efficient training. The time required for training and the memory required in the process are two factors that contribute to efficient training.

Keywords

binary neural network neural network training neural network

Cite

@article{arxiv.2310.19815,
  title  = {Training binary neural networks without floating point precision},
  author = {Federico Fontana},
  journal= {arXiv preprint arXiv:2310.19815},
  year   = {2023}
}

Comments

74 pages, Master's thesis

Related papers

View all related →