Bi-Real Net: Binarizing Deep Network Towards Real-Network Performance

Volume: 128, Issue: 1, Pages: 202 - 219
Published: Sep 9, 2019
Abstract
In this paper, we study 1-bit convolutional neural networks (CNNs), of which both the weights and activations are binary. While being efficient, the lacking of a representational capability and the training difficulty impede 1-bit CNNs from performing as well as real-valued networks. To this end, we propose Bi-Real net with a novel training algorithm to tackle these two challenges. To enhance the representational capability, we propagate the...
Paper Details
Title
Bi-Real Net: Binarizing Deep Network Towards Real-Network Performance
Published Date
Sep 9, 2019
Volume
128
Issue
1
Pages
202 - 219
Citation AnalysisPro
  • Scinapse’s Top 10 Citation Journals & Affiliations graph reveals the quality and authenticity of citations received by a paper.
  • Discover whether citations have been inflated due to self-citations, or if citations include institutional bias.