English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Binary Neural Networks (BNNs) show promising progress in reducingcomputational and memory costs but suffer from substantial accuracy degradationcompared to their real-valued counterparts on large-scale datasets, e.g.,ImageNet. Previous work mainly focused on reducing quantization errors ofweights and activations, whereby a series of approximation methods andsophisticated training tricks have been proposed. In this work, we make severalobservations that challenge conventional wisdom. We revisit some commonly usedtechniques, such as scaling factors and custom gradients, and show that thesemethods are not crucial in training well-performing BNNs. On the contrary, wesuggest several design principles for BNNs based on the insights learned anddemonstrate that highly accurate BNNs can be trained from scratch with a simpletraining strategy. We propose a new BNN architecture BinaryDenseNet, whichsignificantly surpasses all existing 1-bit CNNs on ImageNet without tricks. Inour experiments, BinaryDenseNet achieves 18.6% and 7.6% relative improvementover the well-known XNOR-Network and the current state-of-the-art Bi-Real Netin terms of top-1 accuracy on ImageNet, respectively.

Back to Simplicity: How to Train Accurate BNNs from Scratch?