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 (BNN) have been studied extensively since they rundramatically faster at lower memory and power consumption than floating-pointnetworks, thanks to the efficiency of bit operations. However, contemporaryBNNs whose weights and activations are both single bits suffer from severeaccuracy degradation. To understand why, we investigate the representationability, speed and bias/variance of BNNs through extensive experiments. Weconclude that the error of BNNs is predominantly caused by the intrinsicinstability (training time) and non-robustness (train & test time). Inspired bythis investigation, we propose the Binary Ensemble Neural Network (BENN) whichleverages ensemble methods to improve the performance of BNNs with limitedefficiency cost. While ensemble techniques have been broadly believed to beonly marginally helpful for strong classifiers such as deep neural networks,our analyses and experiments show that they are naturally a perfect fit toboost BNNs. We find that our BENN, which is faster and much more robust thanstate-of-the-art binary networks, can even surpass the accuracy of thefull-precision floating number network with the same architecture.

Binary Ensemble Neural Network: More Bits per Network or More Networks per Bit?