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 Network (BNN) represents convolution weights with 1-bit values,which enhances the efficiency of storage and computation. This paper ismotivated by a previously revealed phenomenon that the binary kernels insuccessful BNNs are nearly power-law distributed: their values are mostlyclustered into a small number of codewords. This phenomenon encourages us tocompact typical BNNs and obtain further close performance through learningnon-repetitive kernels within a binary kernel subspace. Specifically, we regardthe binarization process as kernel grouping in terms of a binary codebook, andour task lies in learning to select a smaller subset of codewords from the fullcodebook. We then leverage the Gumbel-Sinkhorn technique to approximate thecodeword selection process, and develop the Permutation Straight-ThroughEstimator (PSTE) that is able to not only optimize the selection processend-to-end but also maintain the non-repetitive occupancy of selectedcodewords. Experiments verify that our method reduces both the model size andbit-wise computational costs, and achieves accuracy improvements compared withstate-of-the-art BNNs under comparable budgets.

Compacting Binary Neural Networks by Sparse Kernel Selection