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

Compression has emerged as one of the essential deep learning researchtopics, especially for the edge devices that have limited computation power andstorage capacity. Among the main compression techniques, low-rank compressionvia matrix factorization has been known to have two problems. First, anextensive tuning is required. Second, the resulting compression performance istypically not impressive. In this work, we propose a low-rank compressionmethod that utilizes a modified beam-search for an automatic rank selection anda modified stable rank for a compression-friendly training. The resulting BSR(Beam-search and Stable Rank) algorithm requires only a single hyperparameterto be tuned for the desired compression ratio. The performance of BSR in termsof accuracy and compression ratio trade-off curve turns out to be superior tothe previously known low-rank compression methods. Furthermore, BSR can performon par with or better than the state-of-the-art structured pruning methods. Aswith pruning, BSR can be easily combined with quantization for an additionalcompression.

A Highly Effective Low-Rank Compression of Deep Neural Networks with Modified Beam-Search and Modified Stable Rank