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

Pruning is an efficient model compression technique to remove redundancy inthe connectivity of deep neural networks (DNNs). Computations using sparsematrices obtained by pruning parameters, however, exhibit vastly differentparallelism depending on the index representation scheme. As a result,fine-grained pruning has not gained much attention due to its irregular indexform leading to large memory footprint and low parallelism for convolutions andmatrix multiplications. In this paper, we propose a new network pruningtechnique that generates a low-rank binary index matrix to compress index datawhile decompressing index data is performed by simple binary matrixmultiplication. This proposed compression method finds a particularfine-grained pruning mask that can be decomposed into two binary matrices. Wealso propose a tile-based factorization technique that not only lowers memoryrequirements but also enhances compression ratio. Various DNN models can bepruned with much fewer indexes compared to previous sparse matrix formats whilemaintaining the same pruning rate.

Network Pruning for Low-Rank Binary Indexing