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

We present a new approach and a novel architecture, termed WSNet, forlearning compact and efficient deep neural networks. Existing approachesconventionally learn full model parameters independently and then compress themvia ad hoc processing such as model pruning or filter factorization.Alternatively, WSNet proposes learning model parameters by sampling from acompact set of learnable parameters, which naturally enforces {parametersharing} throughout the learning process. We demonstrate that such a novelweight sampling approach (and induced WSNet) promotes both weights andcomputation sharing favorably. By employing this method, we can moreefficiently learn much smaller networks with competitive performance comparedto baseline networks with equal numbers of convolution filters. Specifically,we consider learning compact and efficient 1D convolutional neural networks foraudio classification. Extensive experiments on multiple audio classificationdatasets verify the effectiveness of WSNet. Combined with weight quantization,the resulted models are up to 180 times smaller and theoretically up to 16times faster than the well-established baselines, without noticeableperformance drop.

WSNet: Compact and Efficient Networks Through Weight Sampling