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

While the research on convolutional neural networks (CNNs) is progressingquickly, the real-world deployment of these models is often limited bycomputing resources and memory constraints. In this paper, we address thisissue by proposing a novel filter pruning method to compress and accelerateCNNs. Our work is based on the linear relationship identified in differentfeature map subspaces via visualization of feature maps. Such linearrelationship implies that the information in CNNs is redundant. Our methodeliminates the redundancy in convolutional filters by applying subspaceclustering to feature maps. In this way, most of the representative informationin the network can be retained in each cluster. Therefore, our method providesan effective solution to filter pruning for which most existing methodsdirectly remove filters based on simple heuristics. The proposed method isindependent of the network structure, thus it can be adopted by anyoff-the-shelf deep learning libraries. Experiments on different networks andtasks show that our method outperforms existing techniques before fine-tuning,and achieves the state-of-the-art results after fine-tuning.

Exploring Linear Relationship in Feature Map Subspace for ConvNets Compression