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

Filter is the key component in modern convolutional neural networks (CNNs).However, since CNNs are usually over-parameterized, a pre-trained networkalways contain some invalid (unimportant) filters. These filters haverelatively small $l_{1}$ norm and contribute little to the output(\textbf{Reason}). While filter pruning removes these invalid filters forefficiency consideration, we tend to reactivate them to improve therepresentation capability of CNNs. In this paper, we introduce filter grafting(\textbf{Method}) to achieve this goal. The activation is processed by graftingexternal information (weights) into invalid filters. To better perform thegrafting, we develop a novel criterion to measure the information of filtersand an adaptive weighting strategy to balance the grafted information amongnetworks. After the grafting operation, the network has fewer invalid filterscompared with its initial state, enpowering the model with more representationcapacity. Meanwhile, since grafting is operated reciprocally on all networksinvolved, we find that grafting may lose the information of valid filters whenimproving invalid filters. To gain a universal improvement on both valid andinvalid filters, we compensate grafting with distillation(\textbf{Cultivation}) to overcome the drawback of grafting . Extensiveexperiments are performed on the classification and recognition tasks to showthe superiority of our method. Code is available at\textcolor{black}{\emph{https://github.com/fxmeng/filter-grafting}}.

Filter Grafting for Deep Neural Networks: Reason, Method, and Cultivation