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

A wide range of verification methods have been proposed to verify the safetyproperties of deep neural networks ensuring that the networks functioncorrectly in critical applications. However, many well-known verification toolsstill struggle with complicated network architectures and large network sizes.In this work, we propose a network reduction technique as a pre-processingmethod prior to verification. The proposed method reduces neural networks viaeliminating stable ReLU neurons, and transforming them into a sequential neuralnetwork consisting of ReLU and Affine layers which can be handled by the mostverification tools. We instantiate the reduction technique on thestate-of-the-art complete and incomplete verification tools, includingalpha-beta-crown, VeriNet and PRIMA. Our experiments on a large set ofbenchmarks indicate that the proposed technique can significantly reduce neuralnetworks and speed up existing verification tools. Furthermore, the experimentresults also show that network reduction can improve the availability ofexisting verification tools on many networks by reducing them into sequentialneural networks.

Expediting Neural Network Verification via Network Reduction