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

Although neural networks are widely used, it remains challenging to formallyverify the safety and robustness of neural networks in real-world applications.Existing methods are designed to verify the network before deployment, whichare limited to relatively simple specifications and fixed networks. Thesemethods are not ready to be applied to real-world problems with complex and/ordynamically changing specifications and networks. To effectively handle suchproblems, verification needs to be performed online when these changes takeplace. However, it is still challenging to run existing verification algorithmsonline. Our key insight is that we can leverage the temporal dependencies ofthese changes to accelerate the verification process. This paper establishes anovel framework for scalable online verification to solve real-worldverification problems with dynamically changing specifications and/or networks.We propose three types of acceleration algorithms: Branch Management to reducerepetitive computation, Perturbation Tolerance to tolerate changes, andIncremental Computation to reuse previous results. Experiment results show thatour algorithms achieve up to $100\times$ acceleration, and thus show apromising way to extend neural network verification to real-world applications.

Online Verification of Deep Neural Networks under Domain Shift or Network Updates