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

Unlike traditional static deep neural networks (DNNs), dynamic neuralnetworks (NNs) adjust their structures or parameters to different inputs toguarantee accuracy and computational efficiency. Meanwhile, it has been anemerging research area in deep learning recently. Although traditional staticDNNs are vulnerable to the membership inference attack (MIA) , which aims toinfer whether a particular point was used to train the model, little is knownabout how such an attack performs on the dynamic NNs. In this paper, we proposea novel MI attack against dynamic NNs, leveraging the unique policy networksmechanism of dynamic NNs to increase the effectiveness of membership inference.We conducted extensive experiments using two dynamic NNs, i.e., GaterNet,BlockDrop, on four mainstream image classification tasks, i.e., CIFAR-10,CIFAR-100, STL-10, and GTSRB. The evaluation results demonstrate that thecontrol-flow information can significantly promote the MIA. Based onbackbone-finetuning and information-fusion, our method achieves better resultsthan baseline attack and traditional attack using intermediate information.

A Novel Membership Inference Attack against Dynamic Neural Networks by Utilizing Policy Networks Information