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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Due to its powerful capability of representation learning and high-efficiencycomputation, deep hashing has made significant progress in large-scale imageretrieval. However, deep hashing networks are vulnerable to adversarialexamples, which is a practical secure problem but seldom studied inhashing-based retrieval field. In this paper, we propose a novelprototype-supervised adversarial network (ProS-GAN), which formulates aflexible generative architecture for efficient and effective targeted hashingattack. To the best of our knowledge, this is the first generation-based methodto attack deep hashing networks. Generally, our proposed framework consists ofthree parts, i.e., a PrototypeNet, a generator, and a discriminator.Specifically, the designed PrototypeNet embeds the target label into thesemantic representation and learns the prototype code as the category-levelrepresentative of the target label. Moreover, the semantic representation andthe original image are jointly fed into the generator for a flexible targetedattack. Particularly, the prototype code is adopted to supervise the generatorto construct the targeted adversarial example by minimizing the Hammingdistance between the hash code of the adversarial example and the prototypecode. Furthermore, the generator is against the discriminator to simultaneouslyencourage the adversarial examples visually realistic and the semanticrepresentation informative. Extensive experiments verify that the proposedframework can efficiently produce adversarial examples with better targetedattack performance and transferability over state-of-the-art targeted attackmethods of deep hashing. The related codes could be available athttps://github.com/xunguangwang/ProS-GAN .

Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing