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

In this work, we study the robustness of a CNN+RNN based image captioningsystem being subjected to adversarial noises. We propose to fool an imagecaptioning system to generate some targeted partial captions for an imagepolluted by adversarial noises, even the targeted captions are totallyirrelevant to the image content. A partial caption indicates that the words atsome locations in this caption are observed, while words at other locations arenot restricted.It is the first work to study exact adversarial attacks oftargeted partial captions. Due to the sequential dependencies among words in acaption, we formulate the generation of adversarial noises for targeted partialcaptions as a structured output learning problem with latent variables. Boththe generalized expectation maximization algorithm and structural SVMs withlatent variables are then adopted to optimize the problem. The proposed methodsgenerate very successful at-tacks to three popular CNN+RNN based imagecaptioning models. Furthermore, the proposed attack methods are used tounderstand the inner mechanism of image captioning systems, providing theguidance to further improve automatic image captioning systems towards humancaptioning.

Exact Adversarial Attack to Image Captioning via Structured Output Learning with Latent Variables