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/

Zendy Plus

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

Zendy Tools

Zendy Open

Adversarial attacks have become a major threat for machine learningapplications. There is a growing interest in studying these attacks in theaudio domain, e.g, speech and speaker recognition; and find defenses againstthem. In this work, we focus on using representation learning toclassify/detect attacks w.r.t. the attack algorithm, threat model orsignal-to-adversarial-noise ratio. We found that common attacks in theliterature can be classified with accuracies as high as 90%. Also,representations trained to classify attacks against speaker identification canbe used also to classify attacks against speaker verification and speechrecognition. We also tested an attack verification task, where we need todecide whether two speech utterances contain the same attack. We observed thatour models did not generalize well to attack algorithms not included in theattack representation model training. Motivated by this, we evaluated anunknown attack detection task. We were able to detect unknown attacks withequal error rates of about 19%, which is promising.

Representation Learning to Classify and Detect Adversarial Attacks against Speaker and Speech Recognition Systems