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

A reliable deepfake detector or spoofing countermeasure (CM) should be robustin the face of unpredictable spoofing attacks. To encourage the learning ofmore generaliseable artefacts, rather than those specific only to knownattacks, CMs are usually exposed to a broad variety of different attacks duringtraining. Even so, the performance of deep-learning-based CM solutions areknown to vary, sometimes substantially, when they are retrained with differentinitialisations, hyper-parameters or training data partitions. We show in thispaper that the potency of spoofing attacks, also deep-learning-based, cansimilarly vary according to training conditions, sometimes resulting insubstantial degradations to detection performance. Nevertheless, while aRawNet2 CM model is vulnerable when only modest adjustments are made to theattack algorithm, those based upon graph attention networks and self-supervisedlearning are reassuringly robust. The focus upon training data generated withdifferent attack algorithms might not be sufficient on its own to ensuregeneraliability; some form of spoofing attack augmentation at the algorithmlevel can be complementary.

Spoofing attack augmentation: can differently-trained attack models improve generalisation?