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

Most previous works usually explained adversarial examples from severalspecific perspectives, lacking relatively integral comprehension about thisproblem. In this paper, we present a systematic study on adversarial examplesfrom three aspects: the amount of training data, task-dependent andmodel-specific factors. Particularly, we show that adversarial generalization(i.e. test accuracy on adversarial examples) for standard training requiresmore data than standard generalization (i.e. test accuracy on clean examples);and uncover the global relationship between generalization and robustness withrespect to the data size especially when data is augmented by generativemodels. This reveals the trade-off correlation between standard generalizationand robustness in limited training data regime and their consistency when datasize is large enough. Furthermore, we explore how different task-dependent andmodel-specific factors influence the vulnerability of deep neural networks byextensive empirical analysis. Relevant recommendations on defense againstadversarial attacks are provided as well. Our results outline a potential pathtowards the luminous and systematic understanding of adversarial examples.

Towards Understanding Adversarial Examples Systematically: Exploring Data Size, Task and Model Factors