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

Deep neural networks have been widely adopted in recent years, exhibitingimpressive performances in several application domains. It has however beenshown that they can be fooled by adversarial examples, i.e., images altered bya barely-perceivable adversarial noise, carefully crafted to misleadclassification. In this work, we aim to evaluate the extent to whichrobot-vision systems embodying deep-learning algorithms are vulnerable toadversarial examples, and propose a computationally efficient countermeasure tomitigate this threat, based on rejecting classification of anomalous inputs. Wethen provide a clearer understanding of the safety properties of deep networksthrough an intuitive empirical analysis, showing that the mapping learned bysuch networks essentially violates the smoothness assumption of learningalgorithms. We finally discuss the main limitations of this work, including thecreation of real-world adversarial examples, and sketch promising researchdirections.

Is Deep Learning Safe for Robot Vision? Adversarial Examples against the iCub Humanoid