all InfoSec news
A law of adversarial risk, interpolation, and label noise. (arXiv:2207.03933v1 [stat.ML])
July 11, 2022, 1:20 a.m. | Daniel Paleka, Amartya Sanyal
cs.CR updates on arXiv.org arxiv.org
In supervised learning, it has been shown that label noise in the data can be
interpolated without penalties on test accuracy under many circumstances. We
show that interpolating label noise induces adversarial vulnerability, and
prove the first theorem showing the dependence of label noise and adversarial
risk in terms of the data distribution. Our results are almost sharp without
accounting for the inductive bias of the learning algorithm. We also show that
inductive bias makes the effect of label noise …
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
SOC 2 Manager, Audit and Certification
@ Deloitte | US and CA Multiple Locations
Cyber Security Cloud Solution Architect
@ Microsoft | London, London, United Kingdom
Compliance Program Analyst
@ SailPoint | United States
Software Engineer III, Infrastructure, Google Cloud Security and Privacy
@ Google | Sunnyvale, CA, USA
Cryptography Expert
@ Raiffeisen Bank Ukraine | Kyiv, Kyiv city, Ukraine
Senior Cyber Intelligence Planner (15.09)
@ OCT Consulting, LLC | Washington, District of Columbia, United States