all InfoSec news
How to Craft Backdoors with Unlabeled Data Alone?
April 11, 2024, 4:10 a.m. | Yifei Wang, Wenhan Ma, Yisen Wang
cs.CR updates on arXiv.org arxiv.org
Abstract: Relying only on unlabeled data, Self-supervised learning (SSL) can learn rich features in an economical and scalable way. As the drive-horse for building foundation models, SSL has received a lot of attention recently with wide applications, which also raises security concerns where backdoor attack is a major type of threat: if the released dataset is maliciously poisoned, backdoored SSL models can behave badly when triggers are injected to test samples. The goal of this work …
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
Assistant Manager, IT Security
@ CIMB | Cambodia
IT Security Engineer - GRC
@ Xtremax | Bandung City, West Java, Indonesia
Senior Engineer - Application Security
@ ANZ Banking Group Limited | Quezon City, PH
Penetration Tester Manager
@ RSM | USA-IL-Chicago-30 South Wacker Drive, Suite 3300
Offensive Security Engineer, Device Wireless Connectivity
@ Google | Amsterdam, Netherlands
IT Security Analyst I
@ Mitsubishi Heavy Industries | Houston, TX, US, 77046