all InfoSec news
Safeguarding Data in Multimodal AI: A Differentially Private Approach to CLIP Training
March 4, 2024, 5:10 a.m. | Alyssa Huang, Peihan Liu, Ryumei Nakada, Linjun Zhang, Wanrong Zhang
cs.CR updates on arXiv.org arxiv.org
Abstract: The surge in multimodal AI's success has sparked concerns over data privacy in vision-and-language tasks. While CLIP has revolutionized multimodal learning through joint training on images and text, its potential to unintentionally disclose sensitive information necessitates the integration of privacy-preserving mechanisms. We introduce a differentially private adaptation of the Contrastive Language-Image Pretraining (CLIP) model that effectively addresses privacy concerns while retaining accuracy. Our proposed method, Dp-CLIP, is rigorously evaluated on benchmark datasets encompassing diverse vision-and-language …
arxiv cs.cr cs.it cs.lg data data privacy images information integration language math.it privacy private safeguarding data sensitive sensitive information stat.ml text training
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
Cyber Security Analyst
@ Dane Street | Palm Beach Gardens, Florida, United States
Program Information System Security Manager (ISSM) - onsite Tucson, AZ - TOP SECRET required
@ RTX | AZ855: RMS AP Bldg M05 1151 East Hermans Road Building M05, Tucson, AZ, 85756 USA
Lead - Business System Service (Workday HR Functional Consultant)
@ Freshworks | Bengaluru, India
Cloud Security Engineer
@ ButterflyMX | United States - Remote
Compliance Specialist
@ Airtable | Austin, Texas or San Francisco, California
Cyber SCRM Cloud Assessor Lead
@ Accenture Federal Services | Arlington, VA