Back to Search Start Over

Trustworthy AI Using Confidential Federated Learning.

Authors :
Guo, Jinnan
Pietzuch, Peter
Paverd, Andrew
Vaswani, Kapil
Source :
Communications of the ACM. Sep2024, Vol. 67 Issue 9, p48-53. 6p.
Publication Year :
2024

Abstract

The artificial intelligence (AI) revolution is significantly altering industries and daily life, with innovations such as chatbots, personalized systems, and autonomous vehicles becoming increasingly prevalent. As organizations leverage AI to enhance efficiency and spur growth, ensuring the trustworthiness of AI systems is paramount. Trustworthy AI should demonstrate reliability, fairness, transparency, accountability, and robustness. Privacy is a crucial component, particularly in scenarios requiring sensitive data, such as medical or financial applications. Privacy-preserving techniques like federated learning (FL) and confidential computing are emerging to address these challenges. Combining these approaches into confidential federated learning (CFL) offers enhanced security and compliance with privacy regulations while maintaining transparency and accountability.

Details

Language :
English
ISSN :
00010782
Volume :
67
Issue :
9
Database :
Academic Search Index
Journal :
Communications of the ACM
Publication Type :
Periodical
Accession number :
179256396
Full Text :
https://doi.org/10.1145/3677390