Document Type
Presentation
Location
COAS: Math Conference Room
Start Date
27-9-2023 12:00 PM
End Date
27-9-2023 1:00 PM
Description
AI's applicability across diverse fields is hindered by data sensitivity, privacy concerns, and limited training data availability. Federated Learning (FL) addresses this challenge by enabling collaborative machine learning while preserving data privacy. FL allows clients to engage in model training with their local data, avoiding centralized storage. However, even with FL, security threats persist, jeopardizing model integrity and client data privacy. In this presentation, we will explore our latest findings in this area of research, safeguarding sensitive data from attacks through techniques like secure multiparty computation, homomorphic encryption, and differential privacy within the FL framework, enhancing data protection, and expanding AI's reach across industries.
Privacy-preserving Federated Learning
COAS: Math Conference Room
AI's applicability across diverse fields is hindered by data sensitivity, privacy concerns, and limited training data availability. Federated Learning (FL) addresses this challenge by enabling collaborative machine learning while preserving data privacy. FL allows clients to engage in model training with their local data, avoiding centralized storage. However, even with FL, security threats persist, jeopardizing model integrity and client data privacy. In this presentation, we will explore our latest findings in this area of research, safeguarding sensitive data from attacks through techniques like secure multiparty computation, homomorphic encryption, and differential privacy within the FL framework, enhancing data protection, and expanding AI's reach across industries.