Understanding Clipping for Federated Learning: Convergence and Client-Level Differential Privacy

Mathematics

• Clipping 100%
• Privacy 82%
• Learning 58%
• Update 47%
• Learning Algorithm 38%
• Privacy Protection 26%
• Architecture 17%
• Training 17%
• Neural Networks 15%
• Convergence Analysis 14%
• Knowledge 12%
• Gradient 12%
• Line 10%
• Observation 10%
• Relationships 9%
• Demonstrate 9%
• Model 5%

Engineering & Materials Science

• Learning algorithms 53%
• Data privacy 33%
• Neural networks 19%