Privacy for Free: Wireless Federated Learning via Uncoded Transmission with Adaptive Power Control

Keyphrases

• Non-orthogonal multiple Access 100%
• Uncoded Transmission 100%
• Adaptive Power Control 100%
• Wireless Federated Learning 100%
• Federated Learning 75%
• Privacy Constraint 50%
• Potential Gains 25%
• Wireless Systems 25%
• Distributed Protocol 25%
• Online Methods 25%
• Signal-to-noise Ratio 25%
• Privacy Guarantees 25%
• Work Study 25%
• Data Exchange 25%
• Learning Task 25%
• Power Constraint 25%
• Adaptive Resource Allocation 25%
• Constraint Level 25%
• Differential Privacy 25%
• Offline Optimization 25%
• Learning Performance 25%
• Dynamic Power Allocation 25%
• Channel Noise 25%
• Over the Air 25%
• Masking Mechanism 25%
• Optimality Gap 25%
• Common Learning 25%
• Local Data 25%
• Formal Privacy 25%
• Induced Mechanism 25%
• Distributed Gradient Descent 25%
• Optimization Setting 25%

Computer Science

• Uncoded Transmission 100%
• Federated Learning 100%
• Multiple Access 40%
• Power Allocation 40%
• Non-Orthogonal Multiple Access 40%
• Wireless System 20%
• Potential Benefit 20%
• Noise-to-Signal Ratio 20%
• Learning Performance 20%
• Power Constraint 20%
• Differential Privacy 20%
• Distributed Protocol 20%
• Gradient Descent 20%
• Dynamic Power 20%