Towards Communication-Efficient Federated Learning via Sparse and Aligned Adaptive Optimization

Keyphrases

• Local Model 100%
• Sparsification 100%
• Adaptive Optimization 100%
• Communication-efficient Federated Learning 100%
• Moment Estimation 75%
• Adaptive Moment Estimation 75%
• Convergence Rate 50%
• Parameter Estimation 50%
• Communication Overhead 50%
• Centralized Server 50%
• Distributed Devices 50%
• Learning Rate 25%
• Performance Degradation 25%
• Fast Convergence 25%
• Learning Performance 25%
• Sparse Representation 25%
• Second Moment 25%
• First Moment 25%
• Nonconvex Objective Function 25%
• Model Updating 25%
• Stochastic Gradient Descent 25%
• Upload 25%
• Error Data 25%
• Deviation Bounds 25%
• Federated Learning 25%
• Test Accuracy 25%
• Adam Algorithm 25%
• Adaptive Moment Estimation Algorithm 25%
• Convergence Bounds 25%
• Stochastic Gradient Descent Algorithm 25%
• Imbalanced Distribution 25%
• Uplink Communication 25%

Computer Science

• Federated machine learning 100%
• Sparsification 100%
• Communication Overhead 50%
• Convergence Rate 50%
• Stochastic Gradient Descent 50%
• Experimental Result 25%
• Performance Degradation 25%
• Objective Function 25%
• Fast Convergence 25%
• Imbalanced Data 25%
• Sparse Representation 25%
• Estimation Algorithm 25%
• Data Distribution 25%
• Learning Rate 25%
• Learning Performance 25%