Communication-Efficient Federated Learning via Sparse Training with Regularized Error Correction

Keyphrases

• Federated Learning 100%
• Error Correction 100%
• Communication-efficient Federated Learning 100%
• Sparse Training 100%
• Sparsity 60%
• Sparsification 40%
• Error Correction Mechanism 40%
• Staleness 40%
• Parameter Server 40%
• Learning Process 20%
• Error Feedback 20%
• Computational Cost 20%
• Downscaling 20%
• Accuracy Improvement 20%
• Communication Resources 20%
• Communication Cost 20%
• Privacy-preserving 20%
• Novel Algorithm 20%
• Model Complexity 20%
• Multiple Models 20%
• Limited Training Data 20%
• Updating Model 20%
• Computational Resources 20%
• Training Approaches 20%
• Emergent Paradigm 20%
• Sparsity Level 20%
• Federated Learning System 20%
• Training Scheme 20%
• DNN Model 20%
• Decentralized Machine Learning 20%

Computer Science

• Federated Learning 100%
• Error Correction 100%
• Sparsity 57%
• Parameter Server 28%
• Correction Method 28%
• Communication Cost 14%
• Deep Neural Network 14%
• Privacy Preserving 14%
• Learning Process 14%
• Complex Task 14%
• Computational Cost 14%
• Regularization 14%
• Learning System 14%
• Decentralized Machine Learning 14%