FedCSpc: A Cross-Silo Federated Learning System with Error-Bounded Lossy Parameter Compression

Keyphrases

• Lossy 100%
• Cross-silo Federated Learning 100%
• Parameter Compression 100%
• Federated Learning Framework 100%
• Error-bounded 100%
• Model Accuracy 66%
• FedAvg 66%
• Large-scale Model 66%
• Gradient Compression 66%
• Large Volume 33%
• Error Bound 33%
• Networked Publics 33%
• Compression Techniques 33%
• Volume Model 33%
• Limited Bandwidth 33%
• CPU Resource 33%
• Adjustment Algorithm 33%
• Parameter Adaptation 33%
• Communication Performance 33%
• Data Privacy 33%
• Comprehensive Evaluation 33%
• Gbps 33%
• Compression Error 33%
• Data Silos 33%
• Compression Strategy 33%
• Affect Communication 33%
• Wall Clock Time 33%
• Bound Adjustment 33%
• Scaling Deep 33%
• Adaptive Compression 33%
• Model Training 33%
• Deep Neural Network Training 33%
• GPU 33%
• GPU Cluster 33%

Computer Science

• Learning System 100%
• Federated machine learning 100%
• Model Accuracy 66%
• Graphics Processing Unit 66%
• Deep Neural Network 33%
• Public Networks 33%
• Wide Area Network 33%
• Adaptive Parameter 33%
• Explainable Artificial Intelligence 33%
• Data Privacy 33%
• Neural Network Training 33%
• Compression Technique 33%