Robust Training of Federated Models with Extremely Label Deficiency

Keyphrases

• Unlabeled Data 100%
• Labeled Data 100%
• Robust Training 100%
• Federated Model 100%
• Federated Semi-supervised Learning 100%
• Gradient Conflict 100%
• Semi-supervised Method 50%
• State-of-the-art Techniques 50%
• Benchmark Dataset 50%
• Substantial Evidence 50%
• Comprehensive Experiment 50%
• Machine Learning Models 50%
• Unsupervised Model 50%
• Feature Extracting 50%
• Neighborhood Relation 50%
• Single Model 50%
• Training Machine 50%
• Neighborhood Preserving 50%
• Twin Models 50%
• Mutual Guidance 50%
• Supervised Model 50%

Computer Science

• Objective Function 100%
• Unlabeled Data 66%
• Semisupervised Learning 66%
• Extracted Feature 33%
• Substantial Evidence 33%
• Machine Learning Model 33%