Incorporating unlabeled data into distributionally-robust learning

Computer Science

• Real Data Sets 100%
• Active Learning 100%
• Unlabeled Data 100%
• Performance Guarantee 50%
• Learning Problem 50%
• Learning Performance 50%
• Risk Minimization 50%
• Learning Algorithm 50%
• Data Distribution 50%

Keyphrases

• Unlabeled Data 100%
• Distribution Learning 100%
• Adversary 42%
• Tight 14%
• Performance Guarantee 14%
• Active Learning 14%
• Learning Problems 14%
• Computable 14%
• Learning Performance 14%
• Stochastic Gradient 14%
• Empirical Risk Minimization 14%
• Future Performance 14%
• Model Change 14%
• Gradient-based Optimization 14%
• Data Distribution 14%
• Distributionally Robust 14%
• Active Learning Algorithm 14%

Mathematics

• Real Data 100%
• Data Distribution 50%
• Empirical Risk Minimization 50%
• Stochastics 50%
• Constrains 50%
• Robust Version 50%