Information-theoretic generalization bounds for meta-learning and applications

Keyphrases

• Information Theory 100%
• Meta-learning 100%
• Generalization Bounds 100%
• Mutual Information 50%
• Generalization Gap 50%
• Learning Algorithm 33%
• Training Data 33%
• Meta-training 33%
• Task Training 33%
• Key Performance Indicators 16%
• Reptiles 16%
• Individual Tasks 16%
• Average Loss 16%
• Information Bound 16%
• Tight Bounds 16%
• Learning Procedure 16%
• Sampling Efficiency 16%
• Inductive Bias 16%
• Model-agnostic Meta-learning 16%
• Task Learning 16%
• Learning to Learn 16%
• Conventional Learning 16%
• Noisy Iterative Algorithms 16%
• Task Uncertainty 16%

Computer Science

• Mutual Information 100%
• Learning Algorithm 66%
• Training Data 66%
• Performance Measure 33%
• Individual Task 33%
• Iterative Algorithm 33%
• Task Uncertainty 33%
• Conventional Learning 33%