Advanced Learning Strategies for AI Systems

Inhalt

• Learning with limited information – methods for effective learning from scarce, weak, or uncertain supervision.
  
Topics include: self supervised, weakly supervised, semi supervised, reinforcement, multiple instance, Bayesian learning.

• Data efficient learning – strategies to minimize data and labeling needs through selective and structured training.
  
Topics include: active learning, curriculum learning, few or zero shot learning, machine teaching, data influence analysis.

• Non-stationary learning – approaches for adapting to evolving data and tasks.
  
Topics include: online learning, continual learning, federated learning, causal learning.

• Learning with human feedback – integration of human preferences and guidance into model learning.
  
Topics include: preference learning, human guided learning, recommender systems.

1. explain and distinguish the main paradigms of advanced learning strategies in artificial intelligence
   
2. assess the suitability of learning methods for tasks with limited data, dynamic conditions, or human feedback, 
   
3. critically analyze and discuss current research literature in AI learning, 
   
4. synthesize complex concepts and connect theoretical frameworks with practical challenges, 
   
5. present and defend findings clearly in written and oral form, and provide constructive feedback through peer review, 
   
6. reflect on the broader implications of modern AI systems regarding efficiency, robustness, interpretability, and ethical aspects.
   

Zusätzliche Informationen

Termine