Introduction to Artificial Inteligence - Spring 2001

Learning

1. General models for learning
   • Structure of a learning agent
   • Performance element components
   • Types of learning: supervised/unsupervised, reinforcement
   • Inductive learning
2. Learning decision trees
   • Decision trees as classifiers
   • Expressiveness and efficiency of decision trees
   • Learning decision trees
   • Performance evaluation of decision tree learning algorithms
   • Applications
   • Information theory in learning
   • Information and information gain
   • Noise and overfitting
   • Extensions of the decision-tree framework: missing data, multivalued and continuous attributes, predictive trees
3. Artificial neural networks
   • Natural neural networks - the brain
   • Artificial neural networks
     • Unit (computing element) types
     • Network structures
     • Network structure optimization
   • Perceptrons
     • Definition and representation strength
     • Learning for perceptrons
   • General feed-forward networks
     • Training ANN using genetic algorithms
     • Back-propagation algorithm
   • Evaluation of the neural networks paradigm
   • Applications: pronounciation, OCR, ALVINN