Aims: To provide students with an understanding of some of the principles of Artificial Intelligence.
Learning Outcomes:
After completing this module, students should be able to: construct a simple rule based expert system; explain the major components of a fuzzy logic system and conduct fuzzy inference; describe the major type of neural networks and their learning algorithms; construct multilayer neural networks for pattern classification; apply a simple genetic algorithm to solve optimisation problems; construct and solve game trees for single player and multi-player games.
Skills:
Application of the techniques introduced in the lectures to AI problems: taught, facilitated and tested.
Content:
Expert Systems: Overview, rules , inference, knowledge aquisition, forward and backward chaining. Fuzzy Logic: Comparison with crisp logic. Linguistic variables, Degree of Membership, fuzzy rules, defuzzification. Neural Networks: MCP neuron, geometric interpretation. XOR problem. 1, 2, and 3 layer feed-forward networks. The Hebb rule, sigmoid function, backpropagation. Genetic Algorithms: Overview, the Schema Theorem, representation, populations, selection, crossover mutation. GameTheory: One and two player perfect information games, AND/OR game trees, depth first and breadth first searches, min-max search, alpha-beta pruning, proof number searching.
|
