CSC443 Reasoning under uncertainty

Course outline:
This subject gives students an overview of work done in the Artificial Intelligence community in the area of reasoning under uncertainty.

The first part of the course will focus on Bayesian (or Belief) networks. Bayesian networks have rapidly become one of the leading technologies for applying AI to real world problems. This follows the work of Pearl, Lauritzen, and others in the late 1980s showing that Bayesian reasoning in practice could be tractable (although in principle it is NP-hard). This in turn has led to the rapid growth in applying machine learning methods to the automated building of Bayesian nets. In the first part of this class we survey both aspects of Bayesian AI: the use of Bayesian nets for modeling reasoning under uncertainty and the use of statistical methods, including Minimum Message Length inference (MML), for learning Bayesian nets from observational data. We begin with a brief examination of the philosophy of Bayesianism, motivating the use of probabilities in decision making, agent modeling and data analysis, and contrasting Bayesian methods with certainty factors, fuzzy logic and the Dempster-Shafer calculus. We introduce Bayesian networks, their inference techniques and approximation methods. We illustrate their use in various applications (robotics and planning, medical decision making, plan recognition, natural language generation and game playing). We then look at some of the main machine learning techniques available for learning Bayesian net structures.

The second part of the course will look at another representation of uncertainty, Markov Decision Processes, which are in turn the basis for reinforcement learning. Reinforcement learning is a computational approach to learning whereby an agent tries to maximise the total amount of reward it receives when interacting with a complex, uncertain environment. We will look at basic solutions methods, including dynamic programming, Monte Carlo methods and temporal-difference learning. Then we will look at extensions to these methods, including approximate methods and Partially-Observable MDPs.

Lectures:
Tuesday 11.00 - 1.00pm (26-135)

Assessment:
Assignment 1 (5%) due 5 pm Thursday 23 March (week 4) - electronic submission
Assignment 2 (15%) due 5 pm Thursday 30 March (week 5) - electronic submission
Assignment 3 (15%) due 5 pm Thursday 20 April (week 8) - electronic submission
Assignment 4 (15%) due 5 pm Thursday 18 May (week 11) - electronic submission
Examination (50%) 11.00 am Tuesday 6 June (S13)

Courseware

Back to 2000 Honours subjects