Unit 8: Complex Decision-Making

Commentary

Decision-making is one of the basic intelligent activities of human beings in everyday life and business, and thus has become an active branch of AI that is widely applied in many fields, including decision support systems (DSS) and control engineering. This unit focuses on decision-making principles and techniques, by which an agent can make rational decisions under uncertainty based on what it believes (i.e., beliefs) and what it prefers (i.e., utilities). In addition to introducing utility functions and decision networks, this unit addresses complex decision-making problems, such as sequential decision problems, Markov decision process (MDP), game theory, and multiagent decision-making.

Unit Purpose

When you complete this unit, you will be able to

• Define the concepts of decision theory, utility theory, and game theory.
• Describe preference structure and decision networks.
• Explain sequential decision problems and their algorithms, such as value iteration and policy iteration.
• Explain decision-making principles in multiagent systems based on game theory and mechanism design.

Section 1: Utility functions and decision networks
Section 2: Markov decision processes (MDPs)
Section 3: Multiagent decision-making and game theory

Readings

Supplemental Unit Readings

Books:

Bishop, C. M. (2006). Recognition and machine learning. Springer. ISBN 0-387-31073-8. (refer primarily to the sections covering Graphical Models)

Pearl, J. (1988). Probabilistic reasoning in intelligent systems: Networks of plausible inference. San Mateo, CA: Morgan Kaufmann.

Activities

• Explore other decision-making and MDP algorithms, and make simple comparisons between them. Report your findings in the course conference.
• Explore other game theory and mechanism design research and applications to see if it is possible to adapt and apply them to your own work.