INF2D: Week 10: Rational Decision Making
Welcome to the final week of lectures for Inf2D!
Having studied how to reason about your beliefs in the face of uncertainty, both in a static environment (BNs) and in a dynamic one (DBNs or the special case, HMMs), we are finally in a position to merge those probabilistic inferences with a numeric representation of preferences to show how an agent can make rational decisions: recall the principle of Maximising Expected Utility, which captures an optimal trade off between what the agent prefers and what it thinks it can achieve.
First, however, we have one last section on Dynamic Bayesian Networks: DBNs. The purpose of Section 28 is to give you an idea of when it might be useful to revise the design of your graphical model:
28: Dynamic Bayesian Networks
Then, in sections 29 and 30 we merge reasoning about belief under uncertainty with preferences to compute rational action: Section 29 deals with static environments (and so the belief component of the model is a BN), while Section 30 deals with dynamic environments (and so the belief component of the model is a DBN). Section 30 also contains a final lecture in which we explore the ethical issues and challenges connected with AI, and in particular with building AI agents.
29: Decision Making under Uncertainty
30: Markov Decision Processes and Ethics
As always, these lectures will be delivered in person and live streamed. But in addition, each of the above includes pre-recorded videos of the lectures (with edited captions), the slides that were used in the videos, required readings, and a post-lecture quiz (for 28 and 29 only). The quiz is a chance for you to gauge your understanding of the material presented here, and so we strongly encourage you to review this content in the above order, and then complete the quiz. If there is anything you don't understand, then you have several options:
- Post a question on piazza;
- Ask a question at the in-persion lecture; and/or
- Ask your tutor.
Good luck with your assignments and your exam, and I very much hope you will continue to pursue further AI courses on your degree programme next year.