Review on Reasoning about Uncertainty
Artificial Intelligence
COSC1127/1125
Semester 2, 2021
Prof. Sebastian Sardina
Some news…
Preliminary contest ranking here.
Base marks using Python script provided in #271
Marks may be adjusted by contributions.
Bonus Project 3 to be marked this week
Final Pacman Contest:
Agent system due Week 12
Instructions on Wiki and video coming soon
CES Survey closes soon?
Are you a better CS after this course?
Have you learnt new things as a CS?
THE Review: 2nd lectorial today!
*
Congratulations
Mint to be 3!
Remaining assessments…
Instructions and Wiki template to be provided soon…
Use feedback contests….
Remember Pacman Dashboard!
Better access and download to your logs and replays…
Better individual stats
THE Review today 4:30pm
Join me at 4:30pm!
Next week ….
@
Questions?
*
‹#›
What have we seen?
Search: as a general problem solving technique.
Knowledge Representation: rational behavior requires knowledge! Beyond databases…
Automated Planning: what plan should I execute?
mixing search + KR
Probabilities: basic tool for reasoning under uncertainty.
Bayesian Networks: knowledge representation for probabilistic reasoning.
MDP: decision making under uncertainty
Reinforcement Learning: learn environment and how to act rationally.
*
Whiteboard used for reviewing…
Semantics of Bayes Nets
If we ask for P(x1, x2,…, xn) we obtain
assuming an ordering consistent with network.
By the chain rule, we have:
P(x1, x2,…, xn) =
= P(xn | xn-1, … , x1) P(xn-1 | xn-2, …, x1) … P(x1)
= P(xn | Par(xn)) P(xn-1 | Par(xn-1)) … P(x1)
Thus, the joint is recoverable using the parameters (CPTs) specified in an arbitrary BN