CPSC 522 - Artificial Intelligence 2
AI for Decision-Making and Acting

Overview

This is an advanced AI course on what should an agent do based on its ability, its beliefs, its perception and its values/goals. We will assume basic logic, probability, search, and supervised machine learning such as covered in an introductory AI or machine learning course.

We will cover three main topics: probabilistic graphical models (models of belief), decision making (utility, rewards, planning, reinforcement learning), and relational models (entities and relations, knowledge graphs, making predictions based on relational data, existence and identity uncertainty).

This course builds on the foundations of dynamic systems, decision theory, knowledge representation and machine learning all of which will be presented in enough detail to understand the rest of the course.

Structure of the Course

Instructor: David Poole, poole@cs.ubc.ca. Office hours: After class any day or by appointment (room ICCS 109).

There will be 3 hours of in-class interaction per week. The classes will be a mix of lectures on the foundations and student presentations and discussion of research papers. This is a participatory class; everyone will be expected to participate fully, to have read the reading material before class, and come ready to discuss and critically analyze it.

Topics

• Agents:
  observations, priors -> beliefs
  beliefs, preferences, abilities -> actions
• Beliefs
• Probabilistic models (directed and undirected graphical models): semantics, inference, learning
• Causality, actions
• Models of time
• Relational models: representations, inference, learning. Knowledge graphs. Embedding-based models, graph neural networks, models of interdependencies of relations.
• Identity, number and existence uncertainty
• Action
• Preferences and utilities. Multi-attribute utility theory.
• Multi-stage decision making; decision networks, Markov decision processes.
• Value of information, control and computation (Bounded Rationality)
• Learning how to act; reinforcement learning.
• Relational reinforcement learning
• Partial observability, sensing the environment. (POMDPs)
• Multi-agent systems; cooperative and competitive decision making.

Assessment

The course assessment will be based on in-class presentations/participation, peer reviewing and three assignments/projects. The current plan is for participants to contribute to the online textbook on UBC Wiki. Each wiki page will have (at least) one principal author. Each assignment of for you to:

• Be principal author of (at least) one page
• Be critique of some pages, which involves giving constructive feedback for the page.
• Be an evaluator for all the pages

Each wiki page should include both the prior argument and the evidence that it works.

Last updated: 2023-08-14, David Poole