Artificial Intelligence

1. Administrative
2. Schedule and Readings
3. Lecture Details
4. Assignments

• The course lecturer is Mike Ashcroft:
  • Email: mikeashcroft@inatas.com
  • Phone: 0769 415 418
  • NB Use the given email address for contact. Do not use Mike's Uppsala University address.
• Administration questions should be directed to Roland Bol:
  • Email: Roland.Bol@it.uu.se
  • Phone: 018 471 7606.
  • Room: 1356
• Guest lectures will be given by Olle Gällmo and Ivan Jordanov.
• The course will be given in English. Assignments must be handed in in English.

• Readings refer to chapters (or, in the case of '4.1' chapter and section) in Russell and Norvig, Artificial intelligence : A Modern Approach, 3rd Ed.
• Reading outside brackets are compulsory. Readings in brackets are suggested, and tend to be entire chapters whereas the compulsory readings are selected sections. If you only read the compulsory sections, you may need to look up particular concepts outside these sections.
• The topic schedule is an estimation only. We will proceed as the topics require.
• The assignments are given for the lecture at which work on them will begin.
• The schedule may change depending on lecturer availability.

Week	Day	Time	Room	Topic	Who	Reading	Assignments
36	Tor 6	08:15	Pol_1311	Introduction	Mike	(1,2)
36	Fre 7	15:15	Pol_1311	Search 1	Mike	3,4.1 Lecture notes
37	Tis 11	13:15	Pol_2347	Natural Computation 1	Olle	Slides
37	Tor 13	10:15	Pol_1311	Natural Computation 2	Ivan	Neural network slides Genetic algorithms slides (18.7)
37	Fre 14	15:15	Pol_1311	Search 2	Mike	4.3 Lecture notes	1. Planning
38	Mån 17	15:15	Pol_1211	Propositional and Predicate Logic	Mike	7.3-7.5,8.1-8.2,9.1-9.5 (7,8,9) Lecture slides
38	Ons 19	08:15	Pol_1311	Planning	Mike	10.1-10.3,10.4.4,11.1 (10,11) Lecture slides
39	Mån 24	10:15	Pol_1111	Probability Theory	Mike	13.2-13.5 (13) Probability Theory and Bayesian Network Notes
39	Ons 26	13:15	Pol_2146	Probabilistic Networks	Mike	14.1-14.5,15.3-15.5 (14,15) Probability Theory and Bayesian Network Notes Related Article
39	Fre 28	15:15	Pol_1311	Decision Theoretic Networks	Mike	16,17
40	Tis 2	10:15	Pol_1211	Learning	Mike	20 Probability Theory and Bayesian Network Notes	2. Bayesian Networks
40	Fre 5	08:15	Pol_1311	AI in Computer Games	Olle	Lecture Slides
41	Mån 8	13:15	Pol_1311	Natural Language Processing	Mike	22,23 Lecture Slides
41	Ons 10	10:15	Pol_1311	Spare	Mike	-
41	Fre 12	15:15	Pol_1311	Conclusion/Recap	Mike	Lecture Slides/Exam Guide
43	Mån 22	13:15	Pol_1311	Question Session	Mike	-
43	Ons 24	08:00		Exam
3	Tis 15	08:00		Re-Exam

Lecture 1: Introduction

• Course Administration
• Course Overview
• History of AI

Lecture 2: Search 1

• State Space
• Search Trees
• Tree and Graph Search
• Basic Search Strategies: Breadth-First, Uniform-Cost, Depth-First, Iterative Deepening, Bi-directional
• Heuristic Search: Greedy Best-First, A*

Lecture 3: Natural Computation 1

• TBA

Lecture 4: Natural Computation 2

• Intro to NN - the Perceptron.
• Linear Separability problem.
• Perceptron learning as a minimization problem.
• MLP, NN learning (supervised and unsupervised) and
• generalization, Backpropagation.
• Unsupervised learning - Kohonen's SOMs.
• NN applications.
• GA - intro and terminology.
• A simple GA.
• GA reproductive operators.
• Natural selection and fitness proportional selection - example.
• Other selection techniques.
• Advantages/disadvantages of GA.
• GA applications.

Lecture 5: Search 2 (and Markov Models)

• Hill Climbing
• Simulated Annealing
• Markov Chains/Markov Matrices (in the context of Simulated Annealing)
• Dynamic Programming

Lecture 6: Propositional and First Order Logic

• Semantics (P: Truth Tables and Algebraic. FO: Set-Theoretic and Database)
• Model Searching (P)
• Inference Rules and Proofs
• Substitution and Unification (FO)
• Resolution
• Definite Clauses
• Forward and Backward Chaining
• Possible Worlds and Accessibility Relations (FO)
• Higher Order Logic
• 'Deep' Objections to AI

Lecture 7: Planning

• Actions - Preconditions, Results
• Situational Calculus: Actions as Accessibility Relations
• Planning Domain Definition Language (PDDL)
• Planning Graphs
• Planning as Partial Ordering
• Scheduling
• Resource Constraints

Lecture 8: Probability Theory

• Probability Spaces and Basic Probability Theory
• Random Variables and Probability Distributions
• Bayes Law
• Independence and Conditional Independence
• Conditional Probability Distributions
• The Chain Rule
• Expected Utility

Lecture 9: Markov Models and Bayesian Networks

• Markov Chains
• Bayesian Networks
• Markov Models
• Hierarchical Markov Models
• Dynamic Bayesian Networks

Lecture 10: Markov Decision Processes and Influence Diagrams

• Influence Diagrams
• Markov Decision Processes
• State Models (Dynamic Influence Diagrams)
• Partially Observable Markov Decision Processes
• Kalman Filters
• Partially Observable Dynamic Influence Diagrams
• Pattern Matching

Lecture 11: Learning: Soft-sensing and system-control

• Learning Bayesian Networks/Influence Diagrams - Structural and Parametric
• Hidden Variables
• Pattern Learning
• Soft-Sensing Example
• System Control Example

Lecture 12: AI in Computer Games

• TBA

Lecture 13: Natural Language Processing

• TBA

Lecture 14: Spare

• If the spare is not required (which is unlikely), we will examine natural language processing in further detail.

• There are 2 obligatory assignments, as listed in the schedule. Detailed instructions for each assignment will follow.

Assignment 1: Planning

• Assignment to be undertaken in groups of (up to) four.
• Groups should be formed by Wednesday, September 19.
• Due Date: NEW DUE DATE OCTOBER 8
• Presentation Dates: TBA
• Details given here

Assignment 2: Bayesian Networks

• Due Date: Friday, October 19
• Details here