Download CS 360: Advanced Artificial Intelligence Fall 2014

CS 360: Advanced Artificial Intelligence
Fall 2014
Instructor: Gautam Biswas
Time: Mon-Wed 1:10-2:25 pm
Office: 366 Jacobs Hall (1025 16th Ave S, 401D)
(FGH)
Location: 129 Featheringill Hall
e-mail: [email protected]
Tel: (615)-343-6204
Course Objectives: Artificial Intelligence (AI) is viewed in different ways, which makes it hard
to define in a precise way. However, a majority of computer scientists, engineers, and cognitive
psychologists view AI as a discipline that enumerates and explores tasks that are hard and computationally expensive to solve (sometimes impossible) using conventional algorithms. In other
words, AI deals with complex problems (e.g., playing chess, planning trips, and natural language
understanding) that are best solved by emulating human intelligence algorithmically, and/or devising computational heuristic methods to solve these problems.
We will adopt this viewpoint in this course and study AI both from the computational (how can
we design effective and efficient algorithms to solve problems that typically have exponential
complexity) and cognitive (how do humans solve complex problems effectively and efficiently)
viewpoints. The course will cover fundamental concepts, ideas, techniques, and applications and
provide practical experience by implementing intelligent reasoning techniques.
Text:
Artificial Intelligence A Modern Approach, by Stuart Russell and Peter Norvig, Third edition,
Prentice Hall, NJ, 2009.
References:
G.F. Luger, Artificial Intelligence: Structures and Strategies for Complex Problem Solving, Addison Weseley, 2005.
N.J. Nilsson, Artificial Intelligence: A New Synthesis, Morgan Kaufmann, 1998.
Pre-requisites: CS 260 (Introduction to AI), or equivalent, or permission of instructor.
Syllabus:
No
Topic Name
1
2
Introduction to AI (self study & discussion)
Intelligent Agents (self-study & discussion)
3
Search and Problem Solving
A*, Iterative Deepening A*, Anytime Heuristic Search
Simulated Annealing, Genetic Search Algorithms
Russell & Norvig
Chapter
1 (08/20)
2 (08/20)
3, 4 (08/25, 08/27)
Additional Papers:
1. R.E. Korf, “Depth-First Iterative Deepening: An Optimal Admissible Tree Search,” Artificial Intelligence, vol. 27, pp. 97-109, 1985.
2. E.A. Hansen and R. Zhou, “Anytime Heuristic Search,” Journal of
Artificial Intelligence Research, vol. 28, pp. 267-297, 2007.
4
Adversarial Search (Two person games)
Discussion of algorithms (Alpha-Beta) + applications
5 (09/1)
5
Discussion of Programming Assignment 1
6(09/03)
6
Machine Learning
Supervised Learning (Decision Trees, Linear Regression)
18 (09/08, 09/10)
7
8
9
Constraint Satisfaction Problems (CSP)
Additional Paper:
1. R. Bartak, "Constraint Propagation and Backtracking Search," Lecture
Notes, 2005.
Informed (Heuristic Search)
Discussion of Programming Assignment 2
Knowledge Representation and Reasoning (short)
(Logical Representations)
Propositional Logic (self study)
First Order Logic and Inference (self-study)
Knowledge Representation & Ontologies
Additional Papers:
3. J.A. Hendler, “Agents and the semantic web,” IEEE Intelligent Systems, vol. 16(2), pp. 30-37, March/April 2001.
6(09/15, 09/17)
09/22
7, 8, 9 (Self-Study)
12(09/24, 09/29)
4. D. McGuinness, “Ontologies come of Age,” The Semantic Web:
Why, What, and How, MIT Press, 2001.
10
Discussion of state of the art + applications
Planning & Scheduling (short)
Planning as State Space Search (self-study)
Forward Search & Backward Search (Regression) (self-study)
Planning (GRAPHPLAN) & Partial Order Planning
Additional Paper:
5. D. Bryce and S. Kambhampati, “A Tutorial on Planning GraphBased Reachability Heuristics,” AAAI Magazine, pp. 47-83, Spring
2007. (we will cover some sections of this paper)
10,11 (10/01, 10/06,
10/08)
Discussion of state of the art + applications
11
12
Mid-Term Exam
Discussion of Programming Assignment 3
13
Reasoning under Uncertainty
Bayes Rule and Bayes Nets
Take Home
14, 15, 16
Hidden Markov Models, Dynamic Bayes Nets
Making Simple Decisions (Utility Theory Framework)
Discussion of state of the art + applications
Machine Learning
Learning Probabilistic Models
Unsupervised Learning (Clustering: K-Means, EM algorithm, Conceptual
Clustering)
14
15
16
19, 20, 21 (?)
Additional Topics
Discussion of state of the art + applications
Grading Scheme: The grades for the class will be based on 4 homework + 3 programming assignments, and 2 exams (Midterm and Final: Take home exams), and a paper report (on a topic
of the student’s choice).
Item
Number
Percentage of Grade
Home Work
4
30
Prog Assns
3
30
Exams
2
30
Report
1
10
Notes:
1. There will a number of reading assignments for the course. I will expect you to read the
material and papers before class, so we can use the classroom more for discussion, problem solving, and understanding how the techniques may be applied to solving problems.
2. All the lecture slides, lecture notes, additional reading material, assignments, exams, solutions to exams and assignments, etc. will be posted on Oak. Please make sure you keep
up with all of the postings on Oak.
3. Programming assignments can be solved in a language of your choice (C++, Java, Python, Lisp).
4. All home work, program assignments, and exam work (exams will be take home) turned
in has to be your own. Check the details of the honor code on Oak.