ECS170 Introduction to A.I. Course Overview Lecture What is A.I.? Who can tell me some well known applications? Watson (see http://www.aaai.org/Magazine/Watson/watson.php) Go around the room and introduce yourself, say why you chose this class, what it means for an algorithm to be intelligent?

Course Overview Course Logistics 1

Well Known Applications

• Checkers (world champion is a machine) – http://webdocs.cs.ualberta.ca/~chinook/

• Chess (beat world champion a few times) • Watson (question and answering) • Internet analysis (page rank, recommender systems, analysis of social networks) • DARPA Grand Challenge • SPAM filtering

2

What is A.I.? Typically in computer science classes, you will program a computer to perform some well defined computation/task in some closed/static environment. Example? A.I. problems are more challenging: Win against many opponents at checkers/chess. Learn to recognize what differentiates a male from a female face. Rank order the pages on the internet in terms of authority.

As per text, modern A.I. is designing computer programs to behave rationally. 3

Acting Like a Human

Many prizes (millions of $$$) for passing the Turing test … but not a lot of effort CSI 535 - Introduction to A.I. Lecture 1

4

Key Tasks/Problems in A.I. Searching for shortest paths/solutions in the presence of obstacles/constraints. Search for best sequence of moves in an adversarial environment. Reasoning in the presence of certainty. Reasoning in the presence of uncertainty. Learning from +ve and -ve feedback. Search, Reasoning or Learning?

Route planning? Medical diagnosis systems? Chess, Checkers, Rubik's cube? Theorem proving? Creating a system for voice recognition?

5

Examples of A.I. Applications Practical Motivation S:Planning (i.e. a shortest path planning in video games) R:Medical diagnosis systems L:Voice recognition (i.e. checking flight status) L:Written letter recognition (i.e. scanning of zip codes on letters)

Academic Motivation S:Game playing (i.e. Checkers, Rubik’s cube, Chess) R:Theorem proving A.I. Is practical. But I will give many simple examples for clarity, text book’s co-author (Norvig) is the CTO (aka Director of Search Quality) @ Google. 6

Programming Assignments We'll give you the framework in JAVA, you'll write the intelligent parts. All assignments have a relative performance grading component. 1st Assignment – Searching for the Shortest Route to Climb Mt St Helens. 2nd Assignment – Searching for the best series of moves to win at Connect-4. 3rd Assignment??? - Learning from 1000's of games the best strategy to beat an opponent at Blackjack (aka 21). 7

Many of the Topics Will Involve: a) Clever problem representation b) A well defined objective function c) Algorithms with strong performance bounds/guarantees. Lets go over a) and b) for the three assignments For each break into groups of five and discuss for five minutes and we'll go around the class. 8

Shortest Path to Climb Mt St Helens

9

An Example

10

Searching for the Best Series of Moves for Connect-4

11

Learning to Play Black Jack From 1000's of Games

12

Syllabus For Next 3 Weeks Module Name

R&N 2nd Edition

a) Uninformed/Informed Search Efficient representation of problems Basic uninformed search Informed search – A* Algorithm Analysis of A* algorithm

p59-72 p73-87 Chapter 4 Additional slides

b) Search in an Adversarial Enviornment Game assumptions: zero-sum etc. Basic mini-max search Expansion of game assumptions Alpha-beta pruning Analysis of algorithms

P161-163 P163-167 P171-173 P167-171 Additional slides 13

How Things Work • Larger class so we both need to be efficient • Lectures MWF – I'll give you the reading material for the upcoming and next week. – Read before coming to class

• Homeworks are to reinforce material – Makes the exam easier – On Friday I'll hand out problem sets (typically from previous exams) – You hand them in on Tuesday 5pm – TA will go over solutions during section

• Programming assignments – Learning beyond the book. Helps solidify and internalize the material

14

Logistics and Next Lecture - 1 Text: Russell and Norvig – AI, 2nd Edition or 3rd Edition Mailing List – class announcements, discussion newsgroup, slides on my website: www.cs.ucdavis.edu/~davidson/courses/ECS170-S11.html. TA : Chenguang ([email protected]) (email us both) Meeting times: Class MWF 4:10pm-5pm HARING 1204 Section W 03:10-04:00pm Chem 176 My Office hours:MWF:1:[email protected] Chenguang's office hours TBA by email Your expected to attend lectures. 2 x exams on fundamentals (45%), short problem sets and attendance (based on previous exam questions) (10%), 3 assignments (45%). 15

Logistics and Next Lecture - 2 Collaboration is allowed (except during exams). But your goal is to learn how to do the homeworks yourself. Write up the homework solution by YOURSELF using your own word/style. Your goal is to demonstrate that you understand how to do the problem. Please hand the homeworks in the appropriate homework box in Kemper Hall. Homework will not be accepted in class, email, or delivered in any way other. For programming assignments we will use the “handin” program. Review of this material – Chapter 1 – (RN) 16