CSE 571: Artificial Intelligence - PowerPoint Presentation

Slide1

CSE 571: Artificial Intelligence

Instructor:

Subbarao

Kambhampati

rao@asu.edu

Homepage:

http://rakaposhi.eas.asu.edu/cse571

Office Hours: Right after the class

3:15—4:15pm BY560Slide2

History

At ASU, CSE 471/598 has been taught as the main introductory AI course

Normally taught by either

Rao

or

Huan

Liu

571 has been taught as a graduate level AI course

Didn’t necessarily require 471

Didn’t necessarily have a breadth aspect

Nick

Findler

taught it for a long time and would focus on distributed AI

Chitta

Baral

taught it after Nick and would focus on knowledge representation

Last time

Rao

taught it was in 1996

Looking back at that syllabus, it looks like 571 I taught then is a subset of 471 as I teach nowSlide3

CSE 571 This time?

“Run it as a Graduate Level Follow-on to CSE 471”

Broad objectives

Deeper treatment of some of the 471 topics

More emphasis on tracking current state of the art

Training for literature survey and independent projectsSlide4

Who are you & what do you want?Slide5

What we did in 471

Week

1: Intro; Intelligent agent design [R&N Ch 1, Ch 2]

Week 2: Problem Solving Agents [R&N Ch 3 3.1--3.5]

Week 3: Informed search [R&N Ch 3 3.1--3.5]

Week 4: CSPs and Local Search[R&N Ch 5.1--5.3; Ch 4 4.3]

Week 5: Local Search and Propositional Logic[R&N Ch 4 4.3; Ch 7.1--7.6]

Week 6: Propositional Logic --> Plausible reasoning[R&N Ch 7.1--7.6; [

ch

13 13.1--13.5]]

Week 7: Representations for Reasoning with Uncertainty[ch 13 13.1--13.5]]

Week 8: Bayes Nets: Specification & Inference[ch 13 13.1--13.5]]

Week 9: Bayes Nets: Inference[ch 13 13.1--13.5]] (Here is a fully worked out example of variable elimination)

Week 10: Sampling methods for Bayes net Inference; First-order logic start[ch 13.5; ]

Week 11: Unification, Generalized Modus-Ponens, skolemization and resolution refutation.

Week 12: Reasoning with change



Planning

Week 13: Planning, MDPs & Gametree search

Week 14: Learning Slide6

Chapters

Covered in 471 (Spring 09)

Table of Contents (

Full Version

)

     Preface (

html

);

chapter map

Part I Artificial Intelligence

     1 Introduction

     2 Intelligent Agents

Part II Problem Solving      3 Solving Problems by Searching      4 Informed Search and Exploration      5 Constraint Satisfaction Problems      6 Adversarial Search Part III Knowledge and Reasoning      7 Logical Agents      8 First-Order Logic      9 Inference in First-Order Logic     10 Knowledge Representation Part IV Planning     11 Planning (pdf)     12 Planning and Acting in the Real World

Part V Uncertain Knowledge and Reasoning

    13 Uncertainty

14 Probabilistic Reasoning

 15 Probabilistic Reasoning Over Time

    16

Making Simple Decisions

17

Making Complex Decisions

Part VI Learning

18 Learning from Observations

19 Knowledge in Learning

20 Statistical Learning Methods

21

Reinforcement Learning

Part VII Communicating, Perceiving, and Acting

    22 Communication

    23 Probabilistic Language Processing

    24 Perception

    25 Robotics

Part VIII Conclusions

    26 Philosophical Foundations

   27 AI: Present and Future

   Slide7

Rao:

I could've taught more...I could've taught more, if I'd just...I could've taught more...

Yunsong:

Rao, there are thirty people who are mad at you because you taught too much. Look at them.

Rao:

If I'd made more time...I wasted so much time, you have no idea. If I'd just...

Yunsong:

There will be generations (of bitter people) because of what you did.

Rao:

I didn't do enough.

Yunsong:

You did so much.

Rao:

This slide. We could’ve removed this slide. Why did I keep the slide? Two minutes, right there. Two minutes, two

more minutes.. This music, a bit on reinforcement learning. This review. Two points on bagging and boosting. I could easily have made two for it. At least one. I could’ve gotten one more point across. One more. One more point. A point, Yunsong. For this. I could've gotten one more point across and I didn't. Adieu with an Oscar Schindler Routine.Schindler: I could've got more...I could've got more, if I'd just...I could've got more...Stern: Oskar, there are eleven hundred people who are alive because of you. Look at them.

Schindler: If I'd made more money...I threw away so much money, you have no idea. If I'd just...

Stern: There will be generations because of what you did.

Schindler: I didn't do enough.

Stern: You did so much.

Schindler: This car. Goeth would've bought this car. Why did I keep the car? Ten people, right there. Ten people, ten more people...(He rips the swastika pin from his lapel) This pin, two people. This is gold. Two more people. He would've given me two for it. At least one. He would've given me one. One more. One more person. A person, Stern. For this. I could've gotten one more person and I didn't.

Top few things I would have done if I had more time Statistical Learning Reinforcement Learning; Bagging/Boosting Planning under uncertainty and incompleteness Ideas of induced tree-width Multi-agent X (X=search,learning..) PERCEPTION (Speech; Language…) Be less demanding more often (or even once…)Slide8

Things I Know I want to Cover

Search

Local vs. Systematic

Optimization in continuous domains

Constraint networks

Tree-width concepts; temporal constraint networks

Reasoning: Planning

Temporal planning; belief-space planning, stochastic planning

POMDPs;

DecPOMDPs

?KR: Templated Probabilistic NetworksDynamic probabilistic networks

Relational Probabilistic networks

Learning:

Relational LearningReinforcement learningSlide9

Reading Material…Eclectic

Chapters from the new edition (in preparation) of R&N (in some cases)

First reading: Advanced Search Techniques chapter (Will be distributed in hardcopy)

Chapters from other books

POMDPS from

Thrun

/

Burgard

/Fox

Templated

Graphical models from

Koller &FriedmanCSP/Tree-width stuff from Dechter

Tutorial papers etcSlide10

“Grading”?

3 main ways

Participate in the class actively. Read assigned chapters/papers; submit reviews before the class; take part in the discussion

Learn/Present the state of the art in a sub-area of AI

You will pick papers from IJCAI 2009 as a starting point

http://ijcai.org/papers09/contents.php

Work on a semester-long project

Can be in groups of two (or, in exceptional circumstances, 3)Slide11

Deadlines..

AAMAS deadline: 10/8/09

KR deadline: 11/10/09

ICAPS deadline: 12/16/09

AAAI deadline: 1/15/10

ICML deadline: ~2/10/10Slide12

Discussion

What are the current controversies in AI? What are the hot topics in AI?Slide13

Pendulum Swings in AI

Top-down vs. Bottom-up

Ground vs. Lifted representation

The longer I live the farther down the Chomsky Hierarchy I seem to fall [Fernando Pereira]

Pure Inference and Pure Learning vs. Interleaved inference and learning

Knowledge Engineering vs. Model Learning

Human-aware vs. Slide14

The representational roller-coaster in CSE 471

atomic

propositional/

(factored)

relational

First-order

State-space

search

CSP

Prop logic

Bayes Nets

FOPC

w.o. functions

FOPC

Sit. Calc.

STRIS Planning

MDPs

Min-max

Decision

trees

Semester time



The plot shows the various topics we discussed this semester, and the representational level at which we discussed them. At the minimum

we need to understand every task at the atomic representation level. Once we figure out how to do something at atomic level, we

always strive to do it at higher (propositional, relational, first-order) levels for efficiency and compactness.

During the course we may not discuss certain tasks at higher representation levels either because of lack of time, or because there simply

doesn’t yet exist undergraduate level understanding of that topic at higher levels of representation..Slide15

ideas

Put the

schindler’s

list slide

Make people in the class come up with currents in AI that are most interesting to them

Present the IJCAI statistics

Present the main trends

Present the deadlines

How to make AI commercial?