Markoviana: A reading group

On Probabilistic Reasoning in AI

Spring 2005

Friday 2:30-4:30pm. BY 576

Coordinated by: Subbarao Kambhampati

Mailing List Archive

Weekly meetings. Each meeting approx ~2 hours.

Students encouraged to register for at least 1 credit of independent study (mostly as a sign of serious commitment).

(you can take it also for 2 credits—which will involve making one presentation at least

and 3 credits which will involve making 2 presentations. The grade will be Pass/Fail.)

Everyone is expected to read the week’s material before showing up.

This is not a formal lecture course. We will read papers/chapters together and participants

will give presentations on various weeks.

Each participant must lead at least one reading session.

Current schedule:

1/28: R&N Chapter 15, Sections 15.1-15.3. Speaker: Will Cushing. Here are the notes.

2/4: Rabiner’s tutorial on HMMs (also the chapter on HMMs from Durbin et al). Speaker: Fatih Gelgi. Here are the notes

2/11: Kalman Filters (R&N Chapter 15, Sections 15.4 + Kalman filter tutorial). Speaker: Srini Vadrevu. Here are the notes.

2/18: Particle filters (R&N 15.5 and Particle Filter Tutorial). Speaker: Will Cushing. Here are Slides used

2/25: Monte Carlo Localization for Robots; Inferring transportation routines: Speaker: Dan Bryce . Here are slides used.

3/4: Statistical Learning – R&N 20 (1,2) (Heckerman Tutorial is another reference) [ Jicheng's Slides] (A more complete but uncommented set is here)

3/11 : Statistical learning (EM algorithm) R&N 20.3 + Dellaert's paper). [R&N slides on EM]. [Fatih's Frank Dellart Slides]

3/18: Spring break!

3/25: Cancelled due to AAAI exhaustion

4/1: Structural Probabilistic Relational Models: Inference/Learning (R&N 14.6 + Getoor et. al. chapter) [NamTran]

4/8: Markov Networks (undirected graphical models) (Koller & Friedman, Chapter 5)[Srini Vadrevu]

4/15: Conditional Random Fields

4/22: POMDP (paper?) [Dan Bryce]

4/29: Markov Games

========================

Kernel methods—Support vector machines (paper TBDr) [Mariano]

Utility and MDPs (R&N 16 and 17) [Menkes]

Reinforcement learning (R&N 21) [J Benton]

(semester ends 5/3)

Related Sites:

• Pedro Domingo’s course in U. Wash seems very relevant..
• Max Welling’s Machine learning notes are also quite useful
• See also his Caltech course notes and his Kernel notes..
• Tom Dietterich’s course on probabilistic relational models
• Brian William’s course on robotics/planning
• SLAM summer School

------------Expected Evolution

Topic	Readings	Speakers
Filtering and estimation --Hidden Markov Models --Kalman Filters --Particle Filters --Plan recognition	R&N chapter on Temporal reasoning Chapter 15 (15.1-15.5) [2 weeks] ] Paper on particle filters and kalman filters AAAI 2004 best paper VLDB 2004 best paper	Srini Vadrevu
Statistical learning EM etc Kernel machines and support vector machines	R&N chapter 20 Source for Kernel m/c?
MDP and POMDP --MDP methods --POMDP basic solution methods	MDP: LRTDP or ICAPS 2005 paper from CMU POMDP??
Reinforcement learning	R&N chapter on reinforcement learning
Markov Games (Multi-agent MDPs)
Bayes Nets; MCMC etc First-order probabilistic inference Markov Random Fields	From R&N and Koller book