NCCS/EC lab meeting today, Tue.9.June @ 11a in PII

9.June.2009
11:00 am	to	12:00 pm

Recall that we have our weekly NCCS/EC lab meeting today at 11a in the third floor collaboration area in Partnership II.  We will finalize what kind of summer project we are interested in doing.

Notes on Summer lab meetings

We had a useful kick-off meeting this week, and I thought it would be useful to put a post here for two reasons.  First, there are just a few FYI’s that would be good to know.  Second, the comment thread can be used for discussions of possible summer project, etc.

There were essentially three basic discussion threads at the meeting (as best as I recall):

Weekly Meeting Schedule

Recall that our minimally complicated way to approximate a solution to some of the schedule problems was the following:

• We will meet at IST in the third floor collaboration area of Partnership II for the summer;
• We will meet on Tuesday at 11am on June 9 and June 16;
• The meetings will shift to Friday at 11am from June 26 thereafter.

Relationships in the Various EC-Oriented groups

Dr. Wu and I will likely merge our two groups into one.  Since we more or less operate as one group now, such a merger is mainly logistical / operational involving tedious questions such as what to call the group, how to best coordinate website / email list resources, etc.  Input on these topics is welcome and there’s no urgency.

Also several people during this and earlier meetings have noted in both specific and general ways that we might try to strengthen ties to EPlex.  For one, if we pick a summer project (below) that is interesting to Dr. Stanley and some subset of his students, perhaps we might work together.  For another, we try to increase cross-attendence at lab meetings, etc.  Let’s keep our minds open to ideas on this and when we have a specific notion, Dr. Wu or I can approach Dr. Stanley to get his input, etc.

Summer Project Activities

Recall also that we decided to take a “laid-back / fun” approach to our summer activities, going with Sean’s idea to have some kind of joint project.  Several ideas were discussed but no decisions made.  We are to bring our thoughts to the next meeting and finalize our choice then.  Some ideas that were kicked around:

• Develop an EC/AI pedagogical tool providing interactive and visualization capabilities for explaining how certain algorithms work.  Pro’s: Tool would be useful to all of us at some point and could allow the labs to “produce” something useful for the community; We might interact with Dr. Luke at GMU to see if he is interested in working with us and allowing us to springboard off of all the work he’s done on his “Metaheuristics” notes.  Con’s: There’s no real research problem or interesting intellectual challenge here.
• Consider well-known contest or challenge problems such as those from current or past GECCO, AAAI, and such events … or RoboCup, etc.  Pro’s: Certainly an intellectual challenge; Offers possible comparison to others approaching such problems; Con’s: May require more effort and time than summer provides (allowing for your other duties); Most deadlines for contemporary contests have passed.
• Take a subset of someone’s current research project and focus on helping solve that problem (e.g., Chris H.’s cultural modeling stuff).  Pro’s: Actual research with at least one student already dedicated to the task (justifying the effort); Con’s: Project may not reflect shared interests of the group, asymmetric with respect to projects of the other students.
• Take on a proto-research project on the “inteterested-in” list of either Dr. Wu or myself (e.g., using co-search information extraction methods to learn human choice distinctions; constructing an IEC method for feature extraction & visualization of large, scientific databased; developing a web-based demo for illustrating variational power in collective CASs).  Pro’s: We might be able to use the results to strengthen a future proposal; provides intellectual challenge. Con’s: Problems may still be very ill-posed; Again, interest may be asymmetric, etc.

On that last point, I have several ideas that I can discuss in a bit more detail if people are interestested.  Let me know.

Summer lab meetings start today at 11am

2.June.2009
11:00 am	to	12:00 pm

We kick off our summer lab meetings today at 11am out at IST in research park.  We have a meeting in Partnership II, room 301.  Regular meeting locations will be part of the topics for today, in addition to some of Sean’s comments from the end of Spring semester about how to make the group more interesting, etc.

Adam presents AAMAS paper today, Fri.24.Apr at 2p in HEC 438

24.April.2009
2:00 pm	to	3:00 pm

Today Adam will be presenting a dry-run for his upcoming paper presentation at AAMAS this summer.  The talk is 15 minutes, so there will be time to discus the paper afterward.   We will meet, as usual, in the room HEC 438 from 2p-3p today.

The paper can be downloaded here.  Adam tells us:

“The paper discusses the significant impact that synchroneity (defined

in this paper as the average number of agents that act at any given

time) has on the dynamics of multi-agent systems.  We examine one

problem in particular with both mathematical and algorithmic models

and show how the dynamics become less stable as the number of agents

that act at any time increases.  Just imagine how erratic the behavior

of the stock market would be if everyone bought (or sold) at the exact

same times.  This is the basic idea behind the paper.”

Sean leads Pac-Man AI discussion, Fri.17.Apr, 2p-3p, HEC 438

17.April.2009
2:00 pm	to	3:00 pm

Sean will be leading the discussion this Friday, April 17 at our usual time and location (2p-3p, HEC 438).  He would like to discuss video game AI and how EC techniques can be applied to such, and there are two background articles that will serve as useful pre-readings for this conversation.  The first is an article on Pac-Man, and includes the details of the AI for each of the ghost “personalities” in that game (chapters 2 and 4 are the most relevant).  The second paper is approaches the question of how to evolve team tactics using GP.

Sean would like to lead off the discussion by sharing ideas of how one might develop and/or evolve some of the behaviors of Pac-Man’s foes.

• Pittman, J. (2008) The Pac-Man Dossier. The Gamasutra website.
• Doherty, D., and O’Riordan, C., “Evolving agent-based team tactics for
  combative computer games” [PDF, 200K], Proceedings of AICS 2006 17th Irish
  Artificial Intelligence and Cognitive Science Conference, Belfast,
  Ireland, Sep. 2006, pp 52-61.

Hodge-podge meeting, Fri.10.Apr 2p-3p in HEC 438

10.April.2009
2:00 pm	to	3:00 pm

Sandy will not be able to attend today, so we will not discuss the paper.  We have decided that there are several miscellaneous topics that would be useful to address, so we will meet as usual but the content of the meeting will change.

Sandy will present the paper listed below this Friday, April 10 from 2p-3p in HEC 438.  The technical background for the paper is similar to that of the paper by Keki Burjorjee we discussed several weeks ago.

Differentiable Coarse Graining [PDF]
Jonathan Rowe, Michael Vose, and Alden Wright.

More below the cut.

(more…)

Slides available for Overview of EC Theory

I’ve finally made the few tweaks to the slides for the talk I gave last month at the AI Forum, Overview of EC Theory.  The slides are in PDF, formatted 4up (1.5Mb, 10 pages, 38 slides).

This week’s Friday meeting canceled

There will be no AI forum this week, and we’ve nothing on the EC/NCCS lab schedule.  So concentrate on your studies and GRA duties, instead.

I believe the calendar is open for next week, as well.  Let me know if anyone has something they would like to do.

Discussion thread for room sharing at GECCO

This post begins a discussion thread to help setup potential room-sharing arrangements for GECCO this summer.  It is primarily aimed at student’s in GMU’s GAG, and the various relevant UCF groups:  EPlex, EC lab, and NCCS lab.  Comment directly on this thread if you are interested in finding a roomy.

(Email me if my spam filters are interfering in any way).

Gautham leads discussion today, Fri.27.09 2p-3p in HEC 438

27.March.2009
2:00 pm	to	3:00 pm

Gautham will lead a discussion today in our joint EC/NCCS lab meeting covering several topics from his research including active work and future directions.  As usual, we will meet in HEC 438 from 2:00p to 3:00p.  His discussion will center on two topics:

• Performance of Unbiased Modular Sets: Gautham has sent out a draft copy of a recent paper he, the Dr.’s Garibay, Dr. Wu, and I have had accepted to the GECCO Theory track.  Please read the introduction and conclusion, the discussion be high-level and will not require the technical details.
• Future directions for evolvability research: Gautham will discuss how the OEFF research he has discussed in the past might translate to his more general interest in studying evolvability.  His pre-reading  suggestion is only the Wikipedia page on evolvability