I need you to fill out the student evaluation forms and return them to Student Services (first floor of Swearingen). I have placed them in the folder hanging from the corkboard outside my office. My instructor code is 5023.
I have added the formal description of the final project. Remember that you will need to come talk to me to get it approved.
Here are tests 2 from 2005 and 2006. Since I taught the material in different order you will find that some questions deal with material that we already covered in test 1. Similarly, some of the questions from the previous test-1s I gave you correspond to material for this test. Now that you have all tests, you have all the questions.
I have posted the new and final problem set 3. This should not only be fun but will be useful for future classes.
Looks like everybody is celebrating Logo's 40th birthday. At least, now you can appreciate why everyone is so excited.
If you are an APOGEE student taking this class you need to email me to arrange for taking your test.
In case you didn't catch it in class, here are links to the AAMAS papers and here is a link to the Autonomous Agents and Multi-Agent Systems journal (if you are accessing it from within USC you will be able to download all paper). You will want to start browsing these to get ideas for your final project. Also note that I added a paper to Wednesday's lecture.
The latest issue of IEEE Intelligent Systems contains two papers on real-world applications of multiagent systems.
The first news story, "Multiagent System Helps Train for the Unthinkable," looks at DEFACTO, a prototype firefighter-training system that simulates multiple disasters in a city. The second news story, "The Berkeley Parser: High-Quality Grammar, Automatically," reports on a system that automatically learns a language's grammar and then determines the most likely structure of a sequence of words in that language.
Over time, people change whom they interact with and where they are. Understanding how changes in social and geospatial relations interact is critical to a number of areas--from general social change to counternarcotics and counterterrorism. For instance, as terrorists attempt different tasks, they move to new locations and interact with different groups. This article introduces a simple theoretical multiagent model for reasoning about the criticality of agents and locations as agent interactions coevolve in geographical and social spaces. The model simulates social changes in whom agents interact with and spatial changes in where they relocate to as a function of learning and social-network evolutions. The analysis suggests that terrorists will disperse around the world rather than gather at a specific location. However, terrorists who have been the center of social networks will stay the same. This model helps us gain insights into the complexities of organizations evolving in the social and the geospatial dimensions simultaneously. This article is part of a special issue on social computing.
I have posted PS 2. Let me know if it needs clarifications. Or, rather, let me know which clarifications are needed.
If you are interested in develving deeper into game theory I highly recommend you pickup Binmore's new textbook.
As I have mentioned to several of you, I am flexible with the deadlines. This is a 700-level class and I expect most of you have thesis to work on so as long as you do the work I am happy. Also, note that your really want to use the link primitives in NetLogo for building and layingout your graph. There is a whole section on the NetLogo manual that explains how to use them. There are also sample graph models under the "models library".
The AWCgc implementation on the netlogomas page does not work. I updated to NetLogo 4.0 and in the process somewhow managed to break the program. Luckily, I found an old (3.x) version of the program which does appear to work: AWCgc.nlogo
Seven Minute Madness
Research Presentations of the Faculty
Computer Science and Engineering
In recent years the Department of Computer Science and Engineering has sponsored the Seven Minute Madness presentations where faculty have 7 minutes to overview their research. This years Madness presentations are scheduled for September 7th and 14th. Come hear about some of the exciting research in computing at USC.
September 7, 2007 @ 3:30PM in B201 300 Main StreetI have posted PS1 for those of you who are dying to get started with some serious coding. It is a non-trivial exercise in algorithm design and NetLogo programming. Have fun! I will explain more in class in a couple of weeks after we cover distributed contraint optimization.
As I mentioned in class, PS0 has been posted. You do not need to turn this in but I really encourage you to do it in order to become familiar with NetLogo. PS1 will be a bit more challenging.
The version of the textbook we will be using is now out. See my blog post about it.