Two other presentations at MVIP 2005

Today and Yesterday, 3rd Iranian conference on Machine Vision and Image Processing was held at our university: University of Tehran – Department of Electrical and Computer Engineering. I have not submitted any paper to it; however, I had two presentations!!!
Like the last week, I presented Hossein Mobahi’s papers entitled “A Comparative Study on Geometric and Holistic Representations for Facial Expression Recognition” and “Vision Based Fruit Inspection Using Independent Component Analysis”. I presented both of them well, but some people asked me a few strange questions about the imaging and camera conditions though I had told them before that I did not do anything in this research. These three recent presentations were good practice for me as I understand that not only it is not necessary to do a real research to present a research, but also you can present it rather well by reading its report just before the presentation. (: Nevertheless, I do not intend to do any similar experience in a near future. (;

P.S: Yeap! I do not work on machine vision. But it was a quite natural to understand their concerns and even devise a method in their field if you know that “an image is data to be processed”, “there are some methods to process image data which is 2D extension (or specialization) of general signal processing methods”, “patterns can be recognized”, and more importantly “pattern recognition is a kind of decision making”. Everything else is similar.

Michael Bowling and Manuela Veloso, Multiagent Learning using a Variable Learning Rate

Michael Bowling and Manuela Veloso, “Multiagent Learning using a Variable Learning Rate,” Artificial Intelligence, 2002.

This is the first paper I write about in my weblog in my series of multi-agent reinforcement learning papers. I had seen this paper about a year ago, but I did not read it as I thought that changing learning rate is not a real solution to the problem and I supposed that this paper is a kind of ad-hoc method. In that time, I was not that concerned about learning in multi-agent learning from the game theoretic perspectives, so I was not that aware of this paper. Now, it is apparent that I was not that correct. The results of this paper is interesting, Michael Bowling and Manuela Veloso tried to use as much mathematics as possible, and more importantly, his approach to the problem is really insightful.
I do not discuss the paper much, but I tries to write about my concerns about it. Before going on, it is mandatory to mention that I am new in this MAS learning and game theory; so concepts such as Nash equilibrium and its importance like that are not much clear for me.
Continue reading “Michael Bowling and Manuela Veloso, Multiagent Learning using a Variable Learning Rate”

The Effect of Reinforcement Signal Error in Reinforcement Learning

TITLE: The Effect of Reinforcement Signal Error in Reinforcement Learning (Translated)

ABSTRACT: Designing reinforcement signal is fundamental problem in reinforcement learning. Intelligent agent’s designer can guide the learner agent to its desired behavior by selecting an appropriate reinforcement signal. However, there is no general methodology for designing that signal and the designed signal is different from the unknown ideal one in many cases. In this paper, this difference is considered as a bounded-norm error in reinforcement signal and its effects on the value function and the policy of the agent is calculated as some upper bounds. In the end, the mathematical results are tested in an experiment. (Translated)

This is my The Computer Society of Iran Computer Conference (CSICC) 2005 paper. As it is written in Persian, I translate its abstract and put it in this weblog. It is probable that I revise it and send it to an international conference or even journal.

My Presentations at The Computer Society of Iran Computer Conference (CSICC) 2005

The Computer Society of Iran Computer Conference (CSICC) 2005 was held from Tuesday to Thursday. I had a paper on it which was presented as a poster. Its acceptance as a poster made me quite angry at that time as I became aware that the reason of this kind of acceptance was that one of reviewers did not like my writing style!!! Yes! It is unbelievable that s/he did not evaluate my paper’s scientific qualification, but assessed its style. I can’t explain the exact nature of the problem as the paper is written in Persian but it may be helpful that I note that my writing style was a new methodology of writing in Persian which is used by some well-known and professional authors and poets. Anyway, I had the chance to present my poster to two individuals only: my girlfriend and another person. (:
It is not the end of story … It was 11PM of Wednesday that one of my friends, H.M., mailed me that if I could present his paper. He had submitted a paper to the conference; however, he is not in the country right now. I had written him well before that if he wanted me to present his paper, he would send its PowerPoint file a week before. But he didn’t do so and I received his PPT file at 12PM, Thursday and the session was scheduled to start at 2PM that day!!! Although no wise man would accept to do so, I preferred to present his paper as a very interesting and exciting experience. I read his paper very fast, and then took a look at his PPT once. Then, I rehearsed the presentation for myself. It was 1:25PM that I left home. The situation became more exciting as the traffic was terrific. I arrived the session at 2:25PM and started my presentation 5 min afterward. Fortunately, I presented very well (considering all the conditions, of course!) and I could even answer the questions of Dr.Badi-i and Dr.Meybodi.

Designing a Mobile Robot’s Mind: A New Job!

I have found a new and interesting job: designing a mobile robot’s mind! I have started working on designing necessary decision making system to control a small mobile robot. It ought to act as a vacuum cleaner or something similar. It is not apparent for me what I should do exactly; however, I designed a simple two layer subsumption architecture that wanders around while avoid obstacles. If I am true about my position in the group, I intented to design three other modules in addition to what I have implemented: map building, self-localization, and dirt-cleaning planner to cover the whole map efficiently. I have some ideas on all of them that I will discuss them gradually.

Cybernetical Semiology

Tonight, Ramin and I had a talk on the relation between engineering aspects of context and its semiological implication. The result can be named Cybernetical Semiology.
What is your idea about it? Is there any good relation between these two fields? I believe that what people name “context” in AI (and related fields) is what semiologists call “text”.

Martin Jagersand, Image Based Visual Simulation and Tele-Assisted Robot Control

[In this semester, I want to read many papers on different aspects of artificial intelligence concentrating on machine learning (in general), reinforcement learning (in particular), evolutionary mechanisms, intelligent robotics, and … . In this way, I try to write my notes and comments on the papers that I read. I will submit my comments to the papers’ authors and see if I can get a good feedback.
I will find the optimal way of doing so gradually. Different parameters affect the behavior ranging from my time to the benefit of this work for myself. Any suggestion or comments is highly appreciated.]

Martin Jagersand, “Image Based Visual Simulation and Tele-Assisted Robot Control,” IROS 97.
Continue reading “Martin Jagersand, Image Based Visual Simulation and Tele-Assisted Robot Control”

After Submission days

To build a really intelligent machine, many different problems must be solved (borrowing the idea from The Society of Minds, huh?!). However, it is well-known that 1) we do not know which problems must be solved, and 2) we have not solved many known problems (such as credit assignment in multi-agent systems) yet. So, I cannot go any further and this post will be finished by the end of this dot point! (no! I was joking! There is not dot point nearby!)
At last, I have reported a part of my recent researches to a journal and a conference in the previous week. I feel much better now as the journal one took all my time and energy to be prepared. Actually, it is my first really important journal paper (if it would be published) that its preparation and revision time took five anxious months(!); no! I do not exaggerate. Now, it is finished and I have stopped writing reports for a while and started reading papers and thinking about new problems again. I think that I am from those people who would be wasted by writing long technical reports. In writing them, I move forward very slowly (no! I do not type slowly or even make new sentences that way. However, I think much about what I want to write and this make the process poky) and all my creativity power wastes. However, I know that I must start writing by a few days in order to prepare another paper! ooppss!

~sleeping for 24 hours

It is interesting: I have not slept for 24 hours. I was working on my paper and doing some discussion about it with my advisor. Also I wrote a few e-mails to professors to see whether they need a student or not. I was trying to do so for a professor in the University of Toronto but I could not do so as I was too exhausted. As Marvin Minsky said, tired person cannot use their symbolic manipulators well and probably they do many grammatical mistakes. For me as a non-native English user, this is not only limited to “grammar” but also includes “vocabulary”. For instance, I may think for a two or three seconds to find a simple word (of course this is not for every word!). Summarizing (!), I am too tired and want to sleep as soon as possible.

emmm … a little note about applying … oh! it is a difficult task as most people around the world has no fund, no free position, or something like that!
Note to visitors: if you are a university professor, and want a student interested in machine learning, evolutionary computation, robotics, and …, please drop me a mail at SoloGen@removeSoloGen.net (remove “remove” word from the address). 😀 (We have an idiom which its translation is “Shooting an arrow into the darkness” (its original form is “Tiri dar tariki”) that means trying without generating the whole possible states tree and plan in it using A* search(!).)

Machine Learning Weblog

Believe me or not, but I do not surf to find new technical weblogs these busy days, but surprisingly, they suddenly appear in front of my face.
I found this Machine Learning weblog which is written by John Langford an interesting one. However, I doubt that he could continue this way: he writes as he is writing his “Introduction to Machine Learning” textbook. I guess that he will change this style or the weblog will be stopped by a few months. However, it is only a guess and I hope that it was incorrect as I enjoy a technical weblog like this in my field. A possible solution may be letting other researchers to write in it. For instance, a collection of ten person might pass the critical mass(!) for this technical weblog to be self-living one.