cat /proc/claus

Today professor Fukunaga from the Tokyo University of Engineering (Toukoudai), came to Toudai to make an invited lecture about Heuristics. I had a very nice chance to talk to him after his lecture, where we discussed some questions about my research which are likely to result in a chapter of my thesis (and possibly a publication somewhere), and I also had a chance to pick his brain about his opinions on cooperative research here in Japan and in the United States.

I always had some problems with Cooperative Research. It is in my nature to try and get people to participate in what I’m doing - be it parties, or my hobbies, or anything - and yet I had very little success in trying to do some cooperative research. Until very recently, I was never able to do it at all. Fortunately, I think I’m finally getting the hang of what exactly you can and cannot expect to share with other people in terms of work and research tasks.

===

Later this night I had a meeting with a group called the “Asia Interaction Club”, which was introduced to me by Chaminda. “International Clubs” like this one, for those who have never been in an university in Japan, are student associations which aim to have meetings to integrate foreign and local students. An interesting thing about this one is that they concentrate mostly on
east asian foreign students, while most of the other International Clubs I have seen before seemed to just consider tall blue eyed blond foreigners as “proper” international students. The meeting was quite fun, as they were planning for an “games afternoon” event in the near future and I was able to give many suggestions of what kinds of games to run. Managed to take my head off my many research tasks I’m swamped to these days (the fact that some of the results from my experiments came today and were quite positive also made the night meeting a bit more fun, I guess :-P)

===

After watching the move “Us Now”, like I blogged abotu a few days ago, I made my Couchsurfing profile. Since then (about 3 days ago), I already got more than 10 requests for couchsurfing for the next month - half of them for the next week - WOW! I wonder if october is holiday season in Europe or something like that. Of course, I could not accept many of them because of my thesis or stuff, but I actually got guests staying at home for the most part of the next few days. I’ll report about it as they come.

Late June, I wrote into my calendar “Thesis countdown - 6 months”. Today I stopped to consolitade my tasks, and just realized that I’m down to two and a half months until my first thesis deadline (the pre-thesis comitee - don’t really have a good name for this in english). Of course, I’m nowhere near 50% done.

So things are picking up in the work/research front. Among my current tasks, I’m writing experiments for two different papers, one with two researchers (one from the U.S. and one from Brazil) who I met at GECCO, and the other suggested by a Japanese professor which visited the lab two weeks ago on a local conference. Also, I’m organizing the arrival of a new Post Doc student in november - making sure he has a desk to study, a roof to sleep under, etc. Finally, I’m writing two papers, one which is the translation of a paper recently published by my laboratory - not sure how that will turn out yet - the other one a paper for a local conference in Okinawa in December.

Between all this stuff, my lab days have been quite busy - but that is not to say that I haven’t enjoyed my free time, though. Mainly I have been doing Geocaching in my weekends - a hobby which consists of talking long walks to find small treasures hiding in interesting/difficult places, with the help of a GPS. I also have an RPG campaign going on, which keeps my creative juices flowing. Next on the schedule is a part-time job playing with some kids on a costume at an English School during Halloween, and a gaming party the next week or so.

I’m also halfway through “Salt: A World History”, a genial book from Mark Kurlansky, where he describes the role salt has played in many different societies over the centuries. Sometimes I can’t but feel that he is exaggerating on the importance of salt for
armies, governments and revolutions. I’ll try to write more about the book once I’m done with it (maybe next week or so).

So things are going mostly well - thesis is coming by and the pace is picking up fast. My only options are to ride the wave or get drowned in worries, so …

When I started my research work in the University of Tokyo, my first topic was the study of Ant Colony Clustering algorithms (ACC). I tried to use some Meta-GA to improve ACC, which was believed at the time to be very sensitive on a number of parameters. It was a fun and exciting research, and I read a slew of very interesting papers. However, eventually I hit a wall on that research, in which the clustering generated by ACC would be too fragmented (one class was broken into many subclasses). I couldn’t solve that problem, so I eventually gave up on this line of research and switched to portfolio optimization instead.

Today, a fellow swarm researcher, Vitorino Ramos sent a link via twitter on a new paper about ACC. The paper was a modeling from ACC to SOM that addressed exactly the “too many clusters” problem I was having. Their key idea was that the ACC could be maping as a sample method for neighborhood clustering. So they kinda got rid of the ants, but solved the problem. It was a bit sad to see someone else solve that problem I couldn’t solve a few years ago, but it was also nice that they cited my ACC paper (even including a figure from it), when describing the problem with traditional ACC.

Got a bit Nostalgic with this…

This weekend, I prepared a survey paper for a Japanese Symposium (which, incidentally, will become the basis of the bibliography chapter of my thesis :-P). I decided that it would be a great opportunity to leave the stone age and start using some bibliography management software. Following suggestions from friends, I tried out Mendeley, Zotero and Jabref - these are my impressions:

Mendeley is developed by some people from last.fm and skype. It has many very sexy features, including the ability to drag a PDF into it, which will create a new entry and pre-fill it with information gathered from the PDF. Mendeley can also update an entry’s fields by searching for its title on the web (google scholar), and the accuracy for this one is very good. The interface was clean and very intuitive - I could easily find all functions that I needed to without having to think/click too much about it, specially the addition of tags/comments to and article.

But (unfortunately, there is always a big But), Mendeley had some pretty big problems. The first one is that it would crash too often - After adding just a dozen or so entries, I would find the desktop application crashing without notice - and worse yet, without error messages or memory dumps that would help me track the problem down. In one of these crashes I lost my entire database (fortunately I was just trying the program at the time, no real data was lost). A secondary problem was that the search function was a bit buggy - If I added a keyword, say, finance, to 6 papers, searching for this keyword would only return 3 or 4 of them.

I hope that they fix those bugs eventually, but until they do, Mendeley is unusable for me

Zotero is a firefox add-on that also claims to be able to search for paper details on the net. This is already one letdown (at least for me) - I want a standalone program, not something riding on a Firefox instance (specially because Firefox is SUCH a memory hog!). Zotero won’t even reside inside a tab, it must occupy the root screen of FF. Also, because it is an add-on, its data (including stored pdfs) is saved in a bizarre location inside the hidden mozilla folder - it would be cool if this was configurable, since it would make things easier to back up. These aside, it had many of the same features that I liked in Mendeley, like searching for paper information on the web, and getting info from PDF’s - although they were a bit clunkier to use - for instance, if Zotero couldn’t find the info from a PDF you dragged into it, you couldn’t just enter the info yourself in the entry, you had to create a new one, and drag the pdf into it. They also had a cool feature to link entries as “related” between themselves.

However, all of the above is pointless when Zotero simply chocked on my bibfile. Not sure if it was FF or Zotero’s fault, but trying to import my references file into Zotero would result in FF locking. Next.

Jabref is a java based bib management software. After installation, it read my bibliography file immediately, without glitches. The program was stable, with powerful options to select/search/group papers. However, the interface was very obscure - it took me a while to find the button I had to click to edit an entry (couldn’t just double click it), and there is no option to add tags (I can add keywords to a field and text-search it, but it is just not as easily automateable). Jabref can only search on IEEExplore, and can’t read PDF metadata (it can, interestingly enough, WRITE pdf metadata though - I wonder what is the point of it).

Anyway, after playing with it for about one hour or so, I got the hang of where most of the functions are. It could really use a “search the net for details of this paper” function that searched more than IEEExplore, or a “read PDF metadata function” for convenience of use, but other than that, jabref is functional enough to last me until the other options get their act together.

In the last day of GECCO, there was little in terms of the conference itself. First we had the GECCO business meeting. A number of CFP were announced, including the CFP for GECCO 2010, in Portland, Oregon. It was also said that there was a plan to alternate GECCO between north america and Europe from them on - I hope it works, I think it would be nice. After that, the winners of the different competitions and awards were announced - no big surprises.

After the business meeting we had a plenary talk from professor J. Holland, who talked about the development of evolutionary computation in the past and present, tying it with the idea of complex systems. It reminded me how nice it is to be in a field where you can still hear talks from its pioneers.

After Prof. Holland’s talk, I attended to one of the Late Breaking Papers sessions, to see the work of Nunes that I mentioned in the previous post. It was interesting, and seemed to fit well with the work that other people in my lab are doing for finance.

GECCO was over, and then I went for lunch with many of the people I were hanging with in the conference. Supposedly, there was a “Conference Lunch”, in a restaurant in Chinatown, but when we got there we could see no one from GECCO - just shrimps. Shrimps. That chinese restaurant, for some reason, had 90% of its dishes composed of shrimps - fried shrimp, boiled shrimp, raw shrimp, yakisoba with shrimp, shrimp pie… some people in our group were getting quite sick of it, but I was loving!

After the lunch, Verena, who was with us, said she was going Geocaching - and having heard about it before, I invited myself to accompany her, out of curiosity. I must say it was one of the coolest things I have done in a while. The idea of Geocaching is that people hide small caches in out of the way locations, or in the city, and then publish GPS coordinates for those caches. You get those coordinates and try to find them. When you do, you can sign a log book, and see what other people have left in the same cache.

Sounds simple enough, but for me the most amazing part of the experience was that 1- it took you to places you would never otherwise see in a town - It seems to me a great thing to do when you are in a place you have never been before (like the conference) 2- it also makes you see places you see every day in a very different way - something is hidden somewhere you pass by every day. Verena was quite a pro at it already, and in about 8 hour walking, we found almost a dozen caches in Mount Royal and the old town. We walked a total of almost 15 km! Quite an excercise too!

And that ended my short trip to Montreal It was a good chance to make contacts with new people, re-establish old ones, see new research, and even do quite a bit of tourism.