WADS 2007: Days 2 & 3

D A Y 2

Session 5 (Invited Talk)
The day started off with an invited talk from Michael Langston from the University of Tennesse. The talk was on challenges in data analysis, and how sometimes you can have beautiful data to work with (good correlation, easy to work with, e.t.c.) and sometimes you can have tough data to work with. Although not particularly intriguing to myself (I'm still leaning towards a more theoretical area of research), the things he described are exactly the things I have experienced this summer with my cross-disciplinary work. All and all this was a good talk.

Session 6B
I missed the very first presentation in this session, but the second one was about Steiner trees again (everyone loves a Steiner tree). Not quite computing, because that is NP-Complete The session ended off with computing minimum-depth planar graph embeddings in quartic time. I enjoy graphs so I also enjoyed both of these talks!

Session 7A/7b
This session really wasn't doing it for me so I skipped it. I probably should have looked into the computational geometry presentations taking place in 7A though because it is starting to interest me somewhat.

Session 8B
I checked out some parameterized algorithms and kernelization algorithms. Nothing much to report besides the fact that it wasn't overly exciting! For those who are unfamiliar with kernelization, it's simply taking a problem instance and breaking it down into a smaller problem instance. From here we solve this smaller problem instance and from it we can extract the solution to the original problem. This is a common technique in creating algorithms that have good bounds for running times when analyzed parametrically.

Conference Dinner
Lobster! Nothing to say about the dinner besides for the fact that it was delicious. I joined a few of the PHD/Masters students - Aaron Lee from Carleton University, Cora Borradaile from Brown University and another guy from Calgary whose name escapes me at the moment - at the pub next to the Alexander Keith's brewery after for a couple of drinks. Good times!


D A Y 3

Session 9B
I sort of slept in through these...

Session 10B
Approximation of graph distances and shortest paths. Nothing particularly interesting to point out. I just enjoyed going to all of these graph theory talks because I'm not completely out of the blue when they mention crazy terms and the like.

Session 11B
To end off the conference were two sessions with four presentations. The first two involved suffix arrays. One of the grad students which I was fairly familiar with over the course of the past couple of days, Orgad Keller from Israel, done an excellent job at presenting his work (clarity, leaving out the overly complicated details, e.t.c.). The third talk was interesting; it dealt with streaming algorithms, those which receive input in a stream and use much less space (a fuzzy term, I know) than the actual amount of data. They proposed optimal (or maybe ti was near optimal) ways of solving the straggler identification problem. The last presentation I left for, because it was on TCP acknowledgment.

It was all over and Jason was saddened because he enjoyed his time at Dalhousie. But alas, all good things must come to and end. Back to the ol' grind now and in a couple of weeks school starts again. That's kind of exciting too. . . for now.

WADS 2007: Day #1

I figure I should write a blog entry now before I actually forget what happened yesterday!

Session 1 (Invited Talk)
The day started with a presentation whose major title was "Finding small holes" given by Jeff Erickson. It was a small talk on computational topology. It was surprisingly interesting and I was very intrigued overall by the area. The presentation placed an emphasis on a practical application of itself, which was surface reconstruction from potentially noisy data. The processes described showed how surface reconstruction could be done much better by taking advantage of computational topology. No actual algorithms were presented, but the ideas to establish the algorithms were.

For all of you with low G.P.A.s and thinking you wouldn't really be able to go beyond an undergraduate degree because of it, think again. Jeff Erickson cracked a joke about him being potentially one of the lower G.P.A.s amongst most professors with a 2.4/4.0. Doing well in undergraduate studies doesn't necessarily mean you're prepared to do research and be a professor, and similarly for not doing so well you could be perfect through graduate studies. If it's something you would really like to do, get out there and do it!

Session 2B
Nothing particularly interesting here besides for the last talk of the session. At first I thought it was going to be ridiculous because they themed their talk around Donald Duck and Uncle Scrooge, but it was actually quite interesting. The paper was called "The Stackelberg minimum spanning tree game." Their problem was to take a graph with some known and unknown edge weights, and fill in the edges with unknown weights such that the MST maximizes the sum of the values of the weights given to the unknown edges (sorry if I lost you there). One of the characters which presented this talk was Erik Demaine, a guy who came to Dalhousie at age 14 (I think that's right). He's 26 now and is a professor at MIT. The man is brilliant and quite comical also!

Session 3A
I probably should have went to session B here because this session was completely about graph drawing, which isn't a real particular interest of mine. It was cute, but that's about it.

Session 4A
A couple of interesting computational geometry presentations in this session: steiner trees, art gallery problems, and Delaunay triangulations. The last one was particularly interesting because of the reference to one of its practical applications, which is triangulating GIS data for 3D representation. I enjoyed these talks and I think it gives computational geometry a +1 on my list of things I would potentially do for my thesis/grad studies.

All in all, a great day!

GSoC + WADS 2007

The GSoC is nearing an end and there's not much to do really. I think once I get back home on Saturday I'm going to completely look over my Geometry module and see what little things I can do to fix it up (documentation, refactoring, etc).

The WADS 2007 conference reception was last night, and I met some interesting people from all over the world. The talks seem like they'll be fairly interesting overall, with some neat algorithms coming forth. For now I'm lacking something to blog about, but after the day is over and I have taken in some of these presentations, hopefully I'll have something interesting to write about! Note that I'm really liking this Dalhousie CS building over the CS department at MUN (i.e., our corner in the engineering building)