My Research Journal
--a weekly look into the research process in particular and dmp life in general--

Week 1 | Week 2 | Week 3 | Week 4 | Week 5 | Week 6 | Week 7 | Week 8 | Week 9 | Week 10
Week 1: Monday, June 5th thru Sunday, June 11th
My summer started with 3 frantic days of finals & packing, 20-odd long hours of cross-country driving, and 1 very expensive speeding ticket. After I found my dorm by pure dumb luck (which seems to be more or less how I live these days), my roommate was wise enough to see that I am directionally challenged, and took me to the Bright building on Tuesday morning to officially start my DMP summer. The rest of the work week was spent creating accounts, getting keys, meeting advisors and officemates, lots of reading, and general acclimation to the life of a graduate research office in the summer months. I familiarized myself with Dr. Choe's research to date, and began to zero in on the areas that interested me (namely, SIDA). I wrote a paper commentary summarizing and analyzing one of Dr. Choe's papers and attended a GRE seminar that is apparently going to be a weekly ritual, as well as a USRG luncheon. On Friday I set off to enjoy the weekend, checking out the College Station social scene and going to a baseball game at Minute Maid Park. Although the Astros beat the hell out of my beloved Braves, not to mention smoked our starting pitcher in the temple with a line drive, it was a great day for baseball and I loved the stadium. Jeff Francoeur hit a homer and a triple, the seats were the cheapest in the place and we still had a great view, and I managed to not get lost trying to find it...things are looking up. Next week's plan is to finalize a research plan, do some more background reading, and get my website off the ground. [top]
Week 2: Monday, June 12th thru Sunday, June 18th
I spent most of this week continuing to do background research, getting this website up to date, installing the webcam I'll be using for my research, getting to know its C source code, and re-familiarizing myself with C++, which I haven't used since early in my senior year of high school...now I remember why I switched to Java in the first place. I'll be having nightmares of ampersands and asterisks for weeks to come. Anyway, the biggest thing I did this week was to draft a final research plan and talk it over with Dr. Choe. Long story short, I'm going to be implementing the SIDA algorithm on a webcam, attempting to reconstruct the camera's external input based on its internal state. But, there are still a lot of details about the project to be worked out, the biggest being deciding on the proper representation of the camera's "sensory state". We had another GRE seminar and a couple of luncheons--one on grad school stuff, and one a presentation by a TAMU professor about some of her research. Over the weekend we checked out the local nightlife some more and I headed to my cousin's All-Star baseball game...go Burleson County! Next week I'll continue to get to know the camera code and hopefully start to develop some of my own code, as well as start designing the environment and representations for the camera's "brain" and taking a more careful look at some of the image convolution algorithms. [top]
Week 3: Monday, June 19th thru Sunday, June 25th
This week I actually got my hands dirty with some code. I got into the camera code and worked on extracting the image buffer from it, and then writing a simple convolution function to convolve the grayscale image. The camera code is horribly commented and has only limited documentation, and my knowledge of C/C++ has been slow to come back, so it was frustrating at times and in the end, I had to go to Dr. Choe for help figuring out how the code was storing the RGB values for the pixels. There's a lot left to do even for this simple part of the project, but it's been nice to feel like I'm actually accomplishing something, and the rust is starting to wear off. As far as REU/DMP stuff, we had another GRE seminar which mostly consisted of the instructor making fun of our limited verbal reasoning skills, and a few brown-bag lunches to hear about some more research that's being done and how to write technical papers. The GRE guy is a little over the top but the practice tests and strategy ideas will be nice to have come test time. Outside of work, we continued to meet fun people from other summer programs and watched Edmunton fall apart without Roloson, Stackhouse drop Shaq like a linebacker, the Rocket re-launch, and the US soccer team wallow in yellows and reds...all in all a pretty depressing week of sports but at least there was some good company. Over the weekend we went tubing down a man-made river about 3 hours away with a mix of kids from the CS, physics, biology, and chemistry programs. Next week I'll continue to work on the convolution methods and see where that leads me next. [top]
Week 4: Monday, June 26th thru Sunday, July 2nd
This week I was able to get the convolution filters working, including a Difference-of-Gaussian filter. It's very picky about window size and corresponding thetas; if they're not chosen carefully, the image will come back pure black or pure white, because the zero boundary is not crossed. I then moved on to beginning to implement very basic movement algorithms on the camera, first just acclimating myself to its visual limitations and coordinate system, and then working to make it follow some simple patterns, including a "random walk". The biggest DMP/REU event of the week was a lunch talk by Dr. Bjarne Stroustrup, the designer of C++. I wasn't feeling particularly kindly towards C++ on that morning (or ever), but it was interesting nonetheless, and some of the kids got a chance to ask those nagging questions about why, say, reference parameters need to exist. More informative was a grad student panel over lunch, where we got the chance to ask some current TAMU grad students (PhD and Master's) anything under the sun about grad school, research, funding, etc. They were very candid and it was a great chance to learn more about life as a grad student. Outside of work life, I watched my cousin's All-Star baseball team win the District 13 championship and I continue to be amazed by how intense and demanding the kids and their parents are, even though the players are only 9 years old. What I've seen basically reaffirms everything I've learned about Texas sports from...well, Friday Night Lights. Over the weekend I drove to New Orleans to visit a friend at Tulane. And I'm afraid the details of that trip will have to stay in NOLA. [top]
Week 5: Monday, July 3rd thru Sunday, July 9th
This week I actually started putting together the image processing and camera movement algorithms, and familiarizing myself with the SIDA and network-training code from past research. A lot of the code is Matlab and Perl, neither of which I'm too familiar with, but the Matlab, at least, is fairly easy to figure out. I started coding the SIDA algorithm and it subcomponents in C, trying to keep in mind that I will eventually need to integrate it with the OrbitView code. We got the day off for the 4th of July, and although it rained for a good part of the day, I went to a nearby state park with my cousins and spent the day playing wiffle ball and soccer...nothing like beating up on some 9-year-olds for a little confidence booster, eh? We had the week off from DMP stuff since we had an extra talk last week. Over the weekend, a few of us headed to Arlington to watch the Twins and the Rangers play (ie to watch Fransisco Liriano completely dominate the Rangers lineup), and on Sunday we watched the World Cup finals (ie watched Zidane committ the strangest and possibly stupidest foul in the history of the World Cup). Next week I'll try to finish the SIDA coding by midweek and get it integrated with the OrbitView code, which should leave me free to actually begin running some tests. The program is half-over for me now and although I technically don't have much to show as far as results yet, I think I've made some good progress and hopefully will have some interesting things to report soon. I'm also going to have to start thinking about the poster and the paper by the end of next week. [top]
Week 6: Monday, July 10th thru Sunday, July 16th
Most of this week was spent continuing to code SIDA in C. I ran into a little trouble with the math lurking behind the Gabor filters, but Dr. Choe explained some of the intricacies of the formula to me and I was able to get it down. I'm still working on the learning algorithms, but I will hopefully finish those early next week and still be able to produce some concrete results by next Friday. Dr. Choe suggested that I start by simply creating a reward table, and training the camera on that to see what kind of results I get; if they look promising, I'll go ahead and actually let the agent do its own exploring based on a random reward table. I'll start with simply giving the camera a single frozen image, and then hopefully progress on to allowing it to interact in real-time, although that won't happen next week by any stretch of the imagination, and I think will be the most difficult part of the project. We had our last GRE seminar this week, and our weekly office "lab seminar," where various graduate advisees of Dr. Choe give a quick presentation on their work. Outside of the office, I watched the Home Run Derby (otherwise known as "Pirates of the Allegheny") and we went to Buffalo Wild Wings to watch the All-Star game over wings and beer. Mostly we just got dirty looks for cheering for the American League in the middle of a roomful of Astros fans. As luck would have it, we left after the 8th inning, and just missed Michael Young's heroics in the 9th. I didn't do much over the weekend, just relaxed and hung out with my cousins, who returned my wiffle ball domination favors by thoroughly schooling me in 12 straight games of driveway basketball. [top]
Week 7: Monday, July 17th thru Sunday, July 23th
Same old, same old...this week I continued to code and fine-tune SIDA. The big breakthrough was getting the Gabor filters to work. Finally, given a simple image like a line drawn at 45 degrees, the algorithm can match that image to a given Gabor filter and therefore orientation. This is a key part of SIDA, and one that frustrated me for the first half of the week, since I had been getting the exact same orientation no matter what the input image was. I finally realized two things: (1) doubles do strange things in C, and (2) converting to radians before using C's trig functions might be a good idea. We turned in our research abstracts this week, which was a little unnerving since I still don't have any results to show, and also attended a seminar on presentation skills to prepare us for the poster presentations that are coming up all too quickly. Fortunately it sounds like most of the other interns are at about the same place I am, so I'm not too worried. My second paycheck still hasn't arrived and I'm getting beyond frustrated. Not to mention low on funds. Next week I'll start running experiments and compiling data...for real this time, because I only have a few weeks left to get a poster and a paper put together with some substantial results on it. Over the weekend, since one of the guys is from Philadelphia, we went to B-Dub's again to watch the Phillies/Braves game. We were also trying to watch the Twins/Indians game, play the trivia game, and enjoy the food and drink in front of us, so I pretty much absorbed nothing except wings and grain beverages. My buddy Catfish, who lives in San Angelo, came to visit, so I spent the rest of the weekend showing him around College Station and taking it easy. [top]
Week 8: Monday, July 24th thru Sunday, July 30th
I finally got the filters fully functional this week, and finished the SIDA code. Unfortunately, though, there is still something off in the SIDA code, and it's not returning the results we'd hoped for. So, since time is running very short until the poster needs to be completed, Dr. Choe gave me some reward tables from previous Matlab/Perl-based experiments and I'll be running the algorithm from the camera using those tables, instead of actually training the camera first. Hopefully, though, I'll be able to figure out the bug in the SIDA code before I write the paper for the final week. I also got started on the poster this week, using a LaTex template Dr. Choe gave me. I spent most nights and all of Saturday in the lab this week, but on Sunday we did manage to escape to catch Roger Clemens pitching against the Diamondbacks in Houston. Next week will be an exercise in time management as I try to fully integrate the learning algorithm with the camera code, run tests, finish the poster for presentations on Thursday and Friday, and say goodbye to most of the people I've met down here, as pretty much everyone except myself and my roommate finish on the 4th and take off next weekend. [top]
Week 9: Monday, July 31st through Sunday, August 6th
Wow...what a week. The official poster presentation was set for Friday all day, and they added a second, less formal one on Thursday afternoon. I'm afraid to count the number of hours I spent in the lab this week, but let's just say I didn't make it home before 5 a.m. on Thursday. I cut it absolutely to the wire on the poster both days (on Friday I had to carry it away from my body all the way to the building because the ink was still drying and I was already late), but managed to get it done. Thursday I didn't have the results I wanted, so I went back in and re-ran some tests, fixed one last bug, and got some good results that I was able to put on the poster for Friday. Friday was a looooong day but it was interesting to walk around and see what everyone had been working on, and gratifying to see people interested in/impressed by your own work. Pretty much the only time I got out of the office this week was for a REU/DMP dinner at a nice restaurant downtown. It was a nice break and kind of strange to see all of the professors and advisors out of the office. I spent the weekend catching up on sleep and having one last hurrah with all the kids from the programs, since most everyone left on Saturday or Sunday. After all the goodbyes and subsequent recoveries, I finished writing my paper and cleaning up and documenting my code so people besides me can actually read it. All that remains now is some editing of the paper and getting everything submitted to the right places, and making sure I SCP all my files to my account back at school. Next week is going to be mightly lonely, with everyone except me and Kelli gone. It's sad to see everyone leave, since I was really starting to make some good friends. Hopefully, though, with the wonders that are e-mail, facebook, and cell phones, people will stay in touch. I know several people are applying here for grad school and are already planning reunions during visitation weekend in February, so I'm sure some of them will wind up together again. I'm hoping to be able to go to Pennsylvania to visit the boys at some point, and it's fun to know that should I ever end up in Arizona, or Atlanta, or Lousiana, or Oklahoma, I'll have someone to drop in on. [top]
Week 10: Monday, August 7th thru Friday, August 11th
My last week in College Station...I spent most of Monday, Tuesday, and Wednesday finishing up my paper, editing a few last things on my poster (although those aren't going to make much difference now), and cleaning up all my code. Dr. Choe wants the final code so people can work on it in the future, and I want a copy in my own account back home, so I went through and cleaned out all the obsolete files and added in some comments...one of these days I'll start commenting while I write the code, instead of at the end. I submitted all of my final files, finished up the website, and packed up to head out. I don't get to go home yet, though--Friday is my birthday so it's off to Las Vegas for a weekend with my best friends from school back home and my sister, then probably a night in Denver to check out Coors Stadium, and then on to Minnesota...hopefully with no speeding tickets this time. I'll only have 5 days at home before I have to move in for pre-season, which should be interesting considering I've spent every spare second in the last week and a half in the lab instead of working out. As much as I'm going to miss all the people and the lifestyle here, I'm looking forward to getting home and catching up with everybody, and starting senior year. I'm also looking forward to not having a train roaring by several times a night...oh wait, the train in Northfield is just as loud. Seriously, though, I am sad about leaving this place and all the fun people. It's been a great summer and I'm glad I chose to come down here. [top]