July 11: Moving on for real!?
Today I woke up and had dorm breakfast of natto and siu mai (with rice and miso soup) along with something that had harusame and pork in it... Hmm... ;D I didn't get up early because I had enormous trouble getting to sleep last night! D:< I hate trying to go to sleep at a reasonable time and being unable to actually do so until an unreasonable time, because I can't help but think about how I could have used that time to browse TeamLiqui-- I mean "study"... <_< >_> I am going to try my luck again tonight; here's hoping I meet with more success haha.
I got to work on time and worked on my presentation for a bit. I am happy that I feel like I am starting to understand the graphical representations of models and the "generative processes" that Kabutoya-san gives me. I finished up my slides and let Kabutoya-san take a look, and his only complaints were stylistic ones, so that's cool. I then received a couple of papers from him, one on LDA (Latent Dirichlet Allocation) and one on how to actually do Collapsed Gibbs Sampling. The two go together well, because Latent Dirichlet Allocation requires a method for estimating the parameters that Collapsed Gibbs Sampling fulfills better than the "Variational EM" proposed in the LDA paper, according to Kabutoya-san.
Reading papers is starting to get a liiiiittle bit easier for me, but my reading speed is still disgustingly slow and my retention is equally dismal. I will likely have to read the papers two or three more times in order to actually glean the important content from them. D: I hope that I get better at reading papers soon, as I believe reading many more are is my immediate future.
At lunch I ate mapo-tofu, which was pretty darn good for cafeteria food and also vair vair cheap! I talked about dieting, weight loss and body image with Nishida-san while eating, and Kurauchi-san told me that they have some kind of weight-loss club in our lab, where you don't face any repercussions from missing your goals, and the best dieter or whatever gets treated to yakiniku by the other people. Sounds counterproductive... >_> I tried to play the "are you trying to tell me something?" joke, but it wasn't understood QQ.
After eating I played handiball with my labmates, and I was again actually alright despite the 4-on-4. Apparently my handiball team has a game coming up soon, and there is a company website that will actually tell me about these things, but no one will give me the URL... I can read dates, and I'm pretty sure my coworkers know this, but they seem to like taking care of me. >_< I will have to put extra effort into learning Japanese so they'll stop!
After lunch was spent reading papers, moving desks, editing slides and running additional tests on my models to make more data to include in my presentation. All the references Kabutoya-san gave me to model my presentation after had a "Future Work" section, so I put one in and included "-examine impact of varying number of latent variables in PLSA and Mixture Models". Kabutoya-san saw this when he was reviewing my slides and suggested I just do that right now haha. I don't mind at all, because doing that is really, really easy (famous last words, I'm going to go in tomorrow morning and something will have gone wrong with my tests), and my "Results" section looked pretty anemic, with a single table with 7 pieces of data in it. >_< So, now I'm going to have 20 more pieces of data that will dwarf the amount of information I have about the original focus of my presentation. This is what real research is like, isn't it? ;D
I stayed a little late at work because Kabutoya-san left at 6:30 and told me to leave then as well. I had to finish setting up my tests though, and so I ended up staying a little while longer. I don't regret it at all though, becuase when I was leaving the building I got an absolutely GORGEOUS view of what could only have been Mt. Fuji in the sunset. Everything looked like a watercolour; all washed out and in soft pinks and blues. I tried to take a photo with my camera phone, but that didn't work out so well. I am so frustrated that I can't take my real camera up to the office! It's honestly really, really beautiful up there, and I wish I could show all of you properly. D:
I walked home and had dorm supper of saucy pork strips on cabbage with zaru-soba-esque chilled noodles with broth in a miso soup bowl with puffy tofu slices. I absolutely love the zaru-soba-that-may-not-be-zaru-soba. It's just so perfect in the muggy heat that pervades my existence. On an unrelated note: writing this blog is great because it gives me an opportunity to review my own vocabulary haha. A few years ago I realized I couldn't actually define the word "ardent", despite being able to use it and understanding its meaning when in context. Now, when writing my blog, from time to time I realize that I'm using a word that I would not actually be able to describe the meaning of concisely to someone, so I Wiktionary it haha. Today I was having trouble deciding whether to use "pervade" or "permeate", and I realized that I could not actually tell you the difference, so I looked it up. Turns out "to pervade" is part of the definition Wiktionary provides for "permeate", so I guess I'm glad I couldn't easily think of a difference!
I decided against going to Yokochuu today again because there is apparently another party of some kind tomorrow. Usually company parties end up with me at Yokochuu, so I will see if I can't hit the Seiya after the party. I have my doubts, as often the parties end past 9, but I am going to leave open the opportunity of lucking out and saving 490円 in transit fare haha. Also, in not going to Yokochuu tonight I am making it easier to get to bed early in an attempt to run tomorrow morning.
On that note, I think I will trim my beard and call it a day. Night!
No comments:
Post a Comment