Jun 28: Pre-Nomikai Training
I got up feeling great today because I went to bed so early last night, and I had the lovely experience of not being sleepy all day (except for when I was reading a paper, then I could barely keep my eyes open). I had dorm breakfast of small fishes in chopped vegetables, and some oden-esque tofu with bamboo shoots and thinly sliced meat (with rice and miso soup as always). I had a relatively successful conversation with Oohashi-san, where she apologized for forgetting to turn on the tea pump yesterday, and where I explained that I was not going to be there for dinner due to "sayonara-party", to which she looked like I had said that someone had gotten hit by a car (which maybe sayonara-party implies? Could just be Oohashi-san), so I explained, seemingly successfully, that the party was for people who are switching labs. I then said gochiso-sama and left for work. My timing was dangerously perfect, as I barely waited at the stop for 30 seconds before the bus came along. If I'm not more careful in the future, I could end up on time for work. >_
At work I reported my finished classifiers to Kabutoya-san and received the seminal paper on Probabilistic Latent Semantic Analysis on return. He didn't even check my work this time, so I guess I'm making a good enough impression. He was also quite busy today, and so that may have something to do with it (I am choosing to ignore this fact). I am probably not helping his workload with my TEAM FAST behaviour, but I can't really help it...
At lunch we didn't pay handiball, because there were important presentations to be done by the employees after lunch, and they didn't want to be sweaty and gross (or late). So, I went back to my desk and worked for a little while longer. Also going on throughout the day was me practicing my introduction that I would be doing at tonight's farewell party for departing project members. This one was to be in Japanese, so I definitely had to spend some time memorizing it. Kabutoya-san had told me yesterday that he would prepare a "japanese joke" for me to put into the script, which I was a bit apprehensive about, but what can you do? Today I found out that the "japanese joke" was, ahem, well suited to my talents. In other words, after my actual introduction, I was to read some lines that apparently massacre a common saying along the lines of "when in Rome, do as the Romans do" and turns it into something like "when in Rome, I am Buddha" in as horrible a gaijin accent as I could (especially the line that says "MII WA KONNA WONDERFUL TOKORO DE HATARAKETE TOTTEMO HAPPY DEEEEEESU). >_> I wasn't quite so sure about "japanese joke" anymore, but I realized that if I offended anyone, it wouldn't be my supervisor so it was pretty safe to go ahead with, if a little disconcerting...
During the afternoon Kabutoya-san also gave me a lecture (like a miniature university style one, complete with whiteboard) on Maximum Likelihood estimation, since he could see I was struggling with the PLSA paper, due to me not knowing about a lot of the methods addressed, such as ML estimation, Maximum a Priori Estimation and Expectation-Maximization. He also gave me "homework" to do at my desk, which consisted of deriving the formulae commonly used to for the mean and variance as maximum likelihood estimates. This went swimmingly until I couldn't get rid of a mysterious -2 attached to my result for the variance. It turned out that either he had written the formula for the Gaussian distribution wrong or I had copied it down wrong, as I found a lovely -1/2 waiting for me as I opened the Wikipedia article for it. XD
After finishing my homework, he showed off what I think is one of the most creative and clever circumventions of the language barrier I have ever seen. I showed him my work and he was like "good, good", and then he circled the results I got (the mean and variance) and the formula for the normal distribution at the top of the page, and told me "these are the only things you need to implement this, the parameters and the equation, and also you can understand implementing it with only these. Just like PLSA." It was so cool. Presumably without knowing how to actually express it, he managed to tell me that I needn't concern myself with the details of the derivation of PLSA as long as I understood the underlying probabilistic/statistical model and the basic formulation of the parameters. I will of course continue to worry about the details, but I won't stress over them and I think that I will be able to let go a little more now and let myself learn the details as I get a better pool of background knowledge. The conciseness of his explanation and the minimal use of complicated language made me go "WHOA", and helped me understand why he seems to have 40,000 more meetings to go to than I see the others in our group have, at the tender age of 28.
After that I was left alone to finish reading the PLSA paper and practice my introduction. I wasn't explicitly given time for the introduction, but I finished the paper and everyone in the lab was doing their research proposal presentations in another room at the moment, so I was left with nothing else constructive to do after choosing to read up on Expectation-Maximization on Wikipedia.
At around 6:15 it was time to go to the farewell party. It was in Kurihama, which meant that we took a bus to the YRP Nobi Station and then took the Keikyu line to Kurihama. The party was held at a sushi restaurant that Kabutoya-san told me was famous for its maguro, so I was very excited, considering that Yokosuka itself is already famous for seafood. At the restaurant we were seated, and luckily for the feeling in my legs and my general comfort, western style sitting was adopted by about half the people in the room.
Before the food came I asked Kabutoya-san about the acceptability of beards at the company, since no-one in the entire room of like 40+ people had even a hint of one. He thought I was saying "beer", and said "yes it's ok if you drink". It turned out that he wasn't just mishearing me, but that he simply didn't know the word "beard". I proceeded to teach him about the words "beard", "goatee", "sideburns" and "chinstrap" (the important ones ;D), complete with diagrams drawn on the back of my introduction script. I find it pretty funny that I am able to make Kabutoya-san laugh with jokes like "I am unable to find a solution to the problem of maximizing my work done while minimizing your Expectations", but we have miscommunications about "beer" and "beards". Just goes to show you that you get to know what is relevant to you. Earlier today Nishida-san (one of my friendliest and English-proficient group members) told me that he can read academic papers in English, but can't deal with Harry Potter (books or movies).
The beer came (Kirin), and then some food (Chawan-mushi and tempura, as well as a tiny abalone and a piece of lemon jelly, all very good). This was apparently "practice start" for dinner, and someone did a short toast which I didn't understand except for "kampai!" (cheers). The two groups to left and right of me proceeded to force upon me sake and shochu, respectively, although the sake was 14% alcohol and was very sweet and weak compared to my memory of my limited experience of sake in Canada. The sake camp served me a very reasonable amount, but the shochu camp (of which Kabutoya-san was involved) served me more than half a highball glass full of the stuff (although there was a good deal of ice in it). They were saying what sounded like the equivalent of "oh shit it's way too full who poured that" and stuff and proceeded to give it to me anyways, with a smile of course. I'M ON TO YOU GUYS!!! I was commenting on the fact that I was being attacked from both sides, to which they laughed and said "yes, mixing different alcohol is not so good". I then told them plaintively "that's what you're doing to me! I've practically got one in each hand!" to which they laughed, but kept doing it. >_> Luckily my shochu never got refilled, but my beer and sake did once each.
When the full dinner was ready to come, we had speeches for the departing project members, preceded by me as the opening act. I froze a little in the middle of my introduction, and thus had to consult my script, but the lab members were very supportive and told me all kinds of nice things that weren't true about my ability to speak Japanese. For my "Japanese joke" I decided to go for broke and lowered my voice about 3 octaves (OK OK fine it was more like 8 semitones) and did my best "foreigner fails horribly at Japanese pronounciation" impression, which was ego-crushingly easy. D: All of the people I could see were laughing quite hard once they figured out what was happening, so I guess it wasn't so bad. As long as they like me and continue being as nice to me as they have been so far, I'm OK with looking like a complete tool once in a while. >_>
After the dinner, there was the option to go to a second party, and although I was feeling barely anything (probably due to a good amount of eating) despite my sake, beer and Big Gulp (tm) shochu, I have another party tomorrow night, so I bowed out along with Kabutoya-san who also had a party yesterday along with the one tomorrow. >_< I had had enough adventures for today anyways, with "my" attempt at "humour" and my big gulp shochu and eating something that wasn't very good that Kabutoya-san said "maybe is whale" (he didn't like it either). I had a pleasant ride home on the train and bus, but had decidedly unpleasant beer+raw fish (and maybe whale) burps. D:< EW. Once I got home I ate some senbei to prevent more of those.
I now must go to bed, as this blog has taken far longer to write than I expected, and now I will be tired tomorrow QQ.
Good Night!
No comments:
Post a Comment