the end of the conference

yesterday was the last day of my european summer school in information retrival, and as with most endings, it was bittersweet. i had so much fun, met so many interesting people and learned so much, but i was very tired and ready to have some time to myself. we had lectures for 8 hours everyday (from california time 1am-10am) and social events every night (not that i'm complaining) so it was very energy consuming. there were lectures about all kinds of interesting things, like that 87% of americans can be uniquely identified by their gender, birthday and zip code. with this they can be matched to 'anonymized' government (health, voting, etc) records, and with a little tricky computing, self-declared online information, regardless of identity. so that means your fake profiles (those with pseudonyms, vague location data, etc) can be matched with your real ones along with your shopping habits, medical history and political preferences. the way to combat this? act normally. the more average your activities on the internet the more difficult it is to pick you out from the crowd. ie when you write reviews for movies review the popular ones, have babies on the most common day, live in a city; having specific or abnormal tastes/stats gives identifiable trends to be picked up. ha! that's a lot to think about...

so now i've lost my train...what have i been doing? hanging out with my new forgein friends (who have now all left) and sleeping. but i'm thinking you want to know about what i waaasss doing. a typical day: i get up at 8:10 to get to class late at about 9:15 (by friday i was on time though!). it's about a 10 minute walk from my flat to the lecture hall. lectures were either 1.5 or 2 hours, always with a coffee/tea break between them. lunch continued to be those strange sandwiches throughout the entire week. oh my god i never want another small, triangle, mayonnaise filled sandwich ever again (not that i wanted one in the first place). a few late additions to the menu included tiny shrimp and dill in mayonnaise, chicken in mayonnaise, egg in mayonnaise and salmon in mayonnaise. there would be big long lines around the tables of sandwiches and many, many unmarked trays to choose from. i would always get decision anxiety and end up just grabbing whatever was near me and missing the good ones (ok, better ones). but it literally took 6 or 7 to get filled up. i'm really wishing i had taken pictures of them. all the meals had orange or apple juice and flat or sparkling water, no soda. at the bars they measure out all the liquor in little tiny shot glasses and it is illegal to pour it directly into a glass or to sell more than 2 shots at a time to any person. the beer on the other hand, flows just as in the states. the tap water is pretty good too.

i'm not enjoying telling about an average schedule, it varied too much once the lectures were over. we had a soccer game one night (i cheered), a banquet another night (complete with a paper airplane war) and typically went to a pub after any event. it was really interesting because the first day i was awake in the morning lectures (1am - 4am in ca) and tired in the afternoon. makes sense, i'm a bit of a night owl. the second day i was tired in both sessions and the third day i was tired in the morning (9am - noon here) so i could tell i was getting used to the time finally; tired in the morning like a normal person. regardless i was always awake around social hour, given the activity combined with the time, it was either early afternoon or early evening, both easy to be awake in.

rhiannon came home at 5am on thursday, so i had to kind of grapple with the issue of her wanting me to hang out with her friends and me wanting to hang out with my nerdy conference friends. the first night i just ditched her, the second i tried to mix the two groups, which went surprisingly badly. not like there was any lasting damage to any parties, but mixing 30ish computer scientists (cs) with 21 year old british college girls somehow didn't work as i predicted. the girls wanted to go home and the cs people wanted to go out (the bars closed at midnight). the girls were offended by our complaints about the early closings and comparisons with our home countries. it really convinced me that the uk is pretty much just like home. i mean that i always think places are just like home but here even more so. people are so defensive. and there's a lot of money, which i think makes people somewhat less interesting or motivated or exciting or something.

many of the lectures have been about making distinctions between things: classifiying and comparing documents, web pages, images, music (which was why i was there!). humans are very good at this, but very bad at explaining it. we can take two things and tell you which is which, which belongs in which group or not. computers can be trained to do this fairly easily, but when they make mistakes they are very, very obvious. human mistakes on the contrary, can always be justified in hindsight. (haha, i mean in the context of classification...perhaps in any context? that is debatable. i will amend my earlier statement with nearly always justified.) if i am given a beatles song and asked to supply its genre, i give rock n roll. sure. say we have two systems which we ask the same to. one gives a response of classic american pop and the other reggae. now i tell you the human gave the first answer and a computer the second. we say the first answer is a 'less bad' mistake than the second because its close to the right answer since the beatles are classic and they are pop and famous in the us. reggae on the other hand has nothing to do with the beatles. the problem is there is an endless number of justifiable mistakes which we can't come up with before the fact, but we know how to explain it when we see it. ahh the ultimate computer. so what does this have to do with anything? i'm not sure but it was in my head during the culture clash last night.

how is it some people can into all different kinds of situations, environments and social atmospheres? how is that some people are very bad or very good at it? what's the difference (similarity, haha) between seeing differences and seeing similarities? when you contrast rather than compare, i mean that you look for things which are different rather than similar, and you find them, you feel separate and alone. if you pay attention to the similarities you can fit in and feel comfortable. differences are more critical to us than similarities - not more important but more likely to cause problems (or misclassifications to go back to my original topic). it might not be especially beneficial to the human, but i'm starting to think this is what the computer needs to classify and make those human-like mistakes. if my beatles question had been given to a computer system which made decisions based on differences, without any information of similarities to use at all, it sure couldn't have come up with reggae. unfortunately, training these systems is somewhat of a black box - you can't decide how the computer makes its decisions unless you literally give it an answer for every available option - but you can decide what kind of data it uses to make its decisions.

hmmm exciting blog post? sorry, i haven't been doing much exciting, but there's a piece of my mind either way.



(post script: that was the wrong video, but it took so long to upload i'm a. not going to delete it and b. not going to upload another. see, i'm alive!)