Weather: cloudy, rainy (21 °C) · Lunch: spicy fried chicken · Dinner: — · Sleep: phew, this one was great
So, I finally didn’t go to bed at 4 a.m. It’s starting to come together. It’s definitely great, because in Japan life really begins in the morning, and it’s worth getting up.
So I ran one of the projects again, and meanwhile I’m placing ever more emphasis on the work I was given here. I know, the double life, and you mustn’t forget home — but I think I got very good work and a good opportunity with this.
There really are a lot of results already. Professor Kato’s insights were very important in many respects:
- How do we describe nonlinear functions with HyMeKo — for example, to define a reward (reinforcement learning)?
- How do we describe multiplication? (Now, this is a tough nut: how do we define multiplication on hypergraph edges?)
- What about fully collaborative scenarios? (Fair point.)
- What if we used the new network and the old MLP network together? (A really good idea.)
So yes, I have to say: I did very well with Professor Kato. He keeps my brain spinning constantly.
Lunch
It’s raining and it’s 21 degrees — in contrast to the roughly 35 °C at home. So I’m off to lunch. Beforehand, though, I handed in the paper on which Professor Kato signed that he’s extending my stay.
Two students are at the little shop. I already know them, so we have a chat, and I ask too what good things they’re up to. And I’m hungry by now, so I bought the daily menu (chicken in a spicy coating, with rice), a new Japanese hot dog, and tea.
But then I realise something: I don’t have enough cash on me! I felt very awkward. Luckily the two students stayed, and said they’d wait until I came back with cash. Of course I dashed back. They kindly waited, and even remarked that the Japanese are fundamentally kind people. I can sense it, and I’m very glad of it!
After that we chatted a bit about student daily life. They told me they basically do research, under Professor Kato — mainly in artificial intelligence and machine learning. One of them works on emotion recognition from voice samples, the other on genetic algorithms, precisely. I even forwarded him the papers of my office-mate and pal Boldizsár, who also works on genetic algorithms. In fact, after the student translated his research material, I sent that to Boldizsár separately too.
After that, all that was left was to sit back down and keep running the benchmarks. A new idea came to me: to extend the rotation coding I’ve used so far with a spiral architecture and protruding spikes that encode and transmit the information. I hope it turns out well.
Dinner
Dinner sets off a bit late again, so I went to Saizeriya. This time I had a hamburger steak — it was good for a dinner, though it’s really the pasta and the pizza that are tasty here. In any case, after a long time I had a hot coffee. I like iced coffee too, but I’d honestly been missing hot coffee.
After that, a quick oral exam. As it happens, I’m very satisfied with my students this semester: good and creative games were made.
Photos
