Summer School started
Jul. 9th, 2007 10:59 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
guslThe Summer School started today, bright and early. I arrived at 8:45am, and felt like I was late. It was a long day, full of free food (except for lunch).
Saw Vicente, Kai-Min, Mark Albert. Ran into John Hale, with whom I had lunch. He praised my taste in Summer Schools (all 3 Summer Schools I've ever done have been attended either by him or by a student of his).
There are tons of people here! I was expecting 50-70, but I estimate there are just over 150. All the speakers need to use a microphone.
The size of the group lead to a rather cliquish dynamic, unfortunately. There are several German-speaking cliques. I've hung out with a clique from Brown and the one from Washington@StLouis.
It was a long day... All the lectures were introductory, so I didn't get very much out of them. I skipped about half of the last lecture, on information theory.
It seems that the faculty is very well versed in machine learning, while the students have very mixed backgrounds.
During a break, met a math student from Berkeley, who claims he has trained his ability to visualize things in 4D, by using time.
At dinner, I met this guy Klaus, from Caltech, who offered to fMRI me and give me a digital image of my brain (and give me $40!). I want to do it, although I hear it can be boring and the noise is unpleasant.
At the evening session of probability, I made the point that requiring people to follow the axioms of probability amounts to requiring that they solve NP-Hard problems. After that, I chatted briefly with Josh Tenenbaum, showed him the Turney&Littman paper on analogy, and told him about ICA. I wasn't able to remember how the ICA guys reformulated the source separation problem.
Then I spoke briefly to Brian, who is a philosopher.
---
Highlights of today:
* how do people solve the problem of induction?
* integrating top-level (knowledge of grammar) and bottom-level constraints (observed data): vision, language, causal reasoning
* demo of a vision system that extracts high-level features
---
Things I should be studying:
plate notation (CMU folks: check out the first hit!)
Conjugate prior
Maximum a posteriori
Beta prior
MCMC
---
Nick Chater (a mathematically-sophisticated psychologist) has published with Gruenwald and Vitanyi!
---
My sleep debt has been huge all of today, and it's showing.
And given the schedule, it's looking like I'll be arriving at 9am, everyday... let's see how far I can go.
Saw Vicente, Kai-Min, Mark Albert. Ran into John Hale, with whom I had lunch. He praised my taste in Summer Schools (all 3 Summer Schools I've ever done have been attended either by him or by a student of his).
There are tons of people here! I was expecting 50-70, but I estimate there are just over 150. All the speakers need to use a microphone.
The size of the group lead to a rather cliquish dynamic, unfortunately. There are several German-speaking cliques. I've hung out with a clique from Brown and the one from Washington@StLouis.
It was a long day... All the lectures were introductory, so I didn't get very much out of them. I skipped about half of the last lecture, on information theory.
It seems that the faculty is very well versed in machine learning, while the students have very mixed backgrounds.
During a break, met a math student from Berkeley, who claims he has trained his ability to visualize things in 4D, by using time.
At dinner, I met this guy Klaus, from Caltech, who offered to fMRI me and give me a digital image of my brain (and give me $40!). I want to do it, although I hear it can be boring and the noise is unpleasant.
At the evening session of probability, I made the point that requiring people to follow the axioms of probability amounts to requiring that they solve NP-Hard problems. After that, I chatted briefly with Josh Tenenbaum, showed him the Turney&Littman paper on analogy, and told him about ICA. I wasn't able to remember how the ICA guys reformulated the source separation problem.
Then I spoke briefly to Brian, who is a philosopher.
---
Highlights of today:
* how do people solve the problem of induction?
* integrating top-level (knowledge of grammar) and bottom-level constraints (observed data): vision, language, causal reasoning
* demo of a vision system that extracts high-level features
---
Things I should be studying:
plate notation (CMU folks: check out the first hit!)
Conjugate prior
Maximum a posteriori
Beta prior
MCMC
---
Nick Chater (a mathematically-sophisticated psychologist) has published with Gruenwald and Vitanyi!
---
My sleep debt has been huge all of today, and it's showing.
And given the schedule, it's looking like I'll be arriving at 9am, everyday... let's see how far I can go.