Made a couple more animations for Auger. Got lots of work done this week, including using a suggestion of Gunnar's to speed his code up 18%. He was happy but disappointed that it wasn't more. I'll have to go look back at it to see if there was something I can speed up.

Lots of people around here seem to be aiming for SIGGRAPH '06 papers. It seems strange to me to be aiming for conferences months in advance. Speaking of which, I submitted the paper on dialog act classification to ASRU after SigDIAL rejected it (the reviewers gave it a 11, with the more knowledgeable ones scoring it higher, but it needed a 12 to go through). I took the reviewers' comments into account in writing the new version, and used some of the ideas I had while working on Gunnar's stuff, so though ASRU is a different conference, it has a fair chance of getting in.



They're shutting down the computers tomorrow. Oh well, I'll be in Stuttgart watching War of the Worlds and the Brazil fest.

Mark reports some interesting work with the SVMs on his data - they are making better predictions than people thought was theoretically possible with the available features. Interesting.