Skip to main content

Interspeech 2012, day 0: tutorial on domain adaptation for machine learning

research
Uncategorized
interspeech2012
Author

Abe Kazemzadeh

Published

September 9, 2012

I started the interspeech conference with a tutorial on domain adaptation for machine learning. The main theme of the 3 hour tutorial was to present machine learning and speech processing techniques in the same unified taxonomy, which was 3 categories of methods: instance-based methods, feature transformation-based methods, and model-based methods. Fei Sha (USC) and Brian Kingsbury (IBM) were the two presenters. One of the strengths of this tutorial was the comprehensive references given, without a self-promotion bias. It was cool to see my former labmate Abhinav Sethy and classmate Hal Daumé get prominent citations. One thing I’ll be looking into more is the Blitzer reference about domain adaptation for sentiment analysis. The slides are online at https://rainflower.usc.edu/wiki/projects/is2012tutorial/Interspeech_2012_Tutorial.html


Originally published on abewrites.

clustrmaps.html