Magnus Stensmo

I was a Visiting Graduate Student/Research Assistant in Terry Sejnowski's Computational Neurobiology Laboratory at the Salk Institute in La Jolla, California between 1992 and 1996. After that I moved 730 km NNW for a postdoc at the Computer Science Division of the Department of EECS at UC Berkeley. In 1997 I moved 90 km SSE to San Jose where I am a Research Staff Member at the IBM Almaden Research Center.

Contact information

E-mail: magn us@sa l k.ed u

Research at Salk

At CNL I worked on automated medical and machine diagnosis using probability and decision theory. Mixture models and the EM algorithm were used to build a probability model from a database of cases. Temporal-differences (TD) reinforcement learning was used to find good utility values for a decision model.

In another unrelated project I worked on vigilance and attention prediction from electroencephalogram (EEG) and electrooculogram (EOG) signals using neural networks together with Tzyy-Ping Jung and Scott Makeig. We used back-propagation networks for time-series prediction.

Publications

Odds and ends, a long time ago when the web was born...

We connected an IndyCam in the office I was in for all the world to see. I'm in the back of this picture and Tzyy-Ping is close to the camera.

I kept a page for news from home (highly recommended reading). Early on I also made a web interface (now sadly broken) to the Voice of America Newswire to keep me and other people informed on what's going on in this world. It was a great source of on-line news since it was updated very often before the advent of news web sites. Nowadays I get a lot my news from BBC News.