User:TedDunning

Technologist in the areas of data mining and ecommerce.

Committer on Apache Mahout project for scalable data-mining.

Currently Chief Application Architect for MapR Technologies.

Former CTO of DeepDyve responsible for all engineering activities.

Former Chief Scientist at Veoh Networks, responsible for peer to peer protocols and intelligent recommendation systems.

Former Chief Scientist at ID Analytics, Inc responsible for leading the team that built ID's fraud detection system.

Former Chief Scientist at MUSICMATCH software responsible for music recommendation, radio and ecommerce optimization technology. (See also http://www.idanalytics.com and http://www.musicmatch.com).

Will build AI's for food and fun.

I like to find ways to use simple statistical measures to build big things.

I am also very much interested in how Bayesian inference can be used to analyze human behavior. My favorite surrogate for behavior is some sort of sequence of symbols such as a web log, or a history of what music a person has played. The techniques I use are more general than is required for these problems and work well for analyzing human text and genetic sequences. In unpublished work, they have been very useful in authorship detection.