John Platt, a distinguished scientist at Microsoft Research, uses the Machine Learning Blog to provide an overview of more than 20 years creating machine learning systems at Microsoft and applying them to real problems.
It’s a story that begins 1992 with Bayesian networks, language modeling and speech recognition. In that decade, researchers found that many problems, including text categorization and email prioritization, could be solved through a combination of linear classification and Bayes networks. That work produced the first content-based spam detector and a number of other prototypes and products.
Microsoft News Center Staff