Pervasive conference, tutorial on Machine Learning for Ubicomp

Bernt is presenting the tutorial on machine learning for ubicomp. Machine learning comes from artificial intelligence and I think this is needed to make ubicomp systems intelligent and more smart. How to learn to perform a task from experience? Learning has a lot to do with optimization. We can use Bayes decision theory to classify some class based on some previous class. We can also use naive Bayes classifier, we need the probability given a class, and the probability of the class. There are graphical models of Bayesian networks such as Hidden Markov Models (directed, dynamic Bayesian network).