FAST toolkit for Hidden Markov Models with Featureshttp://mloss.orgUpdates and additions to FAST toolkit for Hidden Markov Models with FeaturesenWed, 29 Apr 2015 04:25:23 -0000FAST toolkit for Hidden Markov Models with Features 1.2.1<html><p>FAST, is an toolkit for adding features to Hidden Markov Models (HMM). It implements a recent variation of the Expectation-Maximization algorithm (Berg-Kirkpatrick et al, 2010) that allows to use logistic regression in unsupervised learning. </p> <p>We demonstrate FAST for predicting future student performance. Our toolkit is up to 300x faster than BNT (a Bayesian Network toolkit), and up to 25% better than conventional HMMs (with no features). </p></html>Jose Gonzalez Brenes, Yun HuangTue, 28 Apr 2015 17:48:23 -0000 regressionexpectation maximizationmaxentsequence modeling<b>Comment by Jose Gonzalez-Brenes on 2015-04-29 04:25</b><p>We have a big update of FAST coming soon in terms of documentation and generalizability. The current implementation is optimized for student model.</p> <p>We are uploading this preliminary version for now to be eligible to participate in the MLOSS workshop at ICML.</p> Jose Gonzalez-BrenesWed, 29 Apr 2015 04:25:23 -0000