Hidden Markov Decision Trees

We study a time series model that can be viewed as a decision tree with Markov temporal structure. The model is intractable for exact calculations, thus we utilize variational approximations. We consider three di erent distributions for the approximation: one in which the Markov calculations are performed exactly and the layers of the decision tree are decoupled, one in which the decision tree calculations are performed exactlyand the timesteps ofthe Markov chain are decoupled, and one in which a Viterbi-like assumption is made to pick out a single most likely state sequence. We present simulation results for arti cial data and the Bach chorales. MIT COMPUTATIONAL COGNITIVE SCIENCE TECHNICAL REPORT 9605

• Type: Conference
• Year: 1996
• Where: NIPS
• Authors: Michael I. Jordan, Zoubin Ghahramani, Lawrence K. Saul
• Keywords: Decision Tree, Decision Tree Calculations, Markov Temporal Structure, NIPS 1996, NIPS 2007