Methods and Experiments With Bounded Tree-width Markov Networks
dc.date.accessioned | 2005-12-22T02:19:52Z | |
dc.date.accessioned | 2018-11-24T10:24:20Z | |
dc.date.available | 2005-12-22T02:19:52Z | |
dc.date.available | 2018-11-24T10:24:20Z | |
dc.date.issued | 2004-12-30 | |
dc.identifier.uri | http://hdl.handle.net/1721.1/30511 | |
dc.identifier.uri | http://repository.aust.edu.ng/xmlui/handle/1721.1/30511 | |
dc.description.abstract | Markov trees generalize naturally to bounded tree-width Markov networks, onwhich exact computations can still be done efficiently. However, learning themaximum likelihood Markov network with tree-width greater than 1 is NP-hard, sowe discuss a few algorithms for approximating the optimal Markov network. Wepresent a set of methods for training a density estimator. Each method isspecified by three arguments: tree-width, model scoring metric (maximumlikelihood or minimum description length), and model representation (using onejoint distribution or several class-conditional distributions). On thesemethods, we give empirical results on density estimation and classificationtasks and explore the implications of these arguments. | |
dc.format.extent | 10 p. | |
dc.format.extent | 10714507 bytes | |
dc.format.extent | 473643 bytes | |
dc.language.iso | en_US | |
dc.subject | AI | |
dc.subject | tree-width | |
dc.subject | hypertrees | |
dc.subject | Markov Networks | |
dc.subject | maximum likelihood | |
dc.subject | MDL | |
dc.title | Methods and Experiments With Bounded Tree-width Markov Networks |
Files in this item
Files | Size | Format | View |
---|---|---|---|
MIT-CSAIL-TR-2004-081.pdf | 473.6Kb | application/pdf | View/ |
MIT-CSAIL-TR-2004-081.ps | 10.71Mb | application/postscript | View/ |