Comparing Model Selection Criteria for Belief Networks Export

@misc{citeulike:586528, abstract = {Learning belief network structure involves a trade-o between network complexity and goodness of t. More complex structures allow for a better t to the data, but suer from a decreased ability to generalize to unseen data. This bias-variance trade-o may be operationalized by a scoring function on structures, called a model selection criterion. We describe various criteria for model selection: a prequential, or Bayesian, criterion, 2-fold cross validation, a bootstrap criterion, Akaike's...}, author = {Van Allen, Tim and Greiner, Russ}, citeulike-article-id = {586528}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.38.6203}, keywords = {bayesian, learning, network}, posted-at = {2006-04-14 18:02:48}, priority = {5}, title = {Comparing Model Selection Criteria for Belief Networks}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.38.6203} }

TY - GEN ID - citeulike:586528 L3 - citeulike-article-id:586528 N2 - Learning belief network structure involves a trade-o between network complexity and goodness of t. More complex structures allow for a better t to the data, but suer from a decreased ability to generalize to unseen data. This bias-variance trade-o may be operationalized by a scoring function on structures, called a model selection criterion. We describe various criteria for model selection: a prequential, or Bayesian, criterion, 2-fold cross validation, a bootstrap criterion, Akaike's... TI - Comparing Model Selection Criteria for Belief Networks KW - bayesian KW - learning KW - network AU - Van Allen, Tim AU - Greiner, Russ UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.38.6203 ER -