Ensembles for predicting structured outputs
published: Feb. 16, 2010, recorded: February 2010, views: 3362
Report a problem or upload filesIf you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
In many real-world domains, such as bioinformatics (functional genomics), text classification and image annotation, the goal is to predict a complex output. For example, in functional genomics, the goal is to predict the function of a gene, while the set of functions can be organized as tree (FunCat) or graph (GO ontology). In this talk, we present an approach for predicting structured outputs using ensembles of trees. The proposed approach is scalable to large datasets, different types of outputs and it is applicable to wide range of domains. First, we describe the types of structured outputs that we typically encounter, and then we explain the base classifiers - predictive clustering trees (PCTs). Next, we discuss the ensemble methods that we extended (bagging and random forests) to deal with structured outputs and accordingly adapted the voting schemes. Afterwards, we present experimental evaluation of the proposed approach on wide range of real-world domains. At the end, we present an application of the proposed approach in functional genomics and show that our approach is competitive with state-of-the-art approaches.
Download slides: solomon_kocev_epso_01.pdf (2.4 MB)
Download slides: solomon_kocev_epso_01.ppt (2.7 MB)
Link this pageWould you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !