Evolution of multiple tree structured patterns using soft clustering
Kengo Yoshida, Tetsuhiro Miyahara, Tetsuji Kuboyama · 2010
We propose a new genetic programming (GP) approach to extracting multiple tree structured patterns from tree structured data using soft clustering. We use a set of multiple tree structured patterns, called tag tree patterns, as a combined pattern. A structured variable in a tag tree pattern can be substituted by an arbitrary tree. A set of multiple tag tree patterns matches a tree, if at least one of the set of patterns matches the tree. Using soft clustering is appropriate because one tree structured data is allowed to match multiple tag tree patterns. By soft clustering of positive data and by running GP subprocesses on each cluster with negative data, we make a combined pattern which consists of best individuals in GP subprocesses. Experiments on some glycan data show that our method has a support of about 0.8, while the previous method for evolving single patterns has a support of about 0.5.