Tree-Like Structure Classification Based on Distance Matrix LU Decomposition with Application to Galaxy Profile Data

• Activity Number: 131
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Jianan Hui* and Xinping Cui and James Flegal and Miguel Aragon-Calvo
• Companies: University of California at Riverside and University of California at Riverside and University of California at Riverside and University of California at Riverside
• Keywords: Tree-like structure ; Tree Encoding ; Feature Extraction ; Classification ; Support Vector Machine ; Galaxy Data

Abstract

With the development of highly efficient graph data collection technology in many application fields, classification of graph data emerges as an important topic in the data mining and machine learning community. In this paper, we propose a multi-step approach for representing and classifying tree-like structures from computer simulations of galaxy formation along with galaxy intrinsic properties. This is accomplished by first, embedding the tree-like structure and the galaxy profile data into feature vectors; second, performing the LU decomposition of the distance matrix computed by the feature vector and then using the output of the decomposition as input vectors of support vector machine and k-nearest-neighbors classifiers. Our method demonstrates a significant improvement of prediction as compared to the state-of-the-art methods. Developing such automated tools for the classification of tree-like structures with galaxy profile data can potentially provide insight to the formation process of galaxies.