A New Geometric -Based Clustering Algorithm for Medical Data Analysis

: Many medical data sets are mathematically structured as graphs. For example, AIDS dataset consists of graph characterizing chemical compounds. In this paper, we address the clustering of a collection of graphs, where each data point is represented as a graph. Our proposed method employs the heat kernel algorithm as a geometric approach. We present an embedding technique based on the heat kernel to transform each graph into a feature vector. Subsequently, we cluster the data set of graphs using these feature vectors. Experimental results demonstrate that the proposed method achieves ۹۸.۸۵% accuracy in clustering the AIDS dataset.

کلیدواژه ها:

Clustering ، Heat Kernel ، Graph Embedding