Deep Feature Learning Based Clustering with Application to TCM Data Analysis

Clustering analysis is an important unsupervised learning method for Traditional Chinese Medicine (TCM) data analysis. Due to the variance and heterogeneity of TCM data, features extracted directly from the input space may lead to poor clustering results. With the success of deep learning study and applications, we propose a feature learning framework based on deep network models to improve the quality of the learned features. The proposed framework incorporates deep feature learning and clustering leading to a global optimization solution. The proposed method is applied to two TCM clinical datasets and the results indicate that the proposed method is superior to the current state-of-the-art methods.