Hierarchical fast learning artificial neural network

The hierarchical fast learning artificial neural network (HieFLANN) is proposed as an unsupervised learning model that incorporates a hierarchical approach to address pattern classification for high dimensional data. It utilizes k-means fast learning artificial neural network (KFLANN) subnets and a canonical covariance feature compression (C2FeCom) process. The embedded individual KFLANN subnet autonomously derives the essential localized network parameters from the input data and in the process, builds a hierarchical network. The C2FeCom feature compression process extracts the independent parameters in compact representations from subnets. The proposed algorithm is experimentally evaluated using benchmark datasets.