A Biomimetic Covering Learning Method Based on Principle of Homology Continuity

The Principle of Homology Continuity (PHC) based covering learning method is an effective to solve the pattern recognition problem. However, PHC and existence optimal coverage are not mathematical proven. To address this issue, we firstly give description theoretical proof PHC. On basis, definition introduced. Optimal can determine internal connections among samples as prior knowledge use neurons learn knowledge. Finally, propose a kind neuron model, effectiveness which demonstrated through extensive experiments conducted on CIFAR-10, LFW, YTF datasets.