Novel fast-learning noniterative neural network in pattern recognition
Chia-Lun John Hu · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2001
When the analog-to-digital mapping to be learned by any pattern recognition scheme satisfies a certain PLI condition, a one-layered, hard-limited perceptron (OHP) is enough to be used for recognizing any unlearned patterns with high robustness. Generally, the PLI condition is satisfied for most practical pattern recognition applications. When this condition is satisfied, then an automatic feature extraction scheme can be derived from an N-dimension geometry point of view. This automatic scheme will automatically extract the most distinguished parts of the pattern vectors used in the training. It selects the feature vectors (sub-vectors of the pattern vectors) automatically according to the descending order of the volumes of the parallelepiped spanned by these sub-vectors. Theoretical derivation revealing the physical nature of this process and its effect in optimizing the robustness of this novel pattern recognition system will be reported in detail.