A novel, fast-learning, non-iterative neural network used in pattern recognition

In the author´s previous works (1990-1999), a one-layered, hard-limited perceptron can be used to classify analog pattern vectors if the latter satisfy the PLI condition. For most pattern recognition applications, this condition should be satisfied. When this condition is satisfied, then an automatic feature extraction scheme can be derived using some N-dimension Euclidean geometry theories. The scheme will automatically extract the most distinguished parts of the pattern vectors used in the training. It selects the feature vectors automatically according to the descending order of the volumes of the parallelepiped spanned by these sub-vectors. Theoretical derivation and numerical examples revealing the physical nature of this process and its effect in optimizing the robustness of this novel pattern recognition system are reported in detail. An experiment shows that the system gives the learning of 4 handwritten characters near to real time. The recognition of untrained handwritten characters is above 90% correct and the recognition is in real time