Graphical User Interface Screen Shot

A Character Recognition Problem: This screen shot shows a NeuroSolutions neural network, which is most commonly referred to as a breadboard. This particular example is an optical character recognition (OCR) problem that is solved using a hybrid unsupervised/supervised network. The input to the network is a set of 24x18 images of handwritten digits. Each image has a corresponding desired output, which is an encoding of the digit that the image represents. The unsupervised portion of the network uses Sanger's learning rule to perform principal component analysis (PCA) on the images. The features extracted from this preprocessing stage are then fed into a multilayer perceptron (MLP), which uses backpropagation to perform the image classification. This network has been trained to a relatively low error, such that the network output closely matches the desired output. Note that the network classified the '3' correctly, but it also found that the image also had some characteristics of an '8' and a '5' due to their similarity in shape.

Return to the Graphical User Interface page.