Fast constructive-covering approach for neural networks

We propose a fast training algorithm called fast constructive-covering approach (FCCA) for neural network construction based on geometrical expansion. Parameters are updated according to the geometrical location of the training samples in the input space, and each sample in the training set is learned only once. By doing this, FCCA is able to avoid iterations and is much faster than traditional training algorithms. Given an input sequence in an arbitrary order, FCCA learns ´easy´ samples first and the ´confusing´ samples are easily learned after these ´easy´ samples. This sample reordering process is done on the fly based on geometrical concept. A comparison of this method with a few other methods on the well-known Iris data set is given.