Bigradient learning algorithm for dimension reduction of text document space

This paper shows text document dimension reduction and clustering technique which is called the bigradient learning algorithm. This algorithm is based on the two learning parameters. The results show, that bigradient learning algorithm, used with proper selected values, does almost the same clustering as the other arbitrary PCA learning method by neural network. At the end, the three linear PCA methods for document clustering are compared. They are: linear Hebbian neural network with Oja learning rule, and neural network with bigradient learning algorithm. The results are concluded.