A self-organizing neural net with three functions related to principal component analysis

The most remarkable property of the authors' model is that the three mathematical functions related to principal component analysis (principal component analyzer, orthogonal projection operation, and novelty filter) are acquired by the same network but with different learning parameters.<>