A Neural Learning Rule for CCA Approximation
A new learning rule is implemented for approximating canonical correlation analysis(CCA) with artificial neural networks. A correlation objective function is maximized in order to find identical or correlated item from several sets of data. A simple weight update rule is derived, that is computationally much more inexpensive than the standard statistical technique. We demonstrate the network capabilities on artificial and real-world data. The experimental results show that this method is a good approximator of CCA as well as correlated item identifier.
Unable to display preview. Download preview PDF.