Quotient canonical feature map competitive learning neural network

We present a new learning method called the quotient canonical feature map for competitive learning neural networks. The previous neural network learning algorithms did not consider their topological properties and thus, the dynamics was not clearly defined. We show that the weight vectors obtained by competitive learning decompose the input vector space and map it to the quotient space X/R. In addition, we define ε, the quotient function which maps [1,∝]±Rⁿ) to (0,1), and induce the proposed algorithm from the performance measure with the quotient function. Experimental results for pattern recognition of remote sensing data indicate the superiority of the proposed algorithm in comparision to conventional competitive learning methods