Finding the overlap integral between two classes of object
作者：
Highlights：
•
摘要
Probability densities which are described by hyperellipsoids are transformed into hyperspheres. The hypersphere is topologically equivalent to the hyperellipsoid, hence there is a transformation. The consequences for the design of neural networks is discussed.
论文关键词：Pattern recognition,Neural network,Eigenvalue,Factor analysis
论文评审过程：Received 17 September 1991, Revised 27 May 1992, Accepted 5 January 1993, Available online 19 May 2003.
论文官网地址：https://doi.org/10.1016/0031-3203(93)90011-K
