Page 198 - Introduction to Statistical Pattern Recognition

180 Introduction to Statistical Pattern Recognition

10. In the design of a piecewise linear classifier, propose a way to assign the
desired output so that we can apply the technique of minimizing the
mean-square error.

References

1. H. L. VanTrees, “Detection, Estimation, and Modulation Theory: Part
I,” Wiley, New York, 1968.
2. A. Fisher, “The Mathematical Theory of Probabilities,” Vol. 1, Macmil-
Ian, New York, 1923.
3. T. W. Anderson and R. R. Buhadur, Classification into two multivariate
normal distributions with different covariance matrices, Ann. Math. Stat.,
33, pp. 422-431, 1962.
4. D. W. Peterson and R. L. Mattson, A method of finding linear discrim-
inant functions for a class of performance criteria, Trans. IEEE Inform.
Theory, IT-12, pp. 380-387, 1966.
5. Y. C. Ho and R. L. Kashyap, An algorithm for linear inequalities and its
applications, Trans. IEEE Electronic Computers, EC- 14, pp. 683-688,
1965.
6. C. W. Therrien and K. Fukunaga, Properties of separable covariance
matrices and their associated Gaussian random processes, Trans. IEEE
Pattern Anal. and Machine Intell., PAMI-6, pp. 652-656, 1984.
7. K. Fukunaga and T. Ito, A design theory of recognition functions in
self-organizing systems, Trans. IEEE Electronic Computers, EC- 14, pp.
44-52, 1965.