Statistical Pattern Recognition: A Review
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.03 |
This paper presents a generalization of a recognition algorithm that is able to classify non-deterministic signals generated by a set of Stochastic Processes (SPs), the number of which may be arbitrarily chosen. This generalized recognizer exploits the nondeterministic trajectories generated by the Karhunen-Loève Transform (KLT) with no additional constraints or explicit limitations, and without the probability density function (pdf) estimation. Several experimentations were performed on SPs generated as solutions of non-linear differential equations with parameters and initial conditions being random variables. The results show a recognition rate which is close to 100%, thus demonstrating the validity of the generalized algorithm.