Optimisation of radial basis function classifiers using simulated annealing algorithm for cancer classification

A modified simulated annealing algorithm is developed and combined with the linear least square and gradient descent paradigms to optimise the structure of the radial basis function classifier (RBFC). The optimised RBFC is then applied to cancer classifications and compared with previous methods, such as least square support vector machine and Fisher discriminant analysis. Experimental results show that the optimised RBFC is not only parsimonious but also has better generalisation performance.