The Relationship of Sample Size and Accuracy in Radial Basis Function Networks

Even though radial basis function networks are known to have good prediction accuracy in several domains, it is not known to decide a proper sample size like other data mining algorithms, so the task of deciding proper sample sizes for the networks tends to be arbitrary. As the size of samples grows, the improvement in error rates becomes better slowly. But we cannot use larger and larger samples, because we have limited training examples, and there is some fluctuation in accuracy depending on the sample sizes. This paper suggests a progressive resampling technique to cope with the fluction of prediction accuracy values for better radial basis function networks. The suggestion is proved by experiments with promising results. Key-Words: neural networks, radial basis function networks, sampling