Comments on "A new approach to fuzzy-neural system modeling" [with reply]

In the original paper (Y. Lin and G. A. Cunningham, ibid., vol. 3, p. 190-8, 1995), an approach to fuzzy-neural knowledge extraction starting from multi-input single-output examples is given. The author shows that the performance index given there is almost monotonically decreasing with m/sup -(1/2)/; that is, it is possible to obtain a very small performance index simply by increasing m. This is possible even if at the end of the learning phase the root mean square error remains very high. The original authors acknowledge that this is correct and explain why nevertheless they presented the approach that they did.