Reply to: A Discussion on “Hindcasting of wave parameters using different soft computing methods” [Appl. Ocean Res. (2008), doi:10.1016/j.apor.2008.03.002

Abstract

Tarkan ERDIK provides some comments on our paper. He has some misapprehensions on our paper. The following discussions follow the order of his comments. 1. Response to the first comment Dr ERDIK is asked to reread Section 3 and 4 of the paper [1]. Several types of Fuzzy Inference System have been proposed in the literature. One of these types is the so called Takagi and Sugeno FIS [2], which has been used in the paper. In the original paper,the subtractive clusteringmethodwas introduced in the last paragraph of the Section 3 as a method to estimate the optimum number of IF–THEN rules and to determine the membership functions, not as a Fuzzy Inference System (FIS) algorithm. The discusser has claimed we introduced the ‘‘grid partitioning’’ as a FIS algorithm which is not correct. 2. Response to the second comment In the process of training an ANN for a dataset that you own no prior knowledge about, you ought to generate the initial weights and thresholds randomly, but with a bit of logic. All the synaptic weights and thresholds are picked from a uniform distribution with a mean of zero, and a variance such that the standard deviations of the induced local fields of the neurons lie at the transition between the linear and the saturated parts of