Recognition of handwritten characters using modified fuzzy hyperline segment neural network

In this paper membership function of fuzzy hyperline segment neural network (FHLSNN) proposed by U.V.Kulkarni and T.R.Sontakke is modified to maintain convexity. The modified membership function is found superior than the function defined by them, which gives relatively lower values to the patterns which are failing close to the hyperline segment (HLS) but far from two end points of HLS. The performance of modified fuzzy hyperline segment neural network (MFHLSNN) is tested with the two splits of FISHER IRIS data and is found superior than FHLSNN. The modified neural network is also found superior than the general fuzzy min-max neural network (GFMM), proposed by Bogdan Gabrys and Andrzej Bargiela, and general fuzzy hypersphere neural network (GFHSNN), proposed by U. V. Kulkarni, D.D.Doye and T. R. Sontakke.