Towards Utilization of Neurofuzzy Systems for Taxonomic Identification Using Psittacines as a Case Study

Table 1

Round 1. Data not separated by gender for any species; 2,820 patterns presented for use in training; NEFCLASS-J defaults for learning constraints and learning rate were kept; maximum number of epochs was 500, minimum number of epochs was 10, and number of epochs after optimum was 25.

MF shape	Aggregation function	Learning procedure	Correctly classified %	Num. rules	Mean error
Triangular	Maximum	Best	96.77	1,795	0.570922
Triangular	Maximum	Best per class	96.1	1,776	0.566667
Triangular	Weighted sum	Best	96.77	1,795	0.571277
Triangular	Weighted sum	Best per class	96.17	1,777	0.567021
Trapezoidal	Maximum	Best	97.84	1,931	0.603191
Trapezoidal	Maximum	Best per class	96.13	1,777	0.573759
Trapezoidal	Weighted sum	Best	97.84	1,931	0.610284
Trapezoidal	Weighted sum	Best per class	97.02	1,910	0.612411
Bell-shaped	Maximum	Best	0.71	1	0.992908
Bell-shaped	Maximum	Best per class	0.71	1	0.992908
Bell-shaped	Weighted sum	Best	0.71	1	0.992908
Bell-shaped	Weighted sum	Best per class	0.71	1	0.992908