Study on Ductility of Ti Aluminide Using Artificial Neural Network

Table 4

Final architecture for all datasets.

Data set	SSE	SSW	Number of neurons	Number of epochs	Training functions	Transfer function
Dataset 1	27.2377	20.602	3	143	Tansig	Trainbr
Dataset 2	20.2437	28.4964	3	114	Tansig	Trainbr
Dataset 3	0.7687	186.566	3	152	Tansig	Trainbr

SSE: squared sum of training error, SSW: squared sum of weights.