Validation of Infinite Impulse Response Multilayer Perceptron for Modelling Nuclear Dynamics

Table 3

Structures and learning algorithms of the three neural networks involved in the comparison (i.e., buffered MLP, FIR-MLP, and IIR-LRNN).

Neural architecture	Network structure	Learning algorithm
Buffered MLP	Hidden nodes: 14 (bias included)	Static backpropagation (BP)
	Input delays: 15
	Feedback delays: 0
FIR-MLP	Hidden nodes: 10 (bias included)	Temporal backpropagation (TBP)
	Hidden MA-AR: 12-0
	Output MA-AR: 12-0
IIR-LRNN	Hidden nodes: 6 (bias included)	Recursive backpropagation (RBP)
	Hidden MA-AR: 12-10
	Output MA-AR: 12-10