4. The EFNN
The EFNN is a five-layer hybrid neural network with
the feature to self-organize its activation function based
on GA in the learning process.
4.1. The structure of EFNN
The inputs and output of the EFNN, as shown in Fig. 4,
represent the evaluation factors and the safety stock,
respectively. The mainly layer operations of the EFNN
consist of the following five components:
Layer 1 is the input layer. In this layer, all the evaluation
factors are fed into EFNN.