Neural model of Cry1A(c) protein produced from a B.t. fed batch fermentation

This paper presents a recurrent trainable neural network (RTNN) model of the fed-batch cultivation of Bacillus thuringiensis variety kurstaki HD-73 (B.t.). The proposed RTNN model has eleven inputs, six outputs and seventeen neurons in the hidden layer, with global and local feedback. The learning algorithm is a modified version of the backpropagation through time. The approximation error is below 2% and the generalization error is below 3%. The learning is performed in 51 epochs, 145 iterations each one. For the RTNN generalization, experimental data of one fermentation, not included in the learning process, are used