Variants of ANN Modeling

The most common way to use ANN modeling is direct modeling, when output variables are correlated with inputs of the network. For instance, in the batch bulk free radical polymerization of methyl methacrylate, using (a,a -azobis(isobutyronitrile) as initiator, monomer conversion, polymerization degrees, and reaction viscosity are related to the working conditions (temperature, initial concentrations of reactants, and initiator and time) (Figiue 12.3a). 

The number of individual networks in a stack is a compromise between the fact that too many networks will greatly increase the computing time and too few networks will not actually benefit from the stacked generalization. The diversity of individual models is a key issue in the stacked generalization and is not easy to achieve because all models are trained to do essentially similar tasks. 

Tian et at [ 1 ] used stacked networks, each of them containing six different networks, for modeling of MMA polymerization. Individual networks of the stack have different model input structures, training data, topologies and initial weights, so that the different networks could perform differently in different operating regions and learn different parts of the problem to form the complementary stacked output. 

---

Handbook of Polymers for Pharmaceutical Technologies 12.2.3 Variants of ANN Modeling... 

---