Neural network modeling direct

A challenging task in material science as well as in pharmaceutical research is to custom tailor a compound s properties. George S. Hammond stated that the most fundamental and lasting objective of synthesis is not production of new compounds, but production of properties (Norris Award Lecture, 1968). The molecular structure of an organic or inorganic compound determines its properties. Nevertheless, methods for the direct prediction of a compound s properties based on its molecular structure are usually not available (Figure 8-1). Therefore, the establishment of Quantitative Structure-Property Relationships (QSPRs) and Quantitative Structure-Activity Relationships (QSARs) uses an indirect approach in order to tackle this problem. In the first step, numerical descriptors encoding information about the molecular structure are calculated for a set of compounds. Secondly, statistical and artificial neural network models are used to predict the property or activity of interest based on these descriptors or a suitable subset. 

The second main category of neural networks is the feedforward type. In this type of network, the signals go in only one direction there are no loops in the system as shown in Fig. 3. The earliest neural network models were linear feed forward. In 1972, two simultaneous articles independently proposed the same model for an associative memory, the linear associator. J. A. Anderson [17], neurophysiologist, and Teuvo Kohonen [18], an electrical engineer, were unaware of each other s work. Today, the most commonly used neural networks are nonlinear feed-forward models. 

A very simple 2-4-1 neural network architecture with two input nodes, one hidden layer with four nodes, and one output node was used in each case. The two input variables were the number of methylene groups and the temperature. Although neural networks have the ability to learn all the differences, differentials, and other calculated inputs directly from the raw data, the training time for the network can be reduced considerably if these values are provided as inputs. The predicted variable was the density of the ester. The neural network model was trained for discrete numbers of methylene groups over the entire temperature range of 300-500 K. The... 

In this approach, connectivity indices were used as the principle descriptor of the topology of the repeat unit of a polymer. The connectivity indices of various polymers were first correlated directly with the experimental data for six different physical properties. The six properties were Van der Waals volume (Vw), molar volume (V), heat capacity (Cp), solubility parameter (5), glass transition temperature Tfj, and cohesive energies ( coh) for the 45 different polymers. Available data were used to establish the dependence of these properties on the topological indices. All the experimental data for these properties were trained simultaneously in the proposed neural network model in order to develop an overall cause-effect relationship for all six properties. 

From these various analyses, it is clear that dopamine regulation of the striatum does not simply control detailed movement, but is involved in the selection and initiation of appropriate goal directed actions (Dunnett and Robbins, 1992 Robbins and Everitt, 1992), as influenced by motor learning (i.e. the acquisition of skills and habits Mishkin et al., 1984 Jog et al., 1999), in the context of motivational information related to needs and rewards (Suri and Schultz, 1999). Theoretical formulations of this process have moved away from the neuropsychological theory, although still conceptually useful, to mathematical and neural network modeling (Houk et al., 1995 Servan-Schreiber et al., 1998), which is beyond the scope of the present review. 

A multilayer perceptron (MLP) is a feed-forward artificial neural network model that maps sets of input data onto a set of suitable outputs (Patterson 1998). A MLP consists of multiple layers of nodes in a directed graph, with each layer fully connected to the next one. Except for the input nodes, each node is a neuron (or processing element) with a nonlinear activation function. MLP employs a supervised learning techruque called backpropagation for training the network. MLP is a modification of the standard linear perceptron and can differentiate data that are not linearly separable. 

Figure 12.3 Direct neural network modeling (a) and inverse neural network modeling (b) for free radical polymerization of MMA.

S. Curteanu, Direct and inverse neural network modeling in free radical polymerization. Cent. Eur. J. Chem., 2 (1), 113-140,2004. 

Lobato J, Canizares P, Rodrigo MA et al (2010) Direct and inverse neural networks modelling applied to study the influence of the gas diffusion layer properties on PBI-based PEM fuel cells. Int J Hydrogen Energy 35 7889-7897... 

Bourquin J, Schmidli H, van Hoogevest P, Leuenberger H. Pitfalls of artificial neural networks (ANN) modelling technique for data sets containing outlier measurements using a study of mixture properties of a direct compressed tablet dosage form. Eur J Pharm Sci 1998 7 17-28. 

Memo. No. 1140. Massachusetts Institute of Technology, Cambridge, MA, 1989. Psichogios, D. C., and Ungar, L. H., Direct and indirect model based control using artificial neural networks, Ind. Eng. Chem. Res. 30, 2564 (1991). 

When an accurate model of the reaction kinetics is not available (e.g., due to the lack of reliable data for identification), the previously developed approach may be ineffective and model-free strategies for the estimation of the effect of the heat released by the reaction, aq, must be adopted. In detail, the approach in [27] can be considered, where aq is estimated, together with the heat transfer coefficient, via a suitably designed nonlinear observer [24], Other model-free approaches can be adopted, e.g., those based on the adoption of universal interpolators (neural networks, polynomials) for the direct online estimation of the heat [16] and purely neural approaches [11], Approaches based on the combination of neural and model-based paradigms [2] or on tendency models [25] can be considered as well. 

A new approach is the application of chemometrics (and neural networks) in modeling [73]. This should allow identification of the parameters of influence in solvent-resistant nanofiltration, which may help in further development of equations. Development of a more systematic model for description and prediction of solute transport in nonaqueous nanofiltration, which is applicable on a wide range of membranes, solvents and solutes, is the next step to be taken. The Maxwell-Stefan approach [74] is one of the most direct methods to attain this. 

At the beginning of this chapter, we introduced statistical models based on the general principle of the Taylor function decomposition, which can be recognized as non-parametric kinetic model. Indeed, this approximation is acceptable because the parameters of the statistical models do not generally have a direct contact with the reality of a physical process. Consequently, statistical models must be included in the general class of connectionist models (models which directly connect the dependent and independent process variables based only on their numerical values). In this section we will discuss the necessary methodologies to obtain the same type of model but using artificial neural networks (ANN). This type of connectionist model has been inspired by the structure and function of animals natural neural networks. 

Apart from immediate release tablet formulations, neural networks have also been applied to modeling immediate release capsule formulations and rapidly disintegrating or dissolving tablets. In the latter, Sunada and Bi from Japan used both statistics and neural networks to optimize both the formulation and processing conditions for rapidly disintegrating tablets developed using both direct compression and wet granulation techniques. 

---