Genetic algorithm experiment

Lindberg N-O, Colbourn EA. Use of artificial neural networks and genetic algorithms—experiences from a tablet formulation. Pharm Tech Eur 2004 16(5) 35-9. 

Simulated Annealing-based solutions [19] are conceptually the same as Genetic Algorithm-based approaches. However, the SA-based techniques, in our experience, are more sensitive to the initial settings of the parameters. Nevertheless, once the correct ones are found, the method can achieve the efficiency of GA-based solutions. We must point out that SA-based solutions have never outperformed the GA-based ones in our studies. Much of what has been mentioned regarding the GA-based solutions is also relevant for the SA technique, particularly, with respect to the cost functions. 

A couple of more recent texts provide gentle introductions to the subject. Evolutionary Computation by De Jong12 looks beyond the basic genetic algorithm, but does so at a studied pace, thus is suitable for those who do not yet have much experience in the field. Finally, Swarm Intelligence, by Kennedy and... 

Two optimization tools can be used for "virtual" catalytic experiments (i) HRS and Genetic Algorithm (GA). We have recently demonstrated [28] that HRS is a faster optimization tool than the GA. The only advantage of GA with respect to HRS is that GA uses a continuous experimental space, while HRS makes use of levels. 

This strategy of integrating neural networks with genetic algorithms has been used to search for the optimal composition of a catalyst for the ammoxidation of propane [62]. In that case, no experiments were performed the network was trained with data published earlier by other authors [63]. However, those data were for only 26 catalysts, thus forming a quite small training set. Even more importantly, the predicted performance of the optimal catalyst, expressed by means of acrylonitrile yield, was not experimentally verified. 

Although in principle a genetic algorithm, or other learning algorithm, shoutd f find the true optimum, the search is limited, either by computer limitations in th i case of numerical studies, or by experimental restrictions in the case of laboratory ir experiments. ij]... 

This method enables prediction of the quahty of a separation on the basis of a relatively hmited number of the experimental data, collected in previous experiments. According to this approach, the chromatographic results are interpreted in terms of the retention functions, valid for each individual solute separately. Some good examples of the interpretative strategy are the so-called window diagrams approach [20] and the search for the extremum of the multiparameter response function with the aid of the genetical algorithm [21],... 

---