In: Biology
7. Show how a GA (Genetic Algorithms) can be used to train a FFNN (Feedforward Neural Networks).
An algorithm is a procedure or formula for problem solving.Genetic Algorithms (GA) can be used for optimization and search problems by relying on bio-inspired operators like mutation, crossover and selection. GA shows the process of natural selection where the fittest individuals are selected for reproduction in order to produce offspring of the next generation. A feed forward neural network is a classified algorithm. It is made up of simple neuron like processing units, organized in layers. Every unit in a layer is connected with all the units in the previous layer. In order to get the expected outputs, neural network has been trained by giving an input. The Genetic Algorithm and FFNN are the important techniques presently used by computer scientists and engineers. It's been found that hybridization of these techniques can be applied to solve complex and challenging problems. They are also considered as important tools for machine learning. GA could be applied for the selection of feature subsets as input variables for back-propagation NNs, to simplify the topology of back-propagation NNs, and to minimize the time taken for learning. The performance of GA depends on the size of population, selection of parents, crossover rate, mutatuon rate, and the number of generations. Finding of suitable GA parameter values is a difficult task. NN optimization is provided through the GA by its advantage of their (NN and GA) strength and eliminating their limitations. It has been experimentally proved that the optimization of NNs by GA converges to a superior optimum solution in less computational time than conventional NNs. Therefore, optimizing NNs using GA is beneficial because the shortcomings attributed to NN design will be eliminated by making it more effective than using NNs of their own.