NEURAL NETWORKS

Similarities between neural networks, GA, ANT Colony, and response surfaces



Similarities between neural networks, GA, ANT Colony, and response surfaces

Part One

Neural Network Concept Basic Idea

The thought of artificial neural systems is dependent upon the outline of the human cerebrum. The human mind is constituted by informative data transforming units (alleged neurons) that are joined by synapses, and it shapes the piece of the human anxious framework. It is fit for transforming includes indicators that are determined from nature and of giving suitable yield signs (e.g. certain activities). The focal points of the human qualified information preparing framework are many sided quality, nonlinearity, and parallelism. A counterfeit neural system takes after the human cerebrum in numerous regards. It is constituted by neurons which are joined by synapses; it has the capacity of mapping enter indicators onto yield indicates and to acclimate to certain errands throughout a preparation stage.

The yield generated by a neural system is called reaction surface. The thought of such a reaction surface in building is then to supplant the deterministic computational model for structural investigation by a neural system. That is, the info signs contain structural parameters, for example stacks, material parameters, and geometrical parameters and the system yield furnishes the partnered reaction surface in the manifestation of burdens, relocations, or distortions. For this reason, the system first ought to study the characteristics of the underlying deterministic computational model. This studying is dependent upon at first performed structural investigations (NeuralAutoTraining Expert Advisor, 2010).

Figure (i)

Source: iticsoftware.com

Genetic algorithms

Genetic algorithms (hereinafter GA) are adaptive search techniques that have recently been more and more widely used for problems of structural and functional diagnostics. They are called genetic because the use of the principle of the evolution of biological organisms Darwin. By which populations evolve over the generations, and, obeying the principle of natural selection, that is, survival of the first. Genetic algorithms are similar to the mechanism work with the population, each of the chromosomes which best possible solution to this problem. Each chromosome is evaluated measure of its fitness fitness function, which is called the optimal function. For example, a measure of fitness of the chromosomes that encode the surface of the aircraft, you can use a common factor streamlining Rx (NeuralAutoTraining Expert Advisor, 2010).



Fittest individuals are able to reproduce offspring with a crossover cross over and mutation, with other individuals of the population. The result is a specific heritage characteristic of the parents. In turn, less fit individuals gradually disappear from the population. The new generation has better performance than the previous ones. Crossover and mutation leads to the fact that the evolution of searches for innovative solutions in a wide area search. Ultimately, the population converges to an optimal solution but, none of genetic algorithm does not guarantee this.

Figure (ii)

Source: www.klopfenstein.net

The genetic algorithm (GA) is a computer model of the evolution of artificial population of “individuals.” Each individual is characterized by its chromosome Sk, the chromosome is the “genome” of an ...