5 – Blue Hawaii: A Natural Selection

Publisher Summary Computer processing speeds had doubled about twenty times since the time the evolutionary programming was first proposed. A typical desktop PC finally offered sufficient computing power to apply the evolutionary approach to solving problems that included predicting financial markets, scheduling for complex factories, and searching for new pharmaceuticals to fight HIV. The evolutionary program evaluated each neural network in terms of classification of malignancies as per the input features. The best five hundred neural networks at each generation were saved and used to create five hundred new neural networks through random mutation, whereby “offspring” neural networks were created from each “parent” neural network by varying every one of the parent's connection weights at random.