Artificial neural networks (ANNs) are useful tools in fields such as controller design and forecasting. However, it is difficult to apply ANNs to specific problems without human intervention, intuition, or prior knowledge of the given problem. Previously constructed ANNs are partially dependent on human intuition to adjust the number of hidden layer neurons or to initialize neuron weights. The reason for using ANNs is that an artificial neuron can be converted into a hyperplane, and the combination of hyperplanes becomes a highly accurate classifier. The Artificial Life Ecosystem (ALE) scheme based on genetic algorithms (GAs), and can generate ANNs without the need for prior knowledge or human intervention. Problem space is divided using Center and Limitation for Locality (CeLL). From randomly placed center, to the sight range, local classification problems are solved which results in parts of neural network. These parts solving local problems are combined to form a neural network that can solve the whole problem using genetic algorithm. Macroevolution adds new layers and microevolution changes weights within the layer. Degeneration removes unnecessary neurons which does not affect the result of classification.