Artificial Neural Networks Based on Fractal Growth

Artificial Neural Networks(ANN) has been increasingly used by researchers. An ANN model based on fractal growth is introduced in this paper. The design was inspired by the physiology of biological nervous system. Driven by data, fractal structure generates and evolves. Classification interface forms from the fractal structure, which makes the classifier have the capacity on following changes. And a tailor method was designed to change the fractal structure which strengthened the following capacity. The voting framework brings the networks a strong antijamming capability by means of considering several targets at the same time. The design was realized on platform VC/MFC and applied on classification of SEMG. The classification accuracy rate was 91.59%. This model is expected to work well in fields like pattern recognition of weak signal, financial forecast, etc..