Exploration of a mechanism to form bionic, self-growing and self-organizing neural network

摘要

Complex biological neural network is the result of long term evolution. The changes in the external environment will prompt the neural network to change its own structure. In this work, we try to build a bionic method to form self-growing and self-organizing neural networks. We have presented method for connection growing, verification and optimization. Paralleled way to simulate this process has also been presented in this work, using CPU \(+\) GPU style. We built a simulation platform to verify and test our methods. We conducted a series of experiments, to verify our method, and also to testify the ability of our algorithm. Results show that our method could generate neural network that could be helpful in detecting moving target.