Gpu-based parallel optimization and embedded system application of immune convolutional neural network

Abstract

Nowadays the image recognition system is applied more and more widely in the security monitoring, the industrial intelligent monitoring, the unmanned vehicle, and even the space exploration. As an image recognition technique, the traditional convolution neural network has some defects such as long training time, easy over-fitting and high misclassification rate. After our analysis on the network structure and parameters of the convolutional neural network, we firstly used the immune mechanism to improve the convolutional neural network and put forward an algorithm of the new immune convolution neural network. Our algorithm not only integrated the network node location and the parameter adjustment, but also dynamically adjusted the base function smoothing factor. In addition, we utilized the NVIDIA GPU to accelerate the parallel computing of the new immune convolutional neural network and built a realtime embedded image recognition system of the new immune convolutional neural network. Experimental results show that our new immune convolutional neural network has higher recognition rate, more stable performance and faster computing speed than the traditional one.