원문정보
초록
영어
Convolutional neural network (CNN) -- the result of the training is affected by of initial value of the weights. It is concluded that the model is not necessarily the best features of expression. The use of genetic algorithm can help choosing the better characteristics. But there almost was not literature study of the combining genetic algorithm with CNN. So this research has a lot of space and prospects. GACNN convolution genetic neural network model based on random sample has a better solution to obtain the unknown character expression. CNN individual training set uses a random data set. At the same time, the crossover and the mutation genetic algorithm bring random factors. There may are unknown feature expressions that may be appropriate. Experiments are based on accepted MNIST data sets, and the experimental results proved the advantages of the model.
목차
1. Introduction
2. Algorithm Constitute
2.1 CNN
2.2 GA Algorithm
2.3 The Algorithm after Combining with Two Parts:
3. Introduction of the Experiment
4. The Experimental Results
5. Analysis of Experimental Results
6. Sum Up the Main Points
References