Neural Network's Investigation And Application Basing On Decision Tree And Genetic Algorithm

Today, there is thousands of data stored by people. Then how to use the data effectively has been paid more and more attention. Data mining is just the way that can obtain useful information from thousands of data and apply it in many fields.At the beginning, this article gives a presentation about Decision Tree, Genetic Algorithm and Neural Network which are used frequently in Data-mining.Then through the analysis of Decision Tree and Neural Network, this article integrates them together and introduces a Neural Network's weights initialization algorithm basing on Decision Tree. This algorithm makes use of Decision Tree and the given data to decide the initial weights of Neural Network. Compare with the traditional algorithm, this method reduces the uncertainty of initial weights and makes the optimization of Neural Network more quickly. At the last of this part, we apply this algorithm on an example about companies'credit and achieve it by Matlab.The fourth part of this article is about an algorithm of Neural Network's architecture optimization. The algorithm integrates Genetic Algorithm and Neural Network skillfully. It uses Genetic Algorithm to solve the problem that how to optimize Neural Network's architecture. On the other hand, it also avoids the difficult matter that how to decide the measure function in Genetic Algorithm through Neural Network. At last, this algorithm is used in a problem about a market's contentment, achieving it by Matlab.