Based On Food Production Bayesian Vector Machine Support

Agriculture is a country prosperous important guarantee, food is the lifeblood of theeconomy, and is an important element of the national economy, it is related to that countrybecome stronger, and economy becomes prosperous and social stability, so each country hasattached great importance to the food problem. In our country, the food production has beenbasically achieved goal in the early1990s that the output keeps balance and annual andsuperabundant, but with the development of the China’s economy, the improvement of theagricultural marketization and industrialization process, the relation between supply and demandin China’s food production has taken a great change. Therefore predicting the food productionbecomes the focus of research.The research work in this article includes the following respects:1. In order to create a new prediction algorithm that can predict the food production better, thisarticle used a method that combines the support vector machine and Bayesian, so this articlestudies the support vector machine theory and Bayesian algorithm firstly so that readerseasier to understand.2. Choice kernel function of support vector machine. When we choice support vector machinefor solving regression problems, kernel function is a key factor, how to choose appropriatekernel function to structure support vector machine is an important matter. Low dimensionalspace vector is complex and difficult to division, so they can be mapped to high dimensionalspace for easy to division. But this method can increase computational complexity, and theintroduction of kernel function solves this problem well. Choice an appropriate kernelfunction can reduce the calculation in high dimensional space greatly. The most commonlyused kernel functions include polynomial kernel function, Gaussian kernel function (RBF)and sigmoid kernel function. With further research, appearance of hybrid kernel functionfurther improves the performance of support vector machine.3. Optimize parameters in support vector machine. Choice and optimize parameters is animportant matter in machine learning, and the performance of support vector machine notonly relates to kernel parameters, but also relies on parameters C and other parameters. Themost commonly used methods of choosing parameters include lattice method and experiencemethod, but the two methods must be tested by making lots of experiments, and theparameters is not often better. In order to improve prediction accuracy, reduce computationalcomplexity and promote generalization ability, this article uses Bayesian evidenceframework which bases on Bayesian theory to optimize parameters.4. Create regression model of support vector machine, and use this model that bases onBayesian evidence framework to predict food production, and then use the standard support vector machine to predict food production, finally compare them to the accuracy of thealgorithm.