Research On The Method Of Hydrocarbon Recognition Based On Support Vector Machine

Support vector machine, based on Statistical learning theory, is one of the new methods in machines learning, which has been a hot spot following neural networks. It has been applied to many fields from pattern recognition to function regression and density estimation, and the results were excellent. Hydrocarbon recognition is a typical problem in pattern recognition. Because the cause for oil is complexity, many methods based on excellent theories including Artificial Neural Network have been applied to it but the results are not good as what we want. In this article, basic theories about support vector machine are fully introduced, and then it is applield to hydrocarbon recognition. To explain that support vector machine is an effective method in hydrocarbon recognition, bayes is also applied to that.The main works in this article include:(1)Machine learning and statistical learning theory, which are basic theory knowledge for support vector machine, are introduced in detail. Including mathmatic formulation in machine learning, VC dimension and structure risk minimization in statistical learning theory and so on.(2)Mathmatic model in support vector machine and kernel functions are discussed.(3)Algorithms for support vector machine are given. Sequential minimization optimization which is suitable for large samples is introduced in detail.(4)Analysed priciples for choosing oil samples, the bayes theory is dicussed briefly and it applied to hydrocarbon recognition.(5)Support vector machine is applied to hydrocarbon recognition, and the result indicate support vector machines is a good method in hydrocarbon recognition compare to that in bayes.