支持向量機用于二噁英類有機污染物的定量構效研究.pdf

1、Chapter 1 Introduction1.1 Introduction to Support vector machines (SVM)Support vector machine (SVM), developed by Vapnik [1-2] , is a new type of statisticallearning model with a lot of advantages such as ：(1) in theory,

2、 a global optimal solutionwill be found, (2) the result is a general solution avoiding overtraining, (3) the solution issparse and only a limited objects of training points contribute to this solution, and (4)nonlinear s

3、olutions can be calculated efficiently due to the usage of inner products [3] . Thebasic idea of applying SVM for regression can be stated briefly as following: First, mapthe input vectors into one feature space (possibl

4、e with a higher dimension), either linearlyor nonlinearly, which is relevant with the selection of the kernel function. Then, within thefeature space from the first step, our goal is to find a linear relationship between

5、 theregressors and the dependent variables. SVM training always seeks a global optimizedsolution and avoids overfitting, so it has the ability to deal with a large number of features.A more particular description to the