Study on Generalization Capability of Support Vector Machine in Splice Site Type Recognition of DNA Sequence. Recently, support vector machine has become a popular model as machine learning. A particular advantage of SVM over other machine learning is that it can be analyzed theoretically and at same time can achieve a good performance when applied to real problems. This paper will describe analytically the using of SVM to solve pattern recognition problem with a preliminary case study in determining the type of splice site on the DNA sequence, particularity on the generalization capability. The result obtained show that SVM has a good generalization capability of around 95.4 %.