中文摘要：

在海洋鱼类胚胎冷冻保存试验中，为了解决试验材料有限，潜在的设计方案巨量的问题，笔者试图将专门解决有限样本下学习规律的支持向量机应用于此领域．采用LIBSVM和BSVM软件包中的多分类算法对不同抗冻剂、不同浓度、不同平衡时间下的组合方案进行筛选，经生物试验验证具有较高的预测准确率，为生物试验设计提供了一种新的思路和方法．

英文摘要：

In order to solve the limited experimental materials and huge potential experimental designs in the field of ocean fish embryo cryopreservation, this article tried to apply the support vector machine specialized in solving the problems of limited samples to this field. The multi-class algorithms in LIBSVM and BSVM were adopted to choose the combinations of different cryoprotectants, different densities and different balancing time. Biological experiments showed this method has high accuracy. This offers a new thinking and method to the field of biological experimental design.