摘要
经验风险与实际风险间的不一致是一个长期困扰机器学习(各种分类或拟合问题)的难题。统计学习理论提供了对这一问题的部分解决方法。本文从理论及现实两方面介绍经验风险与实际风险间的不一致现象,定义了算法的泛化能力,简单介绍了统计学习理论各组成部分的主要结论,并总结了这一理论的应用方向和存在的问题。
The discrepancy between the empirical risk and the true risk is a long-term trouble puzzling the researchers in the field of machine learning. The statistical learning theory (SLT) tries to solve this problem both in theory and in practice. Started with instances displaying such kind of discrepancy, this paper defined the concept of generalization and described the framework and the main results of SLT in brief. The di-rections and problems existing in the application of SLT were also summarized.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2002年第6期712-716,共5页
Computers and Applied Chemistry
基金
国家自然科学基金
宝钢基金(50174038)
