Approximate minimization of the regularized expected error over kernel models

• 成果类型: Article
• 署名作者: Kurkova, Vera; Sanguineti, Marcello
• 署名单位: Czech Academy of Sciences; Institute of Computer Science of the Czech Academy of Sciences; University of Genoa
• 刊物名称: MATHEMATICS OF OPERATIONS RESEARCH
• ISSN/ISSBN: 0364-765X
• DOI: 10.1287/moor.1080.0317
• 发表日期: 2008
• 页码: 747-756
• 关键词: NETWORKS

摘要

Learning from data under constraints on model complexity is studied in terms of rates of approximate minimization of the regularized expected error functional. For kernel models with an increasing number n of kernel functions, upper bounds on such rates are derived. The bounds are of the form a/n+b/root n, where a and b depend on the regularization parameter and on properties of the kernel, and of the probability measure de. ning the expected error. As a special case, estimates of rates of approximate minimization of the regularized empirical error are derived.