摘要
In Model Recognition, although a non-linear discriminating function may be superior to linear or quadraticclassifiers, it is difficult to construct. This paper proposes a constructing method using Legendre polynomials. Theselection of an optimal set of Legendre is determined by the MDL criterion. Results using many data show the effec-tiveness of this method.
In Model Recognition, although a non-linear discriminating function may be superior to linear or quadratic classifiers, it is difficult to construct. This paper proposes a constructing method using Legendre polynomials. The selection of an optimal set of Legendre is determined by the MDL criterion. Results using many data show the effectiveness of this method.
出处
《计算机科学》
CSCD
北大核心
2002年第12期153-154,145,共3页
Computer Science
基金
国家自然科学基金(60173053)
