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...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.展开更多
文摘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.