Fuzzy sigmoid kernel for support vector classifiers
2004; Elsevier BV; Volume: 62; Linguagem: Inglês
10.1016/j.neucom.2004.07.004
ISSN1872-8286
AutoresGustau Camps‐Valls, José D. Martín‐Guerrero, José Luis Rojo‐Álvarez, Emilio Soria‐Olivas,
Tópico(s)Face and Expression Recognition
ResumoThis Letter proposes the use of the fuzzy sigmoid function presented in (IEEE Trans. Neural Networks 14(6) (2003) 1576) as non-positive semi-definite kernel in the support vector machines framework. The fuzzy sigmoid kernel allows lower computational cost, and higher rate of positive eigenvalues of the kernel matrix, which alleviates current limitations of the sigmoid kernel.
Referência(s)