Neuro-Fuzzy System Based Kernel for Classification with Support Vector Machines

Simiński, Krzysztof

Krzysztof Simiński

Abstract:

Selection of proper kernel function for SVM is quite a difficult task. The paper presents the SVM with kernel generated by neuro-fuzzy system (NFS). The kernel function created in this way satisfies the Mercer's theorem. The paper is accompanied by numerical results showing the ability of NFS to approximate the appropriate kernel functions.

Reference:

Krzysztof Simiński, Neuro-Fuzzy System Based Kernel for Classification with Support Vector Machines, [chapter in] Man-Machine Interactions 3, Springer International Publishing, 2014, pp. 415-422. ([10 pkt.])

Bibtex Entry:

@incollection{id:Siminski2014NeuroSVM,
  title={Neuro-Fuzzy System Based Kernel for Classification with Support Vector Machines},
  author={Simi{\'n}ski, Krzysztof},
  booktitle={Man-Machine Interactions 3},
  pages={415--422},
  year={2014},
  publisher={Springer International Publishing},
  abstract={Selection of proper kernel function for SVM is quite a difficult 
  task. The paper presents the SVM with kernel generated by neuro-fuzzy system (NFS).  
  The kernel function created in this way satisfies the Mercer's theorem. 
  The paper is accompanied by numerical results showing the ability of NFS to 
  approximate the appropriate kernel functions.},
  note = {[10 pkt.]},
}