Transformation of Input Domain for SVM in Regression Task

Simiński, Krzysztof

Krzysztof Simiński

Abstract:

Support vector machines (SVM) and neuro-fuzzy systems (NFS) are efficient tools for regression tasks. The problem of the SVMs is the proper choice of kernel functions. Our idea is to transform the task's domain with NFS so that linear kernel can be applied. The paper is accompanied by numerical experiments.

Reference:

Krzysztof Simiński, Transformation of Input Domain for SVM in Regression Task, [chapter in] Man-Machine Interactions 3, Springer International Publishing, 2014, pp. 423-430. ([10 pkt.])

Bibtex Entry:

@incollection{id:Siminski2014Transformation,
  title={Transformation of Input Domain for SVM in Regression Task},
  author={Simi{\'n}ski, Krzysztof},
  booktitle={Man-Machine Interactions 3},
  pages={423--430},
  year={2014},
  publisher={Springer International Publishing},
  abstract={Support vector machines (SVM) and neuro-fuzzy systems (NFS) are 
  efficient tools for regression tasks. The problem of the SVMs is the proper 
  choice of kernel functions. Our idea is to transform the task's domain with 
  NFS so that linear kernel can be applied. The paper is accompanied by 
  numerical experiments.},
  note = {[10 pkt.]},
}