Feature Kernel Functions: Improving SVMs Using High-Level Knowledge

Abstract

Kernel functions are often cited as a mechanism to encode prior knowledge of a learning task. But it can be difficult to capture prior knowledge effectively. For example, we know that image pixels of a handwritten character result from a few strokes from a single writing implement; it is not clear how to express this in a kernel function. We investigate an explanation based learning (EBL) paradigm to generate specialized kernel functions. These embody novel high-level features that are automatically constructed from the interaction of prior knowledge and training examples. Our empirical results showed that the performance of the resulting SVM surpasses that of a conventional SVM on the challenging task of classifying handwritten Chinese characters.

Cite

Text

Sun and DeJong. "Feature Kernel Functions: Improving SVMs Using High-Level Knowledge." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2005. doi:10.1109/CVPR.2005.157

Markdown

[Sun and DeJong. "Feature Kernel Functions: Improving SVMs Using High-Level Knowledge." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2005.](https://mlanthology.org/cvpr/2005/sun2005cvpr-feature/) doi:10.1109/CVPR.2005.157

BibTeX

@inproceedings{sun2005cvpr-feature,
  title     = {{Feature Kernel Functions: Improving SVMs Using High-Level Knowledge}},
  author    = {Sun, Qiang and DeJong, Gerald},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2005},
  pages     = {177-183},
  doi       = {10.1109/CVPR.2005.157},
  url       = {https://mlanthology.org/cvpr/2005/sun2005cvpr-feature/}
}