Tunable Nearest Neighbor Classifier

Yonglei Zhou, Changshui Zhang, Jingchun Wang

Abstract

A tunable nearest neighbor (TNN) classifier is proposed to handle the discrimination problems. The TNN borrows the concept of feature line spaces from the nearest feature line (NFL) classifier, to make use of the information implied by the interaction between each pair of points in the same class. Instead of the NFL distance, a tunable distance metric is proposed in the TNN. The experimental evaluation shows that in the given feature space, the TNN consistently achieves better performance than NFL and conventional nearest neighbor methods, especially for the tasks with small training sets.
BibTeX
@inproceedings{zhou2004tunable,
  author = {Yonglei Zhou and Changshui Zhang and Jingchun Wang},
  title = {{Tunable Nearest Neighbor Classifier}},
  booktitle = {DAGM-Symposium},
  year = {2004},
  doi = {10.1007/978-3-540-28649-3\_25},
}

Loading related content...

Figure