Locally nearest neighbor classifiers for pattern classification
β Scribed by Wenming Zheng; Li Zhao; Cairong Zou
- Publisher
- Elsevier Science
- Year
- 2004
- Tongue
- English
- Weight
- 120 KB
- Volume
- 37
- Category
- Article
- ISSN
- 0031-3203
No coin nor oath required. For personal study only.
π SIMILAR VOLUMES
A novel neuralnet-based method of constructing optimized prototypes for nearest-neighbor classiΓΏers is proposed. Based on an e ective classiΓΏcation oriented error function containing class classiΓΏcation and class separation components, the corresponding prototype and feature weight update rules are
A new nearest-neighbor (NN) rule is proposed. In this rule, the k-nearest neighbors of an input sample are obtained in each class. Two classification examples are presented to test the NN rule proposed. The number of samples misclassified N is evaluated. The minimum of N in the the NN rule proposed