Cluster-Based Nearest-Neighbour Classifier and Its Application on the Lightning Classification
- 58 Downloads
The problem addressed in this paper concerns the prototype generation for a cluster-based nearest-neighbour classifier. It considers, to classify a test pattern, the lines that link the patterns of the training set and a set of prototypes. An efficient method based on clustering is here used for finding subgroups of similar patterns with centroid being used as prototype. A learning method is used for iteratively adjusting both position and local-metric of the prototypes. Finally, we show that a simple adaptive distance measure improves the performance of our nearest-neighbour-based classifier. The performance improvement with respect to other nearest-neighbour-based classifiers is validated by testing our method on a lightning classification task using data acquired from the Fast On-orbit Recording of Transient Events (FORTE) satellite, moreover the performance improvement is validated through experiments with several benchmark datasets. The performance of the proposed methods are also validated using the Wilcoxon Signed-Rank test.
Keywordsnearest-neighbour classifier clustering adaptive distance
Unable to display preview. Download preview PDF.
- Franco A, Maltoni D, Nanni L. Reward-punishment editing. In Proc. International Conference on Pattern Recognition (ICPR04), Cambridge, UK, August 2004, pp.424–427.Google Scholar
- Friedman J. Flexible metric nearest neighbor classification. Tech. Rep.113, Stanford University, 1994.Google Scholar
- Hastie T, Tibshirani R. Discriminant adaptive nearest neighbor classification. IEEE Trans. PAMI, 1996, 18(6): 607–615.Google Scholar
- Domeniconi C, Peng J, Gunopulos D. Locally adaptive metric nearest neighbor classification. IEEE Trans. PAMI, 2004, 24: 1281–1285.Google Scholar
- Zhou Y, Zhang C, Wang J. Tunable nearest neighbour classifier. Lect. Notes Comput. Sci., 2004, 3175: 204–211.Google Scholar
- Eads D, Hill D, Davisa S, Perkinsa S, Maa J, Portera R, Theiler J. Genetic algorithms and support vector machines for time series classification. In Proc. Applications and Science of Neural Networks, Fuzzy Systems, and Evolutionary Computation V, Seattle, USA, December 2002, pp.74–85.Google Scholar
- Briles S, Moore K, Jones R, Blain P, Klinger P, Neagley D, Carey M, Henneke K, Spurgen W. Innovative use of DSP technology in space: FORTE event classifier. In Proc. the International Workshop on Artificial Intelligence in Solar-Terrestrial Physics, Dallas, USA, 1993.Google Scholar
- Moore K, Blain P C, Briles S D, Jones R G. Classification of RF transients in space using digital signal processing and neural network techniques. In Proc. SPIE, 2492, Orlando, USA, 1997.Google Scholar
- Kuncheva L. Combining Pattern Classifiers. John Wiley & Sons, 2004.Google Scholar