摘要: We propose a new method for the construction of nearest prototype classifiers which is based on Gaussian mixture ansatz and can be interpreted as an annealed version learning vector quantization (LVQ). The algorithm performs gradient descent cost-function minimizing classification error training set. investigate properties assess its performance several toy data sets optical letter task. Results show 1) that annealing in dispersion parameter kernels improves accuracy; 2) results are better than those obtained with standard (LVQ 2.1, LVQ 3) equal numbers prototypes; width improved capability. Additionally, principled approach provides explanation number features (heuristic) methods.
sci-hub.se 参考文章(16)
Yoav Freund, H. Sebastian Seung, Eli Shamir, Naftali Tishby, Selective Sampling Using the Query by Committee Algorithm Machine Learning. ,vol. 28, pp. 133- 168 ,(1997) , 10.1023/A:1007330508534
K. Torkkola, J. Kangas, J. Hynninen, T. Kohonen, J. Laaksonen, LVQ_PAK: The Learning Vector Quantization Program Package ,(1996)
S. Katagiri, C.-H. Lee, B.-H. Juang, New discriminative training algorithms based on the generalized probabilistic descent method Neural Networks for Signal Processing Proceedings of the 1991 IEEE Workshop. pp. 299- 308 ,(1991) , 10.1109/NNSP.1991.239512
Sambu Seo, M. Wallat, T. Graepel, K. Obermayer, Gaussian process regression: active data selection and test point rejection international joint conference on neural network. ,vol. 3, pp. 241- 246 ,(2000) , 10.1109/IJCNN.2000.861310
T. Kohonen, Improved versions of learning vector quantization international joint conference on neural network. pp. 545- 550 ,(1990) , 10.1109/IJCNN.1990.137622
Herbert Robbins, Sutton Monro, A Stochastic Approximation Method Annals of Mathematical Statistics. ,vol. 22, pp. 400- 407 ,(1951) , 10.1214/AOMS/1177729586
D. Miller, A.V. Rao, K. Rose, A. Gersho, A global optimization technique for statistical classifier design IEEE Transactions on Signal Processing. ,vol. 44, pp. 3108- 3122 ,(1996) , 10.1109/78.553484
Erik McDermott, Shigeru Katagiri, Prototype-based minimum classification error/generalized probabilistic descent training for various speech units Computer Speech & Language. ,vol. 8, pp. 351- 368 ,(1994) , 10.1006/CSLA.1994.1018
Thore Graepel, Matthias Burger, Klaus Obermayer, Phase transitions in stochastic self-organizing maps Physical Review E. ,vol. 56, pp. 3876- 3890 ,(1997) , 10.1103/PHYSREVE.56.3876
B.-H. Juang, S. Katagiri, Discriminative learning for minimum error classification (pattern recognition) IEEE Transactions on Signal Processing. ,vol. 40, pp. 3043- 3054 ,(1992) , 10.1109/78.175747