Efficient Approximations of Kernel Robust Soft LVQ
作者： Daniela Hofmann , Andrej Gisbrecht , Barbara Hammer
DOI: 10.1007/978-3-642-35230-0_19
关键词: Kernel (image processing) 、 Gramian matrix 、 Data structure 、 Probabilistic logic 、 Mixture model 、 Limit (mathematics) 、 Algorithm 、 Support vector machine 、 Learning vector quantization 、 Computer science
摘要: Robust soft learning vector quantization (RSLVQ) constitutes a probabilistic extension of (LVQ) based on labeled Gaussian mixture model the data. Training optimizes likelihood ratio and recovers variant similar to LVQ2.1 in limit small bandwidth. Recently, RSLVQ has been extended kernel version, thus opening way towards more general data structures characterized terms Gram matrix only. While leading state art results, this drawback that models are no longer sparse, quadratic training complexity is encountered. In contribution, we investigate two approximation schemes which lead sparse models: k-approximations prototypes Nystrom matrix. We behavior these approximations couple benchmarks.
springer.com
sci-hub.st 参考文章(18)
Daniela Hofmann, Barbara Hammer, Kernel robust soft learning vector quantization artificial neural networks in pattern recognition. ,vol. 7477, pp. 14- 23 ,(2012) , 10.1007/978-3-642-33212-8_2
Paulo J.G. Lisboa, Jose D. Martin Guerrero, Alfredo Vellido Alcacena, Making machine learning models interpretable the european symposium on artificial neural networks. pp. 163- 172 ,(2012)
Elzbieta Pekalska, Robert P W Duin, The Dissimilarity Representation for Pattern Recognition: Foundations and Applications World Scientific. ,(2005) , 10.1142/5965
Thomas Gärtner, Kernels for Structured Data ,(2008)
Xibin Zhu, Andrej Gisbrecht, Frank-Michael Schleif, Barbara Hammer, Approximation techniques for clustering dissimilarity data Neurocomputing. ,vol. 90, pp. 72- 84 ,(2012) , 10.1016/J.NEUCOM.2012.01.033
Barbara Hammer, Alexander Hasenfuss, Topographic mapping of large dissimilarity data sets Neural Computation. ,vol. 22, pp. 2229- 2284 ,(2010) , 10.1162/NECO_A_00012
P.N. Suganthan, A.K. Qin, A novel kernel prototype-based learning algorithm international conference on pattern recognition. ,vol. 4, pp. 621- 624 ,(2004) , 10.1109/ICPR.2004.78
Teuvo Kohonen, Panu Somervuo, How to make large self-organizing maps for nonvectorial data Neural Networks. ,vol. 15, pp. 945- 952 ,(2002) , 10.1016/S0893-6080(02)00069-2
Romain Boulet, Bertrand Jouve, Fabrice Rossi, Nathalie Villa, Batch kernel SOM and related Laplacian methods for social network analysis Neurocomputing. ,vol. 71, pp. 1257- 1273 ,(2008) , 10.1016/J.NEUCOM.2007.12.026
Marie Cottrell, Barbara Hammer, Alexander Hasenfuß, Thomas Villmann, Batch and median neural gas workshop on self-organizing maps. ,vol. 19, pp. 762- 771 ,(2006) , 10.1016/J.NEUNET.2006.05.018