Improved neural network for SVM learning

IEEE Transactions on Neural Networks - Tập 13 Số 5 - Trang 1243-1244 - 2002
D. Anguita1, A. Boni2
1Department of Biophysical and Electronic Engineering, University of Genova, Genoa, Italy
2Department of Information and Communication Technologies, University of Trento, Trento, Italy

Tóm tắt

The recurrent network of Xia et al. (1996) was proposed for solving quadratic programming problems and was recently adapted to support vector machine (SVM) learning by Tan et al. (2000). We show that this formulation contains some unnecessary circuits which, furthermore, can fail to provide the correct value of one of the SVM parameters and suggest how to avoid these drawbacks.

Từ khóa

#Neural networks #Support vector machines #Support vector machine classification #Quadratic programming #Machine learning #Circuits #Hardware #Proposals #Very large scale integration #Differential equations

Tài liệu tham khảo

10.1109/72.977319 schölkopf, 2002, Learning with kernels stern, 1965, Theory of Nonlinear Networks and Systems tan, 0, neural network realization of support vector methods for pattern classification, Proc Int Joint Conf Neural Networks IJCNN 2000, 411 tan, 0, neural network realization of support vector methods for pattern classification, IEEE Transactions on Neural Networks vapnik, 1998, Statistical Learning Theory 10.1109/72.548188 burges, 1999, uniqueness of the svm solution, Advances in Neural Information Processing Systems NIPS, 223 10.1023/A:1009715923555 10.1007/BF00994018 cichocki, 1993, Neural Networks for Optimization and Signal Processing 10.1023/A:1012431217818 cristianini, 2000, An Introduction to Support Vector Machines 10.1109/IJCNN.2001.938746 10.1049/el:19981092 10.1023/A:1012479116909