Improvement of generalization ability for identifying dynamical systems by using universal learning networks

Neural Networks

Volumn 14, Issue 10, 2001, Pages 1389-1404

  Hirasawa, Kotaro ^a   Kim, Sung Ho ^a   Hu, Jinglu ^a   Murata, Junichi ^a   Han, Min ^a   Jin, Chunzhi ^a  

^a KYUSHU UNIVERSITY   (Japan)

Author keywords

Generalization ability; Regularization technique; Robustness; Second order derivatives; Universal learning networks

Indexed keywords

COMPUTATIONAL METHODS; COMPUTER SIMULATION; LEARNING ALGORITHMS; LEARNING SYSTEMS;

UNIVERSAL LEARNING NETWORKS (ULN);

NEURAL NETWORKS;

ACCURACY; ALGORITHM; ARTICLE; DYNAMICS; LEARNING; MODEL; NERVE CELL NETWORK; NOISE; PRIORITY JOURNAL; SIMULATION; TASK PERFORMANCE; TECHNIQUE;

COMPUTER SIMULATION; LEARNING; NEURAL NETWORKS (COMPUTER);

EID: 0035200301     PISSN: 08936080     EISSN: None     Source Type: Journal    
DOI: 10.1016/S0893-6080(01)00117-4     Document Type: Article

References (26)

1. Computing second derivatives in feed-forward network: A review
2. Networks and the best approximation property
3. On fuzzy neuron models
4. Generalization ability of universal learning network by using second order derivatives
5. Chaos control on universal learning networks
6. Universal learning networks and its application to robust control
7. Universal learning networks and its application to chaos control
8. Identification with dynamic neural networks - Architecture, comparisons, applications
9. Structure learning with forgetting
10. Automatic task decomposition in modular networks by structure learning with forgetting
11. Trajectory production with the adaptive time-delay neural network
12. Networks for approximation and learning
13. Regularization algorithms for learning that are equivalent to multilayer networks
14. The perceptron: A probabilistic model for information storage and organization in the brain
15. Learning representation by back-propagation errors
16. Overtraining, regularization and searching for a minimum with application to neural networks
17. Nonlinear black-box modeling in system identification: A unified overview
18. Solution of incorrectly formulated problems and the generalization method
19. Generalization by weight-elimination with application to forgetting
20. A learning algorithm for continually running fully recurrent neural networks