Dynamic tunneling based regularization in feedforward neural networks
Artificial Intelligence
Handwritten Chinese Radical Recognition Using Nonlinear Active Shape Models
IEEE Transactions on Pattern Analysis and Machine Intelligence
On Generalization and K-Fold Cross Validation Performance of MLP Trained with EBPDT
AFSS '02 Proceedings of the 2002 AFSS International Conference on Fuzzy Systems. Calcutta: Advances in Soft Computing
Offline handwritten Chinese character recognition by radical decomposition
ACM Transactions on Asian Language Information Processing (TALIP)
Global descent methods for unconstrained global optimization
Journal of Global Optimization
An effective and efficient two stage algorithm for global optimization
ICMLC'05 Proceedings of the 4th international conference on Advances in Machine Learning and Cybernetics
A new constructing auxiliary function method for global optimization
Mathematical and Computer Modelling: An International Journal
Hi-index | 0.00 |
An algorithm based on gradient descent techniques with dynamic tunneling methods fur global optimization is proposed. The proposed algorithm consists of gradient descent for local search and a direct search scheme, based on dynamic tunneling technique, for repelling away from local minimum to find the point of next local descent. This search process applied repeatedly finds the global minimum of an objective function. The convergence properties of the proposed algorithm is validated experimentally on benchmark problems. A comparative computational results confirm the importance of dynamic tunneling in gradient descent techniques