Verifying the Accuracy of GDAM algorithm on Multiple Classification Problems
Nazri Mohd Nawi, M.Z Rehman, Abdullah Khan
Nazri Mohd Nawi
Available Online January 2014.
- neural networks, gradient descent, adaptive momentum, adaptive gain
- Back-Propagation Neural Network (BPNN) algorithm is a widely used technique implemented in many engineering disciplines. Despite solving several practical problems around the globe, BPNN still faces problems like slow convergence, network stagnancy and convergence to local minima. Many alternative ways of improving the convergence rate in BPNN are suggested by previous researchers such as the careful selection of input weights and biases, learning rate, momentum, network topology, activation function and value for ‘gain’ in the activation function. This research propose an algorithm for improving the working performance of back-propagation algorithm which is ‘Gradient Descent with Adaptive Momentum (GDAM)’ by keeping the gain value fixed during all network trials. The performance of GDAM is compared with ‘Gradient Descent with fixed Momentum (GDM)’ and ‘Gradient Descent Method with Adaptive Gain (GDM-AG)’. The results show that GDAM is a better approach than previous methods and shows better accuracy on selected classification problems like Wine Quality, Mushroom, Thyroid disease Breast Cancer, IRIS, Australian Credit Card Approval, Pima Indian Diabetes, and Heart Disease.
- Open Access
- This is an open access article distributed under the CC BY-NC license.
Cite this article
TY - CONF AU - Nazri Mohd Nawi AU - M.Z Rehman AU - Abdullah Khan PY - 2014/01 DA - 2014/01 TI - Verifying the Accuracy of GDAM algorithm on Multiple Classification Problems BT - Proceedings of the 2013 International Conference on Advances in Intelligent Systems in Bioinformatics PB - Atlantis Press SP - 51 EP - 57 SN - 1951-6851 UR - https://www.atlantis-press.com/article/11357 ID - Nawi2014/01 ER -