Journal Articles
Permanent URI for this communityhttp://197.211.34.35:4000/handle/123456789/1
Journal Articles
Browse
2 results
Search Results
Item A Comparative Analysis of Gradient Descent-Based Optimization Algorithms on Convolutional Neural Networks(IEEE, 2018) Dogo, E. M.; Afolabi, O. J.; Nwulu, N. I.; Twala, B.; Aigbavboa, C. O.In this paper, we perform a comparative evaluation of seven most commonly used first-order stochastic gradient-based optimization techniques in a simple Convolutional Neural Network (ConvNet) architectural setup. The investigated techniques are the Stochastic Gradient Descent (SGD), with vanilla (vSGD), with momentum (SGDm), with momentum and nesterov (SGDm+n)), Root Mean Square Propagation (RMSProp), Adaptive Moment Estimation (Adam), Adaptive Gradient (AdaGrad), Adaptive Delta (AdaDelta), Adaptive moment estimation Extension based on infinity norm (Adamax) and Nesterov-accelerated Adaptive Moment Estimation (Nadam). We trained the model and evaluated the optimization techniques in terms of convergence speed, accuracy and loss function using three randomly selected publicly available image classification datasets. The overall experimental results obtained show Nadam achieved better performance across the three datasets in comparison to the other optimization techniques, while AdaDelta performed the worst.Item EXTENDED ACCELERATED OVER-RELAXATION (EAOR) METHOD FOR SOLUTION OF A LARGE AND SPARSELINEAR SYSTEMS(Federal University of Technology, Minna, Nigeria, 2021-06-14) Khadeejah James Audu; Yahaya, Y. A.; Adeboye, K. R.; Abubakar, U. Y.In this research, we introduce a stationary iterative method called Extended Accelerated Over Relaxation (EAOR) method for solving linear systems. The method, an extension of the Accelerated Over Relaxation (A OR) method, was derived through the interpolation procedure with respect to the sub-matrices in application of a genera/ linear stationary schemes. We studied the convergence properties of the method for special matrices such as L-, H- and irreducible diagonally dominant matrices and proposed some convergence theorems. Some numerical tests were carried out to test the efficiency of the proposed method with existing methods in terms of number of iterations, spectral radius and computational time. The results revealed the superiority of the proposed EAOR method over the AOR method in terms of convergence rate.