An Experimental Approach towards the Performance Assessment of Various Optimizers on Convolutional Neural Network

Artificial Intelligence is a technique of modeling a computer, a computer administered-robot, in the indistinguishable manner the acute humans reflect. Machine Learning is a mechanism of data evaluation that automatizes rational model development. It is a branch of Artificial Intelligence based on the objective that systems can imbibe from the input, discover patterns and resolve with nominal human intrusion. Deep Learning is a sub-discipline of Machine Learning where Layers are used in it to create Artificial Neural Network (ANN). Convolutional Neural Network (CNN), a Deep Learning Architecture has been the most influential innovations in the discipline of Computer Vision. Optimizers shape and mold the model into its most accurate form by futzing with the weights. In this paper, seven optimizers namely Stochastic Gradient Descent (SGD), RMSProp, Adam, Adamax, Adagrad, Adadelta, and Nadam are implemented in CNN on Indian Pines Dataset and accuracy comparison results are shown graphically where Adamax outperforms with 99.58% accuracy.