In the training process of an artificial neural network (ANN) model, optimizers play a pivotal role by determining how the ANN model’s parameters are adjusted during each iterative step. Within the machine learning (ML) literature, a range of gradient-based optimizers are commonly employed, including stochastic gradient descent (SGD), momentum-based methods, mini-batch gradient descent, as well as more advanced techniques such as AdaGrad, AdaDelta, RMSProp, and Adam. In this work, the performance of these optimizers is studied for a deep neural network (DNN) model. The DNN model is used to predict the number of Gauss quadrature points to compute the element stiffness matrix in isogeometric analysis.