Cost Optimization in Neural Network using Whale Swarm Algorithm with Batched Gradient Descent Optimizer

Abstract

Optimization algorithms are liable for sinking the losses and to give the most precise outcomes conceivable. Optimizers are utilized to modify the properties of neural network, for example, training rate and weights are used to reduce the losses. Optimization means a procedure of obtaining a global optimal solution for a given problem under given conditions. The real-world problems in the scientific fields, such as engineering design and economic planning, mostly are multimodal, high-dimensional, disconnected, and oscillated optimization problems. These complex problems cannot be solved well within reasonable time using traditional method based on gradient. Nature-inspired algorithms are becoming delightful in resolving mathematical optimization problems, like multiprocessor scheduling problem, vehicle routing and classification problems etc. In this manuscript, Whale Swarm Optimization algorithm on optimizing the neural networks, one of the meta-heuristic algorithms is applied to analysis of the cardiovascular disease dataset and compares the performance with Gradient Descent and RMSprop optimization techniques.

How to perform a simultaneous optimization with several response variables

“…Currently, the most used process optimization method in scientific works is the joint employment of the Desirability agglutination method with the Generalized Reduced Gradient (GRG) mathematical search method (Bell et al, 2020;Goffe et al, 2020;Karthikeyan et al, 2020;Lebron et al, 2020). A viable explanation for this fact is that this combination is present in the Optimizer function of the Minitab® software, which would facilitate the execution of this method (Gomes, Pereira, Marins, et al, 2019).…”

Scientific research trends about metaheuristics in process optimization and case study using the desirability function

“…Multi-response optimization problems often involve conflicting objectives, making them difficult to solve (Muniswamaiah et al, 2020), for example, the minimization of production time versus the minimization of equipment cost in manufacturing processes or the maximization of biomass production versus the minimization of substrate consumption in biotechnological processes. Currently, the most widely used process optimization method in scientific work is the joint employment of the Desirable agglutination method with the mathematical search method Generalized Reduced Gradient (GRG) (Bell et al, 2020;Goffe et al, 2020;Karthikeyan et al, 2020;Lebron et al, 2020).…”

Bending Force of Hot Rolled Strip Based on Improved Whale Optimization Algorithm and Twinning Support Vector Machine

“…Generally, it is complex to obtain the configuration model of hot rolling bending force because some rolling factors related to the model are nonlinear and have strong coupling, as well as the very large detection error, 2 of 15 which seriously affects the improvement of the accuracy of the bending force setting model. The mathematical model established by the traditional theory has low control accuracy in production practice [3][4][5]. Moreover, the development of the traditional bending force setting theory research has not improved the deficiency of the machine model in the nonstationary process, and this situation has seriously restricted the control accuracy and product quality.…”