Metaheuristic Algorithms in Optimizing Deep Neural Network Model for Software Effort Estimation

Effort estimation is the most critical activity for success of overall solution delivery in software engineering projects. In this context, paper’s main contributions to literature on effort are twofold. First, paper examines application meta-heuristic algorithms have a logical and acceptable parametric model estimation. Secondly, unravel benefits nature-inspired usage optimizing Deep Learning (DL) architectures estimation, presents Neural Network (DNN) based algorithms. paper, Grey Wolf Optimizer (GWO) StrawBerry (SB) applied having To validate performances these two algorithms, set nine benchmark functions wide dimensions applied. Results from GWO SB compared with five other used Experimental results showed that has comprehensive superiority terms accuracy The proposed DNN (GWDNNSB) using initial weights learning rate selection, produced better existing work