Year: 2025
Type: article
Abstract: In this paper, we aim to minimize the function via gradient descent. However, the study involves many variables that play an important role in how the algorithm performs, such as step size, number of iterations, tolerance for the stopping criterion, and the starting value. To improve these drawbacks, this research proposes adding momentum to increase the convergence rate and gain more stability for this algorithm. This can be fixed by properly tuning the momentum parameters. This paper provides a thorough analysis of the choice of parameters and suggests guidelines for choosing an optimal configuration that can provide valuable insights into the optimization problem as a whole and practical advice for other application domains.
Source: Theoretical and Natural Science
Authors: Shiyun Xia, Yuming Cai, Xuanting Wang, Shengjie Lin, Libiao Ling
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4410129246