An optimizer derived from Halpern’s method for enhanced neural network convergence and reduced carbon emissions
Open Access
- 28 July 2025
- journal article
- research article
- Published by Springer Nature in Journal of Intelligent Information Systems
- Vol. 64 (1) , 77-96
Abstract
This work examines the Halpern’s iterative method as a means to create new neural network optimizers that could surpass many current existing approaches. We introduce HalpernSGD, an innovative network optimizer that leverages Halpern’s technique to enhance the rate of convergence of the Stochastic Gradient Descent (SGD), leading to reduced carbon emissions in neural network training processes. The combination of Halpern’s iterative method and Gradient Descent (GD) has led to an algorithm with a quadratic rate of convergence compared to the simple GD. Experimental comparisons between their stochastic versions show that HalpernSGD achieves greater efficiency than SGD by requiring fewer training epochs, thus reducing energy consumption and carbon footprint while maintaining model accuracy. We also compare HalpernSGD with ADAM, identifying potential improvements to ADAM’s approach in terms of stability and convergence, and suggesting a future direction for the development of combined optimizers. The implementation for HalpernSGD can be accessed at: https://github.com/EttoreRitacco/HalpernSGD.gitKeywords
Funding Information
- Università della Calabria
This publication has 48 references indexed in Scilit:
- HalpernSGD: A Halpern-Inspired Optimizer for Accelerated Neural Network Convergence and Reduced Carbon Footprint2024
- Machine learning methods for generating high dimensional discrete datasetsWIREs Data Mining and Knowledge Discovery, 2022
- Knowledge Distillation: A SurveyInternational Journal of Computer Vision, 2021
- On the convergence rate of the Halpern-iterationOptimization Letters, 2020
- Non-convergence and Limit Cycles in the Adam Optimizer2019
- On the rate of convergence of Krasnosel’skiĭ-Mann iterations and their connection with sums of BernoullisIsrael Journal of Mathematics, 2013
- Convex Analysis and Monotone Operator Theory in Hilbert Spaces2011
- Geometric Properties of Banach Spaces and Nonlinear Iterations2009
- Iterative Approximation of Fixed Points2007
- Quelques propriétés des opérateurs angle-bornés etn-cycliquement monotonesIsrael Journal of Mathematics, 1977