Biologically inspired neural network layer with homeostatic regulation and adaptive repair mechanisms

This paper presents a pioneering approach to neural network design by incorporating homeostatic regulation and adaptive repair mechanisms inspired by biological nervous systems. The proposed layer architecture demonstrates the ability to maintain stable function and self-repair under perturbations, addressing fundamental brittleness issues in artificial neural networks.

The research validates the approach across critical biomedical datasets including AIDS, HIV, COX2, and protein analysis tasks, showing superior performance compared to conventional architectures. This work establishes a foundation for building more robust and resilient neural networks through biologically-principled design.

Authors:

• MD Azizul Hakim (Lead Author, Independent Researcher, Bangladesh Sweden Polytechnic Institute, Bangladesh)
• Mohammad Ifazul Alam (Co-Author, Bangladesh Sweden Polytechnic Institute, Bangladesh)

Journal: Scientific Reports (Nature Portfolio)

Volume: 15

Pages: 9114

Year: 2025

Publisher: Nature Publishing Group

DOI: 10.1038/s41598-025-09114-8

Keywords: Homeostatic regulation, adaptive repair mechanisms, biologically-inspired neural networks, molecular property prediction, robust AI systems

Recommended citation: Hakim, MD Azizul, & Alam, Mohammad Ifazul. (2025). "Biologically inspired neural network layer with homeostatic regulation and adaptive repair mechanisms." Scientific Reports, 15, 9114.
Download Paper