Developing Artificial Neural Networks with Complex Network Structure

• Grant number: 25/04836-3
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: September 01, 2025
• End date: December 31, 2025
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Odemir Martinez Bruno
• Grantee: Paulo Ricardo Sturion
• Host Institution: Instituto de Física de São Carlos (IFSC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

The project proposes to develop and investigate the application of connection structures inspired by Complex Network models - Random, Small-World, and Scale-Free networks - in Artificial Neural Networks, focusing on MLP and CNN architectures. The approach involves a review of the fundamental concepts of Complex Networks and the theoretical design of adapted techniques and architectures, integrating the models into the traditional structure of neural networks in a way that is compatible with available backpropagation algorithms. To evaluate the proposed architectures, a series of benchmark datasets will be selected, and the models will be implemented using the PyTorch library framework. The execution of controlled experiments will allow for the collection and analysis of data with the following objectives: (i) to investigate the performance and efficiency of the different proposed architectures, and (ii) to identify the relationship between neural network parameters - such as the number of neurons and layers - and the configurations inspired by Complex Networks. As the main scientific contribution, the project is expected to propose new techniques that improve the efficiency and performance of the networks, enabling advances in both computational performance and the biological plausibility of ANNs, and fostering future applications in artificial intelligence.