Neural Decoding Through Convolutional Networks: Semantic Cross-Subject Generalization in EEG Signals and Spectrograms

• Grant number: 25/14850-3
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: August 01, 2025
• End date: July 31, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Nina Sumiko Tomita Hirata
• Grantee: Raffael Raiél Trindade
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated research grant: 22/15304-4 - Learning context rich representations for computer vision, AP.TEM

Abstract

Brain decoding developed as a promising manner to develop brain computer interfaces (BCIs), aimed mainly to the interpretation of neural signals relative to tasks such as motor imagery and assisted spelling, used primarily by people physically limited. The use of convolutional neural networks (CNNs) emerged as an alternative to methods based on Riemannian geometry, motivated by the success of such networks in the field of computer vision. Hence, more complex tasks were tested using this approach, being the classification of viewed or imagined objects through brain signals one of them. However, challenges such as accuracy and cross-subject generalization still persist. Riemannian methods present good results on both aspects and are still focus of active research. This project aims to investigate and compare the performance of CNNs on those questions, analyzing a possible semantic pairing across subjects, contributing to the development of neural decoding approaches.