Exploring the impact of Learning Rules on the neural dynamic of Face Recognition: Insights from FaceNet and MEG

In this study, we are exploring how Artificial Neural Networks compare to the brain on the task of face recognition. We are using a FaceNet model trained using different loss functions and comparing them to how the brain works. We are curious to see how well these instructions make the computer’s recognition similar to how our brains do it. We are leveraging the spatio-temporal proprieties of magnetoencephalography (MEG) data recorded while subject recognized faces. By studying the temporal, frequency, and spatial dynamics of brain activity, we hope to investigate how the brain process faces.The goal is to use these ANN models that are more smilier to the brains, to answer the “why” question of face recognition. We believe our project will bring us closer to understanding how our brains work and improve both artificial intelligence and neuroscience fields.

Faculty Supervisor:

Karim Jerbi

Student:

Partner:

École Supérieure Privée d'Ingénierie et de Technologies (ESPRIT)

Discipline:

Computer science

Sector:

Artificial Intelligence; Technology

University:

Université de Montréal

Program: