Posts by Collection

portfolio

Extending Bio-Inspired DNNs To Colour Vision

This project aimed at further improving model performance by extending the bio-inspired components (Gabor filters, Gaussian filters ,commonly used in these networks) to include colour opponency channels. This adjustment seeks to better mimic human visual processing and push the network’s performance closer to that of state-of-the-art, end-to-end trained models. The overall goal is to refine how these artificial networks process images to achieve more human-like accuracy and efficiency.

Hybrid Time-Delayed and Uncertain Internally-Coupled Complex Networks

The study integrates the Neural Mass Model (specifically the Jansen-Rit Model) with the Kuramoto model using real human brain data from various imaging techniques to create a comprehensive model that simulates brain dynamics. This advanced model allows for the observation of frequency variations, synchronization states, and electrophysiological activities, potentially improving the simulation and understanding of neurological conditions and cognitive states.

publications

talks

teaching