François Pitié

Advanced AI

Department of Electronic & Electrical Engineering • Trinity College Dublin

EE55C34 Advanced AI Coursework & Generative Architectures

Module Description

An advanced, research-oriented treatment of modern artificial intelligence and machine learning architectures, extending beyond foundational deep learning into contemporary generative systems, high-dimensional representation learning, and multi-modal models.

The curriculum investigates cutting-edge paradigms in visual media synthesis, diffusion models (DDPM, Latent Diffusion), contrastive representation learning (CLIP, SimCLR), attention mechanics in Vision Transformers (ViT), and green / energy-efficient media processing in conjunction with research advancements from the European EMERALD project.

Core Topics Covered

Generative Modeling & Diffusion

Score-based generative models, denoising diffusion probabilistic models, classifier-free guidance, and conditional image synthesis.

Transformers & Vision Transformers

Self-attention mechanisms, cross-attention in multi-modal systems, patch projection, positional encodings, and Vision Transformers (ViT).

Self-Supervised Representation Learning

Contrastive learning, masked autoencoders, foundation models, and zero-shot transfer techniques.

Sustainable & Green Media AI

Pruning, quantisation, architectural efficiency, and reducing compute footprint across virtual production pipelines.

Assessment & Projects

Assessment emphasizes deep individual investigation and reproducible implementation of recent peer-reviewed conference publications, preparing postgraduate students for research careers and high-end engineering roles.