Advanced AI
Department of Electronic & Electrical Engineering • Trinity College Dublin
Module Description
This module proposes an advanced, research-oriented, dive into modern AI, with a particular emphasis on multimodal models that learn and reason across vision, language, and audio. We explore how contemporary foundation models are designed, trained, adapted, and applied to complex real-world problems.
Core Topics Covered
Multimodal AI
Vision, language, and speech models and their shared representations.
Transformers & Foundation Models
Attention, LLMs, tokenisation, positional encodings, and efficient fine-tuning.
Vision & Self-Supervised Learning
ViTs, Swin, CLIP, MAE, I-JEPA, and SAM.
Speech & Audio AI
Wav2Vec 2.0, Whisper, and multilingual models.
Generative AI
Diffusion models, latent diffusion, conditioning, and RAG.
Assessment & Projects
Assessment is based on an end-of-term examination and a group project, focusing on modern coding practices and reproducible implementation. The aim is to prepare students for research careers and high-end engineering roles.