Loic Lorente Lemoine
(MPhil) CompSci @
Cardiff University
Hello there! I'm Loic, a Computer Science student at Cardiff University, currently completing my MPhil and working as a research assistant with the Agile CPS Lab.
I earned my BSc in Computer Science from CU in 2024, graduating with First Class Honours after completing my dissertation on EEG-based seizure detection with TinyML. Currently, my master's projects investigates machine learning for defence against vehicle CAN attacks."
As such, my research focuses are on Machine Learning.
Education
Cardiff University
M.Phil. in Computer Science
Advisor: Dr. Amir Javed, Dr. Nick Pham
Thesis: A Vehicle CAN (Controller Area Network) On-Board Security Operations Center (SOC)
Cardiff University
B.S. in Computer Science
Thesis: Design a tiny machine learning model to detect epileptic seizures on wearables
Experience
Research Assistant — AgileCPS Labs, Cardiff University
Publications
EIFCOM 2024
Stress-GPT: Stress detection with an EEG-based foundation model
Catherine Lloyd, Loic Lorente Lemoine, Reiyan Al-Shaikh, Kim Tien Ly, Hakan Kayan, Charith Perera, Nhat Pham
We fine-tune a large language model for stress detection and evaluate it on a 40-subject open stress dataset. This is followed by additional experiments comparing traditional machine learning methods, with key observations highlighted to guide future research directions.
Awards
2024 - Cardiff University
Best Final Year Project, BSc Computer Science and Variants
2024 - Global Wales, Cardiff University, and Vietnam National University - Ho Chi Minh University of Technology
First Prize, Student Poster Competition on “AI, Smart Healthcare, and IoT”
2024 - Thales Group and Cardiff University
Thales Group MPhil Scholarship
Repos
Epileptic Seizure Detection with Tiny Machine Learning
A study on the use of tiny CNN models to detect three common types of epileptic seizures using melspectrogram EEG data. Quantized and deployed models remain reliable, with tonic-clonic, non-specific, and absence seizures presenting accuracies of 85%, 81%, and 99% respectively. Developed as a part of my undergraduate dissertation.