Nicolas Larue

Nicolas Larue

Final-year PhD Candidate in Machine Learning — Generative AI Robustness, Model Evaluation, Self-Supervised Learning, Computer Vision

CV

Current Position

Final-year PhD Candidate in Machine Learning
ETIS Lab, CY Cergy Paris University / ENSEA / CNRS, France
Computer Vision Laboratory, University of Ljubljana, Slovenia
EUTOPIA co-tutelle PhD programme

Research Interests

Model evaluation, AI safety evaluations, generative AI robustness, synthetic-media detection, self-supervised learning, contrastive representation learning, multimodal systems, efficient neural architectures, and quantitative machine learning.

Education

  • PhD in Machine Learning, CY Cergy Paris University / University of Ljubljana, 2021–present
  • Master 2 Research — Intelligent Communicating Systems, CY Cergy Paris University, 2020–2021
    Research project ranked 1/42
  • Engineering Degree, ENSEA, 2018–2021
    Computer Science and Systems; Artificial Intelligence and Big Data

Selected Publications

  • QN-Mixer: A Quasi-Newton MLP-Mixer Model for Sparse-View CT Reconstruction
    CVPR 2024
    Project page

  • SeeABLE: Soft Discrepancies and Bounded Contrastive Learning for Exposing Deepfakes
    ICCV 2023
    Paper

  • SSD: Neural Collapse Inspired Contrastive Continual Learning
    BMVC 2024
    Paper

Experience

  • PhD Researcher / ML Research Engineer, ETIS Lab and University of Ljubljana, 2021–present
    Research on generative AI robustness, synthetic-media detection, self-supervised representation learning, and model evaluation under distribution shift.

  • Research Engineer Intern, ETIS Lab, 2021
    Contrastive multimodal learning for video representation.

  • Software / FPGA Engineer Intern, Optiver, 2020
    Low-latency FPGA real-time trading system.

Skills

Python, PyTorch, Hydra, Lightning, Slurm, CUDA-aware workflows, C/C++, NumPy/SciPy, model evaluation, self-supervised learning, contrastive learning, generative AI robustness, computer vision, data pipelines, ablations, failure analysis.