Victor Livernoche | Ph.D. Student at Mila
I’m Victor Livernoche, a Montreal-born Ph.D. student at McGill University and Mila, supervised by Prof. Reihaneh Rabbany. Outside of research, I enjoy working out, playing sports, and making music. Academically, my work centers on generative modeling, anomaly and deepfake detection, and temporal graph learning. I’m especially interested in how large-scale generative systems can be used responsibly, and how we can design models and datasets that make AI more trustworthy and socially impactful.
About Me
Education
Ph.D., Computer Science
Machine learning research supervised by Prof. Reihaneh Rabbany.
M.Sc. (Thesis), Computer Science
Machine learning research supervised by Prof. Siamak Ravanbakhsh.
B.Sc., Honours Computer Science (Physics minor)
Experience
Research Scientist Student
Focused on diffusion models and anomaly detection; developed a new anomaly detection method based on diffusion models. Applied models to galactic star anomalies. Member of Mila’s Mental Health Committee.
Research Intern
Parametrized the BabyAI reinforcement learning environment in Prof. Yoshua Bengio’s group.
Undergraduate Research Assistant
Analyzed data compaction methods in large databases (with Prof. Oana Balmau).
Research Intern
Supported research operations (admin tasks, simulations, funding processes, partner communications) with Prof. Pierre‑Majorique Léger.
Research Interests
- Generative modeling for images and multimodal generation
- Energy‑based generative models (theory and applications)
- Anomaly detection and deepfake detection against misinformation
- Temporal graph representation learning
Skills
Publications
OpenFake: An Open Dataset and Platform Toward Large-Scale Deepfake Detection
Large-scale dataset and crowdsourced platform for deepfake detection; coauthors include Arodi, Musulan, Yang, Salvail‑Berard, Marceau Caron, Godbout, and Rabbany.
On Diffusion Modeling for Anomaly Detection
Explores diffusion-model-based anomaly detection, showing strong performance with efficient scoring strategies.
A Reproduction of Automatic Multi-Label Prompting: Simple and Interpretable Few-Shot Classification
Reproduction study evaluating Automatic Multi‑Label Prompting for few-shot classification.
Other Projects
Neural Network from Scratch
A Jupyter notebook implementing a neural network from scratch using NumPy.
Upcoming
...
Get In Touch
I'm always interested in discussing research opportunities, collaborations, or innovative projects in temporal graphs and machine learning.