Sam Urmian
Researcher at SLATE, University of Bergen
I work on theoretical computer science — learning algorithms and automated theorem proving — and more recently on synthetic data, privacy, recommender systems, and adaptive learning.
PhD in Informatics, University of Bergen (advisor: Mateus de Oliveira Oliveira). Thesis: Width-Based Dynamic Programming for Automated Theorem Proving in Graph Theory (PDF). MSc, Sharif University of Technology.
Software
- tabdm python
- Synthetic mixed-type tabular data via diffusion in a transformed feature space.
- orchid-ranker
- Adaptive-learning recommender that picks what a learner should work on next.
- TreeWidzard
- Engine for tree-decomposition-based algorithms.
- Zygosity pace 2023
- Heuristic solver for computing twin-width.
- LUBP
- Learning unitary branching programs.
- Sorna Research ios
- Discover, follow, and save papers from public research sources.
Projects
- ASPIRE
- Synthetic data generation, differential privacy, federated learning.
- EduTrust AI
- Trust, responsibility, and privacy in AI for education.
- AI LEARN
- National AI research centre on hybrid human–AI intelligence.
- AUTOPROVING past
- Automated theorem proving from the mindset of parameterized complexity theory.
Book
Machine Learning by Design — a draft book introducing ML from a concept-first, problem-solving perspective. Open on GitHub and updated as chapters land.
NOKI
I lead the Norwegian Olympiad in Artificial Intelligence, which trains Norway’s team for the International Olympiad in AI (IOAI). ioai.no