Josephine Oettinger

Research Interests

• Geometric Deep Learning with special focus on energy-based canonicalisation, i.e., Diffeomorphism-Equivariant Neural Networks (preprint, PDF)
• Lie group theory in (medical) image computing
• Discrete optimisation in quantum computing, i.e., Non-Binary Quantum Amplitude Amplification For Discrete Optimization (thesis)

Talks and Posters

2026

96th Annual Meeting of the International Association of Applied Mathematics and Mechanics (GAMM), Stuttgart, Germany:
Diffeomorphism-Equivariant Neural Networks (slides)

Talk at the group research seminar of the Computational Imaging group at DESY, Hamburg, Germany:
Towards Diffeomorphism-Equivariant Neural Networks via Canonicalization (slides)

Forty-Third International Conference on Machine Learning (ICML 2026), Seoul, South Korea:
Towards Diffeomorphism-Equivariant Neural Networks via Canonicalization (poster)