NEEDLE
Neural-based diffusion likelihood estimations for scientific computing
NEEDLE (NEural-basEd Diffusion Likelihood Estimations) is a Helmholtz Association grant (2025–2028) that I lead as Principal Investigator at DESY.
The project develops diffusion-based and neural likelihood methods for statistical inference in high-energy physics and related scientific computing settings. Focus areas include calibrated likelihood-ratio estimation, generative modelling of synthetic detector data, and scalable workflows for distributed infrastructure such as the WLCG.
Related work includes Neural Quasiprobabilistic Likelihood Ratio Estimation with Negatively Weighted Data.