Dr. René-M. Kruse
Statistician · Data & AI Scientist · AI Engineer
Making neural networks a little more transparent — and statistical models a little more interesting. ✨
I'm a statistician who fell in love with deep learning. After a PhD in Statistics at the Centre for Statistics, University of Göttingen, I now work as a Data & AI Scientist, where I turn insights from the digital world of numbers and data into something useful for us analog humans.
My research lives in the space between statistics and deep learning — building models that are not just accurate, but interpretable. I'm especially interested in distributional and semi-parametric regression: methods that model the whole conditional distribution, not just its mean.
- Interpretable deep learning — bringing the transparency of additive models to neural networks
- Distributional regression — modeling location, scale and shape, well beyond the mean
- Open-source tooling for statisticians and ML practitioners in R, Python, and Julia
- mltpy — Fit flexible conditional distributions to continuous, censored, or covariate-dependent data using monotone Bernstein polynomial transformations.
Python - ConditionalAIC.jl — Conditional AIC and conditional model selection for mixed-effects models, re-platforming R's
cAIC4ontoMixedModels.jl.Julia - NAMLSS — Neural Additive Models for Location Scale and Shape, a framework for interpretable neural regression beyond the mean (AISTATS 2024).
- 📖 Learning Deep Textwork — a book on deep learning for text data.
Interpretable ML · distributional & additive models · teaching statistics and coding · translating research into production · and, of course, cute kittens 🐱