Machine Learning

This group focuses on methodological research and application of Machine Learning methods

• Extension of existing methods on causal personalized treatment effect estimation with model-based trees and random forests, and to provide user-friendly and very flexible R implementations for extensible trees and forests

• Machine Learning methods for time-to-event data analysis (survival analysis) that go beyond right-censoring and proportional hazards

• Variants of (component-wise) gradient boosting that facilitate interpretability

• Combining ML techniques with geometric numerical methods in phase space with the goal to develop physics-informed machine learning methods. Reduced complexity models as emulators or surrogates that retain mathematical and physical properties of the underlying system (collaboration with the Max Planck Institute for Plasma Physics).

• Applications of Machine Learning in substantive sciences like medicine and psychology

• Providing state of the art software implementation and their integration within a unified framework for Machine Learning using R

Projects and Software

• mlr3: Efficient, object-oriented programming on the building blocks of machine learning

• mlr3proba: Probabilistic Supervised Learning for mlr3

• compboost: Fast and Flexible Component-Wise Boosting Framework

• partykit: A toolkit with infrastructure for representing, summarizing, and visualizing tree-structured regression and classification models

• model4you: Model-based trees and random forest for personalized treatment effect estimation

• palmtree: Partially Additive (Generalized) Linear Model Trees / PALM trees

Members