Neuro-Symbolic Human Activity Recognition
These models make use of prior domain knowledge and thus need less training data than purely data-driven models. Also, we can guarantee that certain constraints hold!
Deep Learning for Tabular and Categorical Data
We try to develop deep neural network models that outperform tree-based models for tabular data. Specifically, we think about how categorical features can be handled best in neural networks.
Tractable Probabilistic Models
These probabilistic models come with guarantees on the computational complexity of inference operations. This is really useful if you want to answer many queries about the same distribution.
Explaining Neural Networks
We work on methods that generate interpretable surrogate models from a given neural network. Specifically, we try to use neural networks for this task as well.
Lifted Marginal Filtering
Probabilistic activity models of dynamic systems often have symmetries. Lifted Marginal Filtering makes use of them to make inference in dynamic systems much more efficient.
Detecting Disorientation of People with Dementia
We investigate how spatial disorientation relates to to gait patterns of people with dementia.