I am a PhD scientist at the working group for Computational Statistics at the Ludwig-Maximilian-University Munich, working on automatic machine learning.
At the moment, I am mainly interested in automatic feature preprocessing, feature extraction, and model selection. This encompasses parameterizing the different possibilities of choosing feature preprocessing and modelling pipelines, using black-box optimization methods (for example model based optimization) to choose well-performing parameters for this pipeline, and trading off the cost and accuracy of different performance evaluation methods in a multi fidelity optimization approach.
I am further interested in transfer learning regarding automatic machine learning: Using knowledge gained from previous evaluations of models to reach an acceptable model performance on novel datasets more quickly.
I also have a growing interest in deep learning, in the context of medical image analysis and reinforcement learning; possibly also general game playing and theorem proving.
Before finding my way into Computational Statistics and Machine Learning, I obtained an MMath degree in Theoretical Physics and an MSc degree in Biostatistics.
Institut für Statistik
Room 148, 1st floor
Phone: +49 89 2180 3521
martin.binder [at] stat.uni-muenchen.de