• Probabilistic models
• Dimensionality reduction techniques
• Information theory
• Applications to biological systems, in particular neural population coding

My main research goal is to develop flexible probabilistic and machine learning methods for analyzing multi-modal and multi-scale signals that follow radically different statistics. I am also interested in synergistic insights that can be drawn from simultaneous recordings of such signals.

Neuroscience is one example where such recordings become more and more important. The complexity of the brain’s organization and function demand that the brain is studied at multiple levels of organization. Different recording techniques have been developed for this purpose. Each technique has advantages and disadvantages in terms of spatial and temporal resolution and therefore contributes complementary information for understanding and decoding brain activity. The major obstacle to building models of the joint statistics of neural activity measured with different techniques or at different scales lies in the radical difference of their nature. My ongoing research is concerned with the goal of providing methods for constructing such models and thus aims at meeting the latest demands of multi-modal and multi-scale recordings.