My main research interest is in developing machine learning models and
algorithms suitable for solving problems arising in computer vision.
High-level computer vision tasks are a unique source of hard machine learning
problems for three reasons.
First, in contrast to physics-based processes, for many tasks we currently do
not have accurate predictive models, which means we have to work in situations
with high model uncertainty and misspecification.
Second, humans excel at high-level vision tasks that were beneficial to
our evolutionary survival and therefore humans can provide accurate ground
truth data and model criticism (ground truth oracle).
Third, image and video data is available for free at an enormous scale
(data availability) with inexpensive cameras enabling novel consumer
These properties make computer vision a particularly attractive area for
machine learning research.
In terms of methodology I am particularly interested in structured
prediction, that is, using statistical models to address structured
In terms of industrial applications of my work, I have worked in
image processing (denoising, demosaicing, deblurring),
time-of-flight imaging, gesture recognition, cloud-based machine learning,