Domain-Adaptation

Validating assumptions on domain relationships is not possible without target labels. We argue that, in order to make domain-adaptive classifiers more practical, it is necessary to focus on robustness; robust in the sense that an adaptive classifier will still perform at least as well as a non-adaptive classifier without having to rely on the validity of strong assumptions.

This poster recaps two collaboration projects I did during my time as Niels Stensen Fellow at the University of Copenhagen. The main …

Suppose one is faced with the challenge of tissue segmentation in MR images, without annotators at their center to provide labeled …

Sequential Domain Adaptive Intelligent Systems focuses on domain-adaptative classification over an ordered sequence of biased samples. An example of such a sequence is medical data from hospital along a geographic path.

NVPHBV is the Dutch Society for Pattern Recognition and Image Processing. During their meetings, researchers from the Netherlands have …

I presented my paper on problems with importance-weighted cross-validation under covariate shift. Under covariate shift, the standard cross-validation estimator is not consistent (i.e. it won’t return optimal hyperparameter estimates). Importance-weighting the cross-validation estimator was deemed to resolve this issue, but we show that it is still not consistent.

NVPHBV is the Dutch Society for Pattern Recognition and Image Processing. During their meetings, researchers from the Netherlands have …

Importance-weighting is a popular and well-researched technique for dealing with sample selection bias and covariate shift. It has …

SNN organized a one day symposium entitled Intelligent Machines, where an overview of recent developments was presented. The meeting aimed to establish a dialogue and to build connections between academic research, industry and public institutions in the Netherlands. I presented my preliminary work on incorporating transfer models in domain-adaptive classifiers.

Domain Adaptive Pattern Recognition explores the limits of generalization for a special case of statistical learning where training data and test data are differently biased samples of some underlying data-generating distribution.