Profile

Karsten Roth is a PhD researcher at the Explainable Machine Learning group as part of the European Laboratory for Learning and Intelligent Systems (ELLIS) and the International Max-Planck Research School for Intelligent Systems (IMPRS-IS) co-supervised by Prof. Zeynep Akata and Hon. Prof. Oriol Vinyals at Deepmind. He is supported by the Qualcomm Innovation Fellowship 2023.

He has completed both Bachelor and Master studies in Physics at Heidelberg University (2021). During that time, Karsten has spent time abroad in Canada as a research intern at the Montreal Institute for Learning Algorithms (MILA) supervised by Dr. Joseph Paul Cohen and Prof. Yoshua Bengio, and the Vector Institute supervised by Prof. Marzyeh Ghassemi, working on all manners of representation learning and their applications to the medical domain.

As research intern, Karsten has also worked at the Amazon AWS research lablet in Tuebingen on Anomaly Detection with Peter Gehler and Thomas Brox, and Meta AI in Paris on Disentangled Representation Learning with Mark Ibrahim, Pascal Vincent and Diane Bouchacourt.

His primary interests cover approaches to effective representation learning under different forms of distribution shifts, including zero-shot, few-shot and continual learning problems, as well as understanding generalisation behaviour of learned (multimodal) representations and foundation models. He is also very interested in their application to medicine and the sciences.

Publications

• "Waffling around for Performance: Visual Classification with Random Words and Broad Concepts", Karsten Roth, Jae Myung Kim, Almut Sophia Koepke, Oriol Vinyals, Cordelia Schmid, Zeynep Akata at International Conference for Comptuer Vision, ICCV 2023

• "Disentanglement of Correlated Factors via Hausdorff Factorized Support", Karsten Roth, Mark Ibrahim, Zeynep Akata, Pascal Vincent *, Diane Bouchacourt * at International Conference on Learning Representations, ICLR 2023

• "Momentum-based Weight Interpolation of Strong Zero-Shot Models for Continual Learning", Zafir Stojanovski *, Karsten Roth *, Zeynep Akata, Best Paper at INTERPOLATE Workshop @ Conference on Neural Information Processing Systems, NeurIPS 2022

• "A Non-isotropic Probabilistic Take on Proxy-based Deep Metric Learning", Michael Kirchhof *, Karsten Roth *, Zeynep Akata, Enkelejda Kasneci, European Conference on Computer Vision, ECCV 2022

• "Uniform Priors for Data-Efficient Learning", Samarth Sinha *, Karsten Roth *, Anirudh Goyal, Marzyeh Ghassemi, Zeynep Akata, Hugo Larochelle, Animesh Garg, Workshop on Learning with Limited Labels @ IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2022

• "Non-isotropy Regularization for Proxy-based Deep Metric Learning", Karsten Roth, Oriol Vinyals, Zeynep Akata, IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2022

• "Integrating Language Guidance into Vision-based Deep Metric Learning", Karsten Roth, Oriol Vinyals, Zeynep Akata, IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2022 (Oral)