|
Dr. Efstratios Gavves is an Associate Professor at the University of Amsterdam, Lecturer of Deep Learning at the MSc in AI, Scholar ELLIS European Network of Excellent in AI, and co-founder of Ellogon.AI and LYDS.AI.
His core scientific hypothesis, and research focus is the computational understanding of time, dynamics, and causality, in particular for robotics and sciences, as time and dynamics is the fundamental missing piece in the AGI puzzle towards genuine extrapolation. On this, he pioneered dynamical video representations in visual spatiotemporal data in 2014 while a postdoctoral researcher at KU Leuven with prof. T. Tuytelaars, while in 2016, he introduced the paradigm-shifting Siamese algorithms in visual object tracking, challenging 30 years of practice, now adopted by virtually all state-of-the-art. Before that, he completed his doctorate at the University of Amsterdam with prof. A. Smeulders and prof. C. Snoek on fine-grained recognition.
Since 2021 he has been spearheading research on Causal Representation Learning for algorithms that learn completely autonomously by integrating cause-and-effect in their core. To research the computational learning of time, he was awarded the ERC Career Starting Grant 2020, the NWO VIDI, and is co-PI in the ELLIOT Horizonflagship project. He further collaborates extensively with industry, including Toyota on Robot AI, Qualcomm on Dynamical Deep Learning, and Elekta on AI-driven Adaptive Radiotherapy. Efstratios currently supervises a team of more than 10 doctoral students, on theory and applications of Deep Learning and Dynamics, and Computer Vision. He has also been advancing AI algorithms for identifying cancer patients that benefit from immunotherapy.
|
