What is the role of dynamics and causality for Deep Learning that not only understands but also interfaces with the physical world?<
This question fascinates me and drives my research on Causal Computer Vision. The vision is Embodied General Intelligence and eventually, a novel paradigm of Cyberphysical AI.
With Cyberphysical AI I want to address the fundamental chasm between information and matter, that is the observation that machines in the digital world and machines in the physical world work fantastically in isolation, but cannot easily be combined. Bridging the digital and physical world will be key in not only Robot Learning but also automating Biomedical, Health, IoT, Manufacting, Sciences, and so on with AI.
My research is supported by an ERC StG and NWO VIDI, and academic-industry ICAI labs: QUVA with Qualcomm, and POP-AART with Elekta and NKI.
→ Ilze presents Modulated Neural ODEs in NeurIPS 2023.
→ Miltos presents Latent Field Discovery in Interacting Dynamical Systems with Neural Fields in NeurIPS 2023.
→ Haochen presents Causal Identifiability from Temporal Intervened Sequences in ICCV 2023 as oral.
→ Mohammad presents Time Does Tell: Self-Supervised Time-Tuning of Dense Image Representations in ICCV 2023.
→ Lucas had Spatio-temporal physics-informed learning: A novel approach to CT perfusion analysis in acute ischemic stroke in IEEE MedIA 2023.
→ Wenzhe had PC-Reg: A pyramidal prediction-correction approach for large deformation image registration in IEEE MedIA 2023.