Konstantinos Kontras

Konstantinos Kontras

Multimodal Machine Learning

PostDoc Researcher KU Leuven

Hi, I'm Konstantinos

I am currently a Postdoc with Maarten De Vos in the Biomed group of STADIUS KU Leuven in Belgium working on multimodal learning.

I was also a PhD in the group working on Multimodal Fusion and Multimodal Competition, issues that arise during training of multimodal models, supervised by Maarten De Vos , Christos Chatzichristos and Johan Suykens

Before the PhD, I was a part of PSI/Visics and TRACE group working on uncertainty for autonomous vehicles, i.e. Bayesian Model Average - Active Learning for semantic segmentation, supervised by Mark Proesmas and Luc Van Gool

News

April 2025: MIT-KU Leuven Seed Fund - Awarded €30,000 (first personally written proposal)
Febr 2025: Defended the PhD, officially a doctor!
Jan 2025: New Arxiv Multimodal Fusion Balancing Through Game-Theoretic Regularization
Dec 2024: Submitted the PhD Thesis
Nov 2024: VSC Tier-1 Compute Grant - Awarded €40,000
Oct 2024: Seminar on Multimodal Competition soon on YouTube
Jul 2024: New paper on BMVC24 Improving Multimodal Learning with Multi-Loss Gradient Modulation !
March 2023: Talk in the EU Parliament Health Tech Summit on the role of ML/AI on future health systems.
March 2023: CoRe-Sleep: A Multimodal Fusion Framework for Time Series Robust to Imperfect Modalities published in IEEE TSNRE.

Publications

(2023). CoRe-Sleep: A Multimodal Fusion Framework for Time Series Robust to Imperfect Modalities . IEEE TSNRE 2024. Cite
(2023). Is there benefit in training multimodal models from scratch? ArXiv . Cite
(2024). Improving Multimodal Learning with Multi-Loss Gradient Modulation BMVC24 . Cite
(2024). Multimodal Competition Regularizer ArXiv. Cite

Contact