dr. ir. Axel Faes

Developing advanced deep learning and foundation models integrating longitudinal MRI data, multi-omics profiles, and clinical variables for early detection and risk prediction of hepatocellular carcinoma (HCC). Building explainable AI (XAI) frameworks that link imaging features and molecular markers to clinically meaningful outcomes in precision oncology within the AI-HCC ZonMw project.

Scientific Coordinator of the Flanders AI Research Program Use Case Real World Evidence, leading multi-institutional research on privacy-preserving healthcare analytics. As Technical Machine Learning Lead of the Biomedical Data Sciences Group, developed federated learning frameworks enabling collaborative clinical research across Belgian healthcare institutions without sharing sensitive patient data. Research encompasses federated tensor regression for longitudinal health data, distributed learning architectures for heterogeneous clinical datasets, and translation of federated AI methods into practical healthcare applications.

Developed advanced machine learning methods for decoding sign language alphabet gestures from high-density intracranial brain activity (ECoG). Combined deep learning architectures with tensor regression models to capture complex spatiotemporal patterns in neural signals associated with linguistic motor control. Technical contributions included implementing block-term tensor decomposition for multi-dimensional neural data and exploring cross-subject generalization for robust BCI performance. The research advances both fundamental understanding of neural encoding of sign language and practical assistive communication technology. Conducted at the Laboratory for Neuro- and Psychophysiology under supervision of Prof. Marc Van Hulle, extending previous doctoral research in motor decoding to linguistic applications.

Provided consultancy for physical experiments using Synamps RT and Neuvo EEG Systems. Supported the data analysis for hybrid BCI with new analytical EEG modeling techniques & signal detection- and classification algorithms. Infer emotion from EEG signals using machine learning techniques. Usage of ECoG signals from semi-invasive recording for imagined movement detection and language processing.

Develop custom software for the control of the Gloreha robotic rehabilitation device, Specifically the R-TOUCH PRO Sinfonia, a custom-built robotic exoskeleton for hand rehabilitation.

Cognitive and Molecular Neuroscience PhD Thesis: Finger Movement Decoding: From Source-Localisation to Tensor Regression Modelling

I worked on the iMinds PRO-FLOW project @ Expertise centre for Digital Media (EDM) focus on the difference between the http versions (http1.1, https, http2).

Interfacing between human entity, a drone and virtual objects @ Expertise centre for Digital Media (EDM). (C++, Optitrack motion capture, custom built drone)

Leading scientists from all over the world will meet together to discuss topical problems in the field of neuroscience and artificial intelligence. The conference is organized by Lobachevsky State University of Nizhny Novgorod (UNN).

Graduates are able to engage in day-to-day conversations with deaf and hard-of-hearing individuals.

Cognitive and Molecular Neuroscience PhD Thesis: Finger Movement Decoding: From Source-Localisation to Tensor Regression Modelling

Engineering and Computer Science (76% - Cum Laude) Thesis: An Information Theoretical Approach to EEG Source-Reconstructed Connectivity (on Github)

Research Assistant: design of type-&-effect system for Eff based on row polymorphism Research Assistent: efficient compilation of algebraic effect handlers (in Eff)

Artificial Intelligence & Theoretical Computer Science (76% - Cum Laude) Thesis: Algebraic Subtyping for Algebraic Effects and Handlers (on Github)

International Business

Physics and General courses (79% - Magna Cum Laude) Thesis: Machine learning techniques for flow-based network intrusion detection systems (on Github)

Role: Scientific coordinator (UHasselt) and AI Researcher Team: approx. 10 researchers across 4 institutions

Role: Consortium partner (UHasselt), focusing on health data re-use and federated analyses Collaborations: KU Leuven (Yves Moreau), University of Antwerp (Geert Vandeweyer), VIB (Frederik Coppens) Context: ELIXIR Belgium is part of ELIXIR Europe, an ESFRI initiative funded by the European Commission and national agencies

FLkit is designed to help life scientists apply federated learning in their research. It offers guidance, techniques, and tools for working with decentralized and sensitive data, enabling privacy-preserving collaboration and deeper insights without sharing raw data. The project is in its early stages and open to contributors. The long-term goal is to make FLkit community-driven, sustained, and governed by researchers and practitioners who want to advance federated learning for life sciences.

Cardinal is a C++20 reimplementation of the Wren VM, designed for clarity, research, and education. It provides a small, well-documented virtual machine and runtime for exploring compilers, concurrency (fibers), and language embedding. The project serves as a platform for student theses and for prototyping new programming language ideas.

The open-source BTTR implementation provides researchers and practitioners with accessible, reproducible tools for tensor-based regression in neuroscience and related domains. It is actively developed and documented to support experimentation, education, and further research.

This repository serves as a step-by-step tutorial for running federated learning studies with the FL4E framework. Using the Wisconsin Breast Cancer dataset as an example, it guides researchers and students through the full workflow. Clear documentation and markdown guides explain each step, making it easier for newcomers to understand and apply federated learning in practice.

A Node.js application to create an iCalendar file for courses at KU Leuven. (>1000 active users)

Developed a room-scale AR game using everyday objects as controllers (broom, fan, vacuum, mousetrap). Demonstrated immersive physical-virtual interactions. Still used in university courses as a teaching example.

Interdisciplinary health-research summer school at the University of Twente.

Supporting high school and university students in mastering mathematics (and other natural sciences) through personalized tutoring

Inspire women to fall in love with programming (Python, Django workshops)

Teach children programming (Scratch, Python, Minecraft and Lego mindstorm).

POC (Education Committee) of Master Computer Science Engineering, POC (Education Committee) of Advanced Master Artificial Intelligence, Member of Departmental council of Computer Science, Department board of Computer Science and Faculty council of Engineering Science

Member of Board of Education, Faculty Council. Board of Student Facilities, Diversity Commission, Temporary representative in VVS (Flemish Union of Students)

Representing students interests in a Computer Science education context. Representing Computer Science education for high school students