About us

Consortium

Team

Meet the Alumni

Flanders AI

Contact

storage/airesearch/gc1.1.jpg

The goal of this work package is to investigate where the pre-processing itself can be&nbsp; automated by means of artificial intelligence and machine learning. To achieve this, leaps&nbsp; with respect to state of the art are required for actions such as: data wrangling, feature&nbsp; generation and selection, contextual segmentation, and automatically deal with&nbsp; imperfections in the data.
WP Lead: Hendrik Blockeel, KU Leuven, DTAI&nbsp;

WP 1 AI-driven data engineering 

The goal of this WP is to increase the common ground between learners and reasoners by
1) developing neurosymbolic AI models that integrate symbolic reasoning with subsymbolic learning approaches,
2) reasoning about learned models in order to verify their&nbsp; properties, assess the risk of making mistakes, check compliance by design, and exploiting&nbsp; pre-existing knowledge,
3) integrate machine learning methods in combinatorial&nbsp; optimisation both to speed up solvers and to learn constraints and (parts of) models
WP Lead: Luc De Raedt, KU Leuven, DTAI

WP 2 Symbolic and Subsymbolic Learning and  Reasoning

Unlock the potential of rich types of data and models when the user&rsquo;s goal is ill-specified&nbsp; and where domain context is important, through the intelligent automation of data&nbsp; exploration methods while enabling user interaction and incorporation of expert&nbsp; knowledge, and with a prime focus on trustworthiness (fairness/bias, privacy, transparency)
WP Lead: Tijl De Bie, Universiteit Gent, AIDA-IDLab

WP 3  AI-assisted data exploration

The goal of this work package is to develop data-efficient tools and techniques to automate&nbsp; key parts of the machine learning pipeline while characterizing and exploiting model&nbsp; uncertainties and incorporating domain and expert knowledge.&nbsp; Achieving this requires leaps from the state-of-the-art for tasks such as: automated data&nbsp; collection, data-efficient learning, expert knowledge-based learning, physics-based learning,&nbsp; automated uncertainty quantification, and automated data-to-decision.
WP Lead: Tom Dhaene, Universiteit Gent, SUMO-IDLab

WP 4 - Automation in Machine Learning

The objective of this WP is to make sense of audiovisual data (images, video, audio, speech&nbsp; and text) in a flexible manner, by reducing the need for supervision, exploiting multiple&nbsp; modalities, incorporating physics-based models and designing models for dense prediction&nbsp; tasks.
WP Lead: Tinne Tuytelaars, KU Leuven, PSI

WP 5 Understanding images and audio

WP 6 The goal of this work package is to provide a bridge between methodological research on&nbsp; the one hand, and exploitation in real world applications on the other. This involves selected focused activities in research areas focused on deployment challenges in artificial&nbsp; intelligence systems. These involve both aspects related to responsible use of AI, such as&nbsp; privacy concerns and explainability, but also aspects related to translating machine learning&nbsp; outputs into quantities interpretable by domain experts, and decision support system&nbsp; frameworks.
WP Leads: Matthew Blaschko, KU Leuven, PSI &amp;&nbsp;Yves Moreau, KU Leuven, STADIUS

WP 6 Responsible AI and Decision Support Systems

Work packages

Management team,&nbsp;<a href="https://www.esat.kuleuven.be/stadius/" target="_blank" rel="external nofollow noopener">KU Leuven - ESAT/STADIUS</a>

Bart De Moor

Management team,&nbsp;<a href="https://idlab.ugent.be/" target="_blank" rel="external nofollow noopener">UGent - IDLab</a>

Piet Demeester

Ann Ackaert

Oscar Mauricio Agudelo

WP1 Lead,&nbsp;<a href="https://wms.cs.kuleuven.be/dtai" target="_blank" rel="external nofollow noopener">KU Leuven DTAI</a>

Hendrik Blockeel 

WP2 Lead,<a href="https://wms.cs.kuleuven.be/dtai" target="_blank" rel="external nofollow noopener">&nbsp;KU Leuven DTAI</a>

Luc De Raedt

WP3 Lead,&nbsp;<a href="https://idlab.ugent.be/" target="_blank" rel="noopener">UGent - AIDA-IDLab </a>

Tijl De Bie

WP4 Lead,<a href="https://sumo.ilabt.imec.be/home" target="_blank" rel="external nofollow noopener">&nbsp;UGent SUMO-IDLab</a>

Tom Dhaene

WP5 Lead,&nbsp;<a href="https://www.esat.kuleuven.be/psi" target="_blank" rel="external nofollow noopener">KU Leuven PSI</a>

Tinne Tuytelaars 

WP6 Lead,&nbsp;<a href="https://www.esat.kuleuven.be/psi" target="_blank" rel="external nofollow noopener">KU Leuven PSI</a>/<a href="https://www.esat.kuleuven.be/stadius/" target="_blank" rel="external nofollow noopener">STADIUS</a>

Matthew Blaschko/Yves Moreau

Multiple research groups collaborate on this research domain. This table mentions the contact person and his/her affiliation.

contacts

Supporting complex decision making and actionable insights creation&nbsp; with powerful and intuitive AI-driven data science

Unlocking the value of data in a trusted and automated manner, supporting complex decision making and providing new insights that will empower individuals and society in generating major advances in healthcare, education, industry 4.0, energy systems and more.