Tom Heskes
Tom Heskes is a professor of data science at Radboud University Nijmegen whose research covers artificial intelligence, machine learning and data science. He completed a PhD thesis on learning processes in neural networks there in 1993, received an NWO Vici grant in 2006, and was research director of the Institute for Computing and Information Sciences from 2009 until 2014. In 2017, he co-founded the spin-off Machine2Learn, which Silo AI announced it was acquiring in 2023, and he initiated Radboud AI. With Evi Sijben, Ioan Gabriel Bucur and Tom Claassen, he published Causal Shapley Values at NeurIPS in 2020, a paper that uses Pearl's do-calculus to explain model predictions with causal knowledge. Since September 2023, he is vice dean of research of the Faculty of Science and an ELLIS Fellow in the Nijmegen unit.
- Now at
- Radboud University Nijmegen and AI for Parkinson lab
- Born
- 1966
Years link to their sources.
Timeline
-
1993
PhD on neural network learning
At Radboud University Nijmegen, he completes his PhD thesis, Learning processes in neural networks, in 1993.
-
1998
Multi-task learning at ICML
In 1998, he publishes 'Solving a huge number of similar tasks: a combination of multi-task learning and a hierarchical Bayesian approach' in the Proceedings of the International Conference on Machine Learning.
-
2006
An NWO Vici grant
In 2006, he receives an NWO Vici grant and publishes 'Convexity arguments for efficient minimization of the Bethe and Kikuchi free energies' in the Journal of Artificial Intelligence Research.
-
2009–2014
Research director of iCIS
From 2009 until 2014, he is research director of the Institute for Computing and Information Sciences (iCIS) at Radboud University Nijmegen.
-
2017–2020
Education director in computing science
From February 2017 to February 2020, he is director of education for computing science and information sciences at Radboud University.
-
2017–2023
Radboud AI and Machine2Learn
In 2017, he co-founds the spin-off Machine2Learn with Ali Bahramisharif, an Amsterdam company that offers machine vision and machine learning solutions, especially for machinery and process quality control. He also initiates Radboud AI. Silo AI announces in June 2023 that it is acquiring Machine2Learn.
-
2020
Causal Shapley values
In 2020, he publishes 'Causal Shapley Values: Exploiting Causal Knowledge to Explain Individual Predictions of Complex Models' at NeurIPS with Evi Sijben, Ioan Gabriel Bucur and Tom Claassen. Using Pearl's do-calculus, the paper derives Shapley values, a method for explaining machine learning predictions, for general causal graphs.
-
Since September 2023
Vice dean for research
In September 2023, he succeeds Richard van Wezel as vice dean of research of the Faculty of Science at Radboud University. His research covers artificial intelligence, machine learning and data science, and he is a professor of data science. He is an ELLIS Fellow in the Nijmegen unit and one of the co-directors of the AI for Parkinson lab.
Career
- Since 2009Radboud University NijmegenResearch Director, Director of Education for Computing Science and Information Sciences
- NowAI for Parkinson labCo-director
- From 2017Machine2LearnCo-founder
Education
- 1993Radboud University NijmegenPhD
Colleagues
Sources and credits
Photo: We have no published photo of Tom Heskes yet. Send us a free-licensed photo.
- Publications of year 1993 (Tom Heskes)
- Publications of year 1998 (Tom Heskes)
- Prof. T.M. Heskes (Tom) | Radboud University
- Publications of year 2006 (Tom Heskes)
- Tom Heskes (personal page, Radboud University)
- Silo AI acquires Amsterdam's Machine2Learn to bolster its presence in Benelux and Western Europe - Silicon Canals
- Silo AI acquires Radboud spin-off Machine2Learn | Radboud University
- Causal Shapley Values: Exploiting Causal Knowledge to Explain Individual Predictions of Complex Models
- Tom Heskes appointed new vice-dean of research Faculty of Science | Radboud University
- Tom Heskes | European Laboratory for Learning and Intelligent Systems