2024 IMCN Best Thesis Award

Pierre-Paul DE BREUCK has received the 2024 IMCN Best Thesis Award on 23 May 2025.

His work was entitled "Small datasets, big predictions: learning methods for uncertainty-aware modelling of multi-fidelity material properties".

This Prize, which is granted yearly by the IMCN Institute, rewards the most outstanding PhD work among those who graduated during the previous civil year (2024 in the present case).  This initiative aims at promoting excellence in scientific research within the Institute.

The jury, chaired by Prof. Bernard NYSTEN, highlights Pierre-Paul DE BREUCK's the exceptional quality of his thesis that tackles a major challenge in materials science, that of using artificial intelligence to accelerate the discovery of new materials based on sparse data.

Pierre-Paul's thesis work was the starting point for a new research theme within the Institute's MODL research division, centred around artificial intelligence. It has led to the development of a code, MODNet, which has been made available to the scientific community as Open Source and which has already been used by numerous users, with over 80 citations to date. It has also resulted in several widely cited publications, including one that has been cited more than 70 times.
The jury would also like to highlight Pierre-Paul de Breuck's deep involvement in the IMCN research association.
 

Abstract

Functional materials play a critical role in various technological applications, and the availability of open databases has paved new paths for materials design. In this thesis, I address the challenge of supervised materials design with limited datasets by investigating various learning methods and proposing an all-encompassing framework called MODNet. Traditional approaches often rely on large datasets, which may not be readily available in practice. MODNet overcomes this limitation by combining a feedforward neural network with physically meaningful feature selection and joint learning. This results in faster training and superior performance on small datasets compared to existing
graph-network models. MODNet showcases excellent performance on the Matbench test suite and achieves a mean absolute test error of 0.009 meV/K/atom on the vibrational entropy of crystals at 305 K, significantly outperforming previous studies.

The thesis also addresses the critical task of uncertainty assessment in material science, utilizing an ensemble MODNet model to build confidence intervals and quantify uncertainty in individual predictions.  Furthermore, the potential of active learning is explored by using Bayesian Optimization to iteratively explore metals within the materials space. The significance of considering imbalance and bias in the training set for successful real-world applications of machine learning in materials science is emphasized.

Additionally, the thesis investigates techniques that leverage multiple quality sources for the same property, with a focus on the electronic band gap. By learning from differences between high- and low-quality values as a correction, substantial improvements in results are achieved compared to relying solely on high-quality experimental data.

This research significantly contributes to advancing machine learning in material property predictions and offers valuable insights into uncertainty assessment and data quality aspects in the field. By providing a comprehensive framework and effectively addressing critical challenges, this research opens new opportunities for accelerating materials design and discovery.

In addition to the winning PhD thesis, the selection committee also recognized the excellent quality of another PhD thesis and highlight its scientific quality. In this respect, the jury emphasizes the thesis of Laura CAPUTO entitled “Theoretical Modelling of Novel Two-Dimensional BCN Nanomaterials”, which represents a significant contribution to the development of advanced nanomaterials for opto-electronics devices.