TY - JOUR A1 - Bonneau, Klara A1 - Lederer, Jonas A1 - Templeton, Clark A1 - Giambagli, Lorenzo A1 - Rosenberger, David A1 - Müller, Klaus-Robert A1 - Clementi, Cecilia T1 - Peering inside the black box by learning the relevance of many-body functions in neural network potentials N2 - Machine learned potentials based on artificial neural networks are becoming apopular tool to define an effective energy model for complex systems, either incorporating electronic structure effects at the atomistic resolution, or effectively renormalizing part of the atomistic degrees of freedom at a coarsegrained resolution. One main criticism regarding neural network potentials is that their inferred energy is less interpretable than in traditional approaches, which use simpler and more transparent functional forms. Here we address this problem by extending tools recently proposed in the nascent field of explainable artificial intelligence to coarse-grained potentials based on graph neural networks.With these tools, neural network potentials can be practically decomposed into n-body interactions, providing a human understandable interpretation without compromising predictive power. We demonstrate the approach on three different coarse-grained systems including two fluids (methane and water) and the protein NTL9. The obtained interpretations suggest that well-trained neural network potentials learn physical interactions, which are consistent with fundamental principles. KW - Neural Network KW - Explainability KW - Machine learning potentials PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-647531 DO - https://doi.org/10.1038/s41467-025-65863-0 SN - 2041-1723 VL - 16 IS - 1 SP - 1 EP - 14 AN - OPUS4-64753 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -