ExplainabilityTransparencyAccountabilityphysics.chem-ph
Deep learning of committor and explainable artificial intelligence analysis for identifying reaction coordinates
Toshifumi Mori, Kei-ichi Okazaki, Kang Kim, Nobuyuki Matubayasi
March 26, 2026
Chemical Physics Reviews
2 citations
Key Takeaways
-
Black-box AI models can be made interpretable through explainability techniques: The paper demonstrates that XAI methods can quantitatively measure which input variables (collective variables) most influence a deep learning model's predictions, addressing a key challenge in responsible AI—understanding why AI systems make specific decisions.
-
Explainability reveals which factors truly matter: By applying XAI to neural networks trained on molecular systems, researchers identified that only certain input variables play dominant roles in predictions, suggesting that explainability can help distinguish between noise and genuine causal factors in complex systems.
-
Interpretable AI enables scientific discovery: The framework shows that combining deep learning with explainability techniques allows researchers to assign meaningful molecular mechanisms and identify well-defined boundaries in data, demonstrating that responsible AI practices can enhance rather than hinder scientific understanding.
-
Explainability is practical for complex systems: The approach is described as "applicable to a wide range of complex molecular systems," suggesting that XAI methods can scale beyond single applications and become standard practice in AI-assisted scientific research.
Abstract
In complex molecular systems, the reaction coordinate (RC) that characterizes transition pathways is essential to understand underlying molecular mechanisms. This review surveys a framework for identifying the RC by applying deep learning to the committor, which provides the most reliable measure of the progress along a transition path. The inputs to the neural network are collective variables (CVs) expressed as functions of atomic coordinates of the system, and the corresponding RC is predicted as the output by training the network on the committor as the learning target. Because deep learning models typically operate in a black-box manner, it is difficult to determine which input variables govern the predictions. The incorporation of eXplainable Artificial Intelligence (XAI) techniques enables quantitative assessment of the contributions of individual input variables to the predictions. This approach allows the identification of CVs that play dominant roles and demonstrates that the committor distribution on the surface using important CVs is separated by well-defined boundaries. The framework provides an explainable deep learning strategy for assigning a molecular mechanism from the RC and is applicable to a wide range of complex molecular systems.