Tabular foundation models predict energetic material properties accurately and quickly
Accurate predictions of properties of energetic materials with tabular foundation models
Machine Learning
Summary
Predicting the behavior of energetic materials like explosives is important but tricky. The authors show that a type of AI model called TabPFN can guess key traits—how sensitive a material is to impact, its decomposition temperature, and its crystal density—better than older methods. This model works well even without adjusting its settings and can also tell how confident its guesses are. It does this faster and more accurately than other machine learning models, including ones that use detailed chemistry knowledge.
What this means in practice
- •For materials engineers: Predict the safety and performance properties of energetic compounds quickly to aid material design and testing decisions.
- •For chemical data scientists: Use TabPFN models with molecular embeddings to improve accuracy and efficiency in property prediction pipelines for energetic substances.
Authors
Vagn Krister Nes Glomsrud, Mathias Bynke, Lars Henry Berge Olsen, Dennis Christensen
Abstract
We use TabPFN, a recently developed tabular foundation model, to predict three properties of energetic materials: impact sensitivity, decomposition temperature and crystal density. Without any hyperparameter tuning, TabPFN outperforms the best published models for all three properties. It also substantially outperforms a wide range of established machine-learning models, including tuned gradient boosting (XGBoost), random forests and Gaussian processes. This holds both when the models are given hand-crafted molecular descriptors and when they are given learned representations from Chemprop's message-passing neural network. In particular, for all three properties, TabPFN applied to Chemprop's own embedding is more accurate than Chemprop's predictions. TabPFN is also among the fastest models we consider, more than an order of magnitude faster than XGBoost. As TabPFN predicts the entire distribution of the target rather than just a single value, it offers built-in uncertainty quantification, which we compare with the empirical distributions of compounds with repeated measurements. Finally, we interpret the predictions using Shapley values, which rank the features by their importance.