Chen, Lung YiLung YiChenYI-PEI LI2023-11-072023-11-072023-01-01978-3-031-37195-0978-3-031-37196-725424491https://scholars.lib.ntu.edu.tw/handle/123456789/636906Kinetic modeling can predict the performance of a reaction system and aids in understanding detailed reaction chemistry. However, high-fidelity reaction simulations require accurate thermodynamic and kinetic parameters of the involved species, which are not always available, especially for complex reaction networks containing hundreds or thousands of chemicals. This chapter aims to survey recent developments in machine learning in predicting molecular thermochemistry and kinetic properties, with an emphasis on advances in deep learning-based molecular property and reaction kinetics prediction. The pros and cons of commonly used conventional theoretical models are also discussed in this chapter.Atomic fingerprints | Machine learning | Reaction kinetics | Thermodynamic properties | Transition state searchbook part10.1007/978-3-031-37196-7_72-s2.0-85174451973https://api.elsevier.com/content/abstract/scopus_id/85174451973