In this work, the implementation of time-lag autoencoders for chemical kinetics reductions is expanded to complex chemical kinetics (i.e, mechanisms with hundreds of species). A sampling methodology is analyzed for efficient hyperparameter optimization, enabling for a data-efficient methodology.
Castellanos, L., & et al. (2025). Deep Learning Dynamical Latencies for the Parameterization of Complex Chemical Kinetics. CYPHER Workshop on Machine Learning for Complex Flows, Madrid, Spain. https://hdl.handle.net/2078.5/267996