Long-term prediction of chaotic systems using neural networks to estimate the state evolution of a variety of chaotic dynamical systems and significantly extend the prediction time. Deep symbolic regression using neural networks to estimate the underlying mathematical expressions describing a given dataset. Interpretable Spatio-Temporal Modeling Identify the main sub-feature sets with predictive ability and interpretability from the original spatio-temporal space via causal analysis. Crime Hot Spot Forecasting using Spatio-temporal Deep Networks to estimate longāterm crime risks.
