Machine learning (ML) methods have provided means to enhance our understanding of subsurface applications including environmental subsurface flow and transport, oil and gas, carbon sequestration, geothermal, waste disposal, and nuclear nonproliferation. Due to the heterogeneous nature of the subsurface, physics-based models must account for a wide range of spatial and temporal scales to predict any quantities of interest., therefore, such models are computationally expensive. The ability of ML to accelerate physics-based models has enabled improvements in parameter estimation, uncertainty quantification, multi-scale modeling, and performance assessment. In addition, ML techniques have been successful in extracting critical features from noisy data sets making it possible to better interpret field observations. This session invites outstanding works in numerical modeling of coupled processes and extraction of material properties in heterogeneous porous media combined with ML techniques. We strongly encourage works utilizing experimental, field, and simulation data combined with ML to investigate subsurface systems.