Surrogate and inverse modeling for two-phase flow in porous media via theory-guided convolutional neural network.