... We introduce a machine-learning augmented inversion framework for geophysical reservoir characterization, an ill-posed multi-parameter inverse problem. One key component is a generative network that directly models subsurface parameter distributions. The network’s augmentation can provide better...

... geophysical data geophysical inversion latent vector generative network upstream oil & gas training set exploration geophysicist 10 international conference latent space training model artificial intelligence stochastic inversion applied geoscience encoder network inversion variational...