Towards Extracting Faithful and Descriptive Representations of Latent Variable Models

Open in new window