Identifiable Deep Generative Models via Sparse Decoding
Presented by Gemma Moran, Postdoctoral Research Scientist, Data Science Institute, Columbia University
Thursday, Feb 2 2023
3:30 PM-4:30 PM ET
Webex Meeting Link
We develop the sparse VAE for unsupervised representation learning on high-dimensional data. The sparse VAE learns a set of latent factors (representations) which summarize the associations in the observed data features. The underlying model is sparse in that each observed feature (i.e. each dimension of the data) depends on a small subset of the latent factors. As examples, in ratings data each movie is only described by a few genres; in text data each word is only applicable to a few topics; in genomics, each gene is active in only a few biological processes. We prove such sparse deep generative models are identifiable: with infinite data, the true model parameters can be learned. (In contrast, most deep generative models are not identifiable.) We empirically study the sparse VAE with both simulated and real data. We find that it recovers meaningful latent factors and has smaller heldout reconstruction error than related methods.
Gemma Moran is a postdoctoral research scientist at the Data Science Institute at Columbia University, mentored by David Blei. She received her PhD in Statistics from the University of Pennsylvania, advised by Edward George and Veronika Rockova. Her research is on probabilistic machine learning and Bayesian inference. She aims to develop interpretable and reliable methods for large-scale data that help researchers gain scientific insight and guide their decision-making.