View publication

This paper was accepted at the workshop on Overparameterization: Pitfalls and Opportunities at the ICML 2021 conference.

Deep linear networks trained with gradient descent yield low rank solutions, as is typically studied in matrix factorization. In this paper, we take a step further and analyze implicit rank regularization in autoencoders. We show greedy learning of low-rank latent codes induced by a linear sub-network at the autoencoder bottleneck. We further propose orthogonal initialization and principled learning rate adjustment to mitigate sensitivity of training dynamics to spectral prior and linear depth. With linear autoencoders on synthetic data, our method converges stably to ground-truth latent code rank. With nonlinear autoencoders, our method converges to latent ranks optimal for downstream classification and image sampling.

Related readings and updates.

Apple at ICML 2021

Apple is sponsoring the thirty-eighth International Conference on Machine Learning (ICML). This conference focuses on the advancement of the branch of artificial intelligence known as machine learning and will take place virtually from July 18 to 24.

See event details

Variational Neural Machine Translation with Normalizing Flows

Variational Neural Machine Translation (VNMT) is an attractive framework for modeling the generation of target translations, conditioned not only on the source sentence but also on some latent random variables. The latent variable modeling may introduce useful statistical dependencies that can improve translation accuracy. Unfortunately, learning informative latent variables is non-trivial, as the latent space can be prohibitively large, and the…
See paper details