Autoencoders Learn Generative Linear Models
arXiv:1806.00572
Abstract
We provide a series of results for unsupervised learning with autoencoders. Specifically, we study shallow two-layer autoencoder architectures with shared weights. We focus on three generative models for data that are common in statistical machine learning: (i) the mixture-of-gaussians model, (ii) the sparse coding model, and (iii) the sparsity model with non-negative coefficients. For each of these models, we prove that under suitable choices of hyperparameters, architectures, and initialization, autoencoders learned by gradient descent can successfully recover the parameters of the corresponding model. To our knowledge, this is the first result that rigorously studies the dynamics of gradient descent for weight-sharing autoencoders. Our analysis can be viewed as theoretical evidence that shallow autoencoder modules indeed can be used as feature learning mechanisms for a variety of data models, and may shed insight on how to train larger stacked architectures with autoencoders as basic building blocks.
arXiv commentExperimental study on synthesis data added. Typos fixed
References in corpus (8)
- Provable Bounds for Learning Some Deep Representations
- Recovery Guarantees for One-hidden-layer Neural Networks
- Learning One-hidden-layer Neural Networks with Landscape Design
- Learning mixtures of separated nonspherical Gaussians
- Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs
- Why are deep nets reversible: A simple theory, with implications for training
- Zero-bias autoencoders and the benefits of co-adapting features
- Why Regularized Auto-Encoders learn Sparse Representation?