Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations

TLDR

Unsupervised learning of hierarchical generative models such as deep belief networks has attracted much interest, but scaling them to full‑size, high‑dimensional images remains challenging. This work introduces the convolutional deep belief network, a hierarchical generative model designed to scale to realistic image sizes. The model is translation‑invariant, supports efficient bottom‑up and top‑down probabilistic inference, and employs probabilistic max‑pooling to shrink higher‑layer representations in a sound manner. Experiments show that the algorithm learns useful high‑level visual features from unlabeled images and achieves excellent performance on several visual recognition tasks while enabling hierarchical inference over full‑size images.

Abstract

There has been much interest in unsupervised learning of hierarchical generative models such as deep belief networks. Scaling such models to full-sized, high-dimensional images remains a difficult problem. To address this problem, we present the convolutional deep belief network, a hierarchical generative model which scales to realistic image sizes. This model is translation-invariant and supports efficient bottom-up and top-down probabilistic inference. Key to our approach is probabilistic max-pooling, a novel technique which shrinks the representations of higher layers in a probabilistically sound way. Our experiments show that the algorithm learns useful high-level visual features, such as object parts, from unlabeled images of objects and natural scenes. We demonstrate excellent performance on several visual recognition tasks and show that our model can perform hierarchical (bottom-up and top-down) inference over full-sized images.