Representation Learning: A Review and New Perspectives

TL;DR

This paper reviews recent advances in unsupervised and deep learning for data representation, emphasizing the importance of representations in explaining data variation and discussing open questions in the field.

Contribution

It provides a comprehensive overview of current methods in representation learning and highlights new perspectives and unresolved questions in the area.

Findings

Advances in probabilistic models and auto-encoders for unsupervised learning

Connections between representation learning, density estimation, and manifold learning

Open questions about objectives and inference in representation learning

Abstract

The success of machine learning algorithms generally depends on data representation, and we hypothesize that this is because different representations can entangle and hide more or less the different explanatory factors of variation behind the data. Although specific domain knowledge can be used to help design representations, learning with generic priors can also be used, and the quest for AI is motivating the design of more powerful representation-learning algorithms implementing such priors. This paper reviews recent work in the area of unsupervised feature learning and deep learning, covering advances in probabilistic models, auto-encoders, manifold learning, and deep networks. This motivates longer-term unanswered questions about the appropriate objectives for learning good representations, for computing representations (i.e., inference), and the geometrical connections between representation learning, density estimation and manifold learning.