Algorithm Unrolling: Interpretable, Efficient Deep Learning for Signal and Image Processing

TL;DR

Algorithm unrolling bridges iterative algorithms and deep neural networks, enhancing interpretability and efficiency in signal and image processing, while reducing training data requirements.

Contribution

This paper reviews the development, techniques, and theoretical foundations of algorithm unrolling in signal and image processing, highlighting its advantages and future challenges.

Findings

Unrolled networks offer interpretability and efficiency.

Connections between iterative algorithms and neural networks are established.

Theoretical insights into unrolling methods are summarized.

Abstract

Deep neural networks provide unprecedented performance gains in many real world problems in signal and image processing. Despite these gains, future development and practical deployment of deep networks is hindered by their blackbox nature, i.e., lack of interpretability, and by the need for very large training sets. An emerging technique called algorithm unrolling or unfolding offers promise in eliminating these issues by providing a concrete and systematic connection between iterative algorithms that are used widely in signal processing and deep neural networks. Unrolling methods were first proposed to develop fast neural network approximations for sparse coding. More recently, this direction has attracted enormous attention and is rapidly growing both in theoretic investigations and practical applications. The growing popularity of unrolled deep networks is due in part to their potential in developing efficient, high-performance and yet interpretable network architectures from reasonable size training sets. In this article, we review algorithm unrolling for signal and image processing. We extensively cover popular techniques for algorithm unrolling in various domains of signal and image processing including imaging, vision and recognition, and speech processing. By reviewing previous works, we reveal the connections between iterative algorithms and neural networks and present recent theoretical results. Finally, we provide a discussion on current limitations of unrolling and suggest possible future research directions.