[论文解读] Algorithm Unrolling: Interpretable, Efficient Deep Learning for Signal and Image Processing

Vishal Monga, Yuelong Li+1 more|arXiv (Cornell University)|Dec 22, 2019

Sparse and Compressive Sensing Techniques被引用 43

一句话总结

本文综述算法展开（unrolling）作为将迭代信号处理算法转换为可训练、可解释的深度网络的原理性方法，具有效率与泛化方面的好处。

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.

研究动机与目标

• 在信号与图像处理中激发对可解释、具泛化能力的深度网络的需求。
• 解释将迭代算法映射到神经网络的算法展开框架。
• 展示展开的网络如何编码领域知识并降低数据需求。
• 综述在成像、视觉和医学成像中的应用。
• 讨论局限性与未来研究方向。

提出的方法

• 描述将算法的每次迭代变为一个网络层的展开过程。
• 通过反向传播端到端学习各层特定参数以优化性能。
• 以 Learned ISTA (LISTA) 为例，ISTA 的迭代被展开为一个可训练的网络。
• 将该方法推广到其他迭代方案，并展示参数共享与逐层变分之间的关系。
• 强调展开网络比通用网络在可解释性和参数效率方面的优势。

实验结果

研究问题

• RQ1 如何通过展开将迭代信号处理算法转化为可训练的神经网络？
• RQ2 与传统深度网络相比，展开网络在可解释性、效率和泛化方面有哪些优势？
• RQ3 在成像、视觉和医学成像领域，哪些应用受益于算法展开，原因是什么？
• RQ4 关于展开体系的收敛性与学习参数存在哪些理论结果？
• RQ5 算法展开的未解决挑战与未来方向是什么？

主要发现

• 展开网络通过嵌入来自迭代算法的领域知识，提供了相比通用神经网络的参数高效替代。
• LISTA 表明学习的 ISTA 可以在显著减少层数（迭代次数）和更快收敛的情况下达到类似的性能。
• 展开不仅限于稀疏编码，还扩展到图像去模糊、超分辨率和医学成像重建等多样化问题。
• 源自展开的网络由于结构化先验，能够在较少的训练样本下实现有竞争力甚至更优的性能。
• 该方法带来更快的推理速度和与底层算法步骤相关的可解释体系结构。