This paper reviews recent studies in emerging directions of understanding neural-network representations and learning neural networks with interpretable/disentangled middle-layer representations. Although deep neural networks have exhibited superior performance in various tasks, the interpretability is always an Achilles' heel of deep neural networks. At present, deep neural networks obtain a high discrimination power at the cost of low interpretability of their black-box representations. We believe that the high model interpretability may help people to break several bottlenecks of deep learning, e.g., learning from very few annotations, learning via human-computer communications at the semantic level, and semantically debugging network representations. In this paper, we focus on convolutional neural networks (CNNs), and we revisit the visualization of CNN representations, methods of diagnosing representations of pre-trained CNNs, approaches for disentangling pre-trained CNN representations, learning of CNNs with disentangled representations, and middle-to-end learning based on model interpretability. Finally, we discuss prospective trends of explainable artificial intelligence.

本文综述了解神经网络表示和学习可解释/解耦的中间层表示的最新研究进展，并重点介绍了卷积神经网络 (CNNs) 的可视化、诊断、解耦、学习及其在可解释人工智能方面的前景趋势。

深度学习的可视化可解释性调查