The challenges of the conventional cloud computing paradigms motivated the emergence of the next generation cloud computing architectures. The emerging cloud computing architectures generate voluminous amount of data that are beyond the capability of the shallow intelligent algorithms to process. Deep learning algorithms, with their ability to process large-scale datasets, have recently started gaining tremendous attentions from researchers to solve problem in the emerging cloud computing architectures. However, no comprehensive literature review exists on the applications of deep learning architectures to solve complex problems in emerging cloud computing architectures. To fill this gap, we conducted a comprehensive literature survey on the applications of deep learning architectures in emerging cloud computing architectures. The survey shows that the adoption of deep learning architectures in emerging cloud computing architectures are increasingly becoming an interesting research area. We introduce a new taxonomy of deep learning architectures for emerging cloud computing architectures and provide deep insights into the current state-of-the-art active research works on deep learning to solve complex problems in emerging cloud computing architectures. The synthesis and analysis of the articles as well as their limitation are presented. A lot of challenges were identified in the literature and new future research directions to solve the identified challenges are presented. We believed that this article can serve as a reference guide to new researchers and an update for expert researchers to explore and develop more deep learning applications in the emerging cloud computing architectures.

常规云计算范式的挑战促使了下一代云计算架构的出现。新兴的云计算体系结构生成大量数据，这些数据超出了浅层智能算法的处理能力。深度学习算法具有处理大规模数据集的能力，最近已开始引起研究人员的极大关注，以解决新兴的云计算架构中的问题。但是，关于深度学习架构在解决新兴云计算架构中的复杂问题方面的应用，尚无全面的文献综述。为了填补这一空白，我们对深度学习架构在新兴云计算架构中的应用进行了全面的文献调查。调查显示，在新兴的云计算架构中采用深度学习架构正逐渐成为一个有趣的研究领域。我们介绍了针对新兴云计算架构的深度学习架构的新分类法，并提供了有关深度学习以解决新兴云计算架构中复杂问题的最新技术研究成果的深刻见解。介绍了文章的综合分析及其局限性。在文献中发现了很多挑战，并提出了解决这些挑战的新的未来研究方向。