Abstract
Recently, deep learning and blockchain technologies have gained successful attention due to the high potential of generating accurate decisions and data security, respectively. The data provenances characteristics such as transparency, traceability, and trustworthiness are provided by the vast majority of centralized server-based deep learning approaches. This article examines the advantages of combining deep learning algorithms with blockchain technology. In addition, the most effective strategy for combining these two technologies to achieve the best result is identified through the most recent state-of-the-art literature. In this manner, the article is divided into seven thematic taxonomies based on the literature review: applications of deep learning and blockchain, deep learning techniques, protocols, domains, types of blockchain, and datasets. We have outlined the advantages and disadvantages of blockchain-based deep learning frameworks to facilitate insightful discussions.
