Blockchain and Machine Learning as Deep Reinforcement

Due to its capacity to make wise decisions, deep learning has become extremely popular in recent years. The current generation of deep learning, which heavily rely centralized servers, are unable to offer attributes like operational transparency, stability, security, and reliable data provenance. Additionally, Single point of failure is a problem that deep learning designs are susceptible since they need centralized data to train them. We review the body of research on the application of deep learning to blockchain. We categorize and arrange the literature for developing topic taxonomy based their criteria: Application domain, deep learning-specific consensus mechanisms, goals for deployment and blockchain type. To facilitate meaningful discussions, we list the benefits and drawbacks of the most cutting-edge blockchain-based deep learning frameworks.