Several successful papers and applications have demonstrated machine learning's limitless potential. However, when we immerse ourselves in powerful machine learning-based systems or applications, two critical research issues arise: how to ensure that the searched results of a machine learning system are not tampered with by anyone and how to prevent other users in the same network environment from easily obtaining our private data. This predicament is similar to that of other modern information systems that face security and privacy concerns. The introduction of blockchain technology presents us with an alternate method of addressing these two challenges. As a result, recent research has sought to construct machine learning systems using blockchain technology or to apply machine learning methods to blockchain systems. In this study, we provided a parallel framework to identify acceptable deep learning hyperparameters in a blockchain context using a metaheuristic algorithm to demonstrate what the combination of blockchain and machine learning is capable of. The suggested system also takes communication costs into consideration by restricting the amount of information transfers between miners and blockchain.