Privacy-Preserving in FinTech using Deep Learning with Federated Learning in Cryptocurrency


  • Shailendra Sharma
  • Bonthu Kotaiah
  • Samarth Singh
  • K.V.Daya Sagar
  • S Durga
  • Nidhi Sree K



Federated learning, Blockchain technology, deep learning, privacy.


Recent developments in deep learning techniques have produced significant improvements in long-standing AI jobs like drug discovery, gene analysis, and speech and image recognition. Although deep learning has numerous benefits, the identical training dataset that has did it so reliable also raises serious privacy concerns to address these privacy concerns, McMahan et al. created Federated Deep Learning, a together distributed deep learning paradigm for the mobile devices (FDL). Deep learning and distributed computation are essentially combined in FDL, where several parties get involved into the process of distributed training and parameter server records track of a deep learning model that needs to be built. The central parameter server first distributes a pre-trained model globally on a common set of data to each participant. Then, in each cycle, each party utilizes a local dataset to the train and improve the current global model. The gradients are gathered from each party by the central parameter server, which then utilises them to build a new global oriented model for upcoming iteration. At final stage the different parties and the central parameter of server repeat the aforementioned procedure till the global modelling attains a particular accuracy or ideal convergence.




How to Cite

Shailendra Sharma, Bonthu Kotaiah, Samarth Singh, K.V.Daya Sagar, S Durga, & Nidhi Sree K. (2022). Privacy-Preserving in FinTech using Deep Learning with Federated Learning in Cryptocurrency. Journal of Pharmaceutical Negative Results, 532–542.