Abstract
We present the Differentially Private Blockchain-Based Vertical FederalLearning (DP-BBVFL) algorithm that provides verifiability and privacyguarantees for decentralized applications. DP-BBVFL uses a smart contract toaggregate the feature representations, i.e., the embeddings, from clientstransparently. We apply local differential privacy to provide privacy forembeddings stored on a blockchain, hence protecting the original data. Weprovide the first prototype application of differential privacy with blockchainfor vertical federated learning. Our experiments with medical data show thatDP-BBVFL achieves high accuracy with a tradeoff in training time due toon-chain aggregation. This innovative fusion of differential privacy andblockchain technology in DP-BBVFL could herald a new era of collaborative andtrustworthy machine learning applications across several decentralizedapplication domains.