Ensuring Data Integrity with Cryptographic Hashing and the Ethereum Blockchain

A method is presented for ensuring data integrity in machine learning by cryptographically hashing a dataset and storing its hash immutably on the Ethereum blockchain. This technique provides a permanent and verifiable record of a dataset's state, which is crucial for distributed teams and ensuring reproducibility. To avoid transaction fees, the process leverages the Sepolia testnet, allowing for free storage of the data's fingerprint in a transaction's input data field. The implementation uses Python libraries like `web3.py` and `hashlib` to create the hash, sign a transaction, and broadcast it to the network. Finally, a metadata record linking the dataset to its on-chain hash is created to complete the versioning process.