Data Science

Build AI Models on Private Data

Build models on valuable, private data that was previously inaccessible. Data is unreasonably effective in improving AI model accuracy and Ocean Protocol’s marketplaces and Compute-to-Data help data scientists & AI practitioners get more data, including the most elusive of all — private data.

AI practitioners and data scientists can access private data via remote execution of AI training algorithms against the data, with computation provided by the data owner. AI practitioners benefit from increased accuracy of their AI models, or perhaps even the ability to model things that were not previously possible. Data owners retain privacy and control of their data.

The blockchain-based system gives an audit trail on data being bought and sold. Compute-to-Data gives proof that algorithms are properly executed, so that AI practitioners can be confident in the results.

How It Works

A typical data science flow usually happens in an Ocean data marketplace where AI practitioners find data to run their algorithms on. Flows without marketplaces are also possible, for example, in a Jupyter notebook.

The AI practitioner visits an Ocean-powered data marketplace, proposes to purchase data, and submits an AI model-training algorithm.

Data Science Resources