Monetize your data and collaborate with data partners to train
machine learning models without exposing, sharing or moving
Virtual Data Collaboration
In a data-driven world, enterprises want access to richer additive external data to build and train their AI and ML models. However, Compliance regulations and process overhead make data sharing so complicated that the time to access this external data renders the data less valuable.
CN-Insight leverages new technologies such as ‘Secure Multi-party Computation’ (SMC) and ‘secret shares cryptography’, this allows separate entities or organizations to ‘virtually share data’ for building and training data science, AI and ML models where the data itself is never shared, combined nor accessed, nor can the data values be derived by the other party.
The Virtual Data Collaboration Process
1. Install CN-Insight at each Division or Data Partner and establish secure communication (not for data exchange)
2. Each Party ingests required datasets into their local CN-Insight application
3. Data Schema Extraction Process into local CN-Insight application
4. Metadata is made available to appropriate parties
5. Secure Multi-Party Computation (SMC) leveraging Secret Shares Splitting process initiated
6. One part of the Secret Shares is exchanged between all parties (not the actual data) for each computation on secret shared data
7. Data science feature engineering, model building, and training occurs
8. Based on the agreement between parties, model results are revealed by sending all shares of the model only to the agreed-upon parties