Gemsen is a machine learning platform that drastically speeds up data exploration and analysis, while completely preserving privacy of the underlying datasets; we do this by converting datasets into powerful abstractions, called GEMs, that massively compress the data while capturing the mathematical relationships between data features. Models build on GEMs are as accurate as if built on the underlying data, but are 1000x faster to train, and offer unbreakable privacy protection for collaborative computing on sensitive data. These capabilities are useful for financial institutions computing on large volumes of data (including real time data), and those looking to collaborate on sensitive data between various institutions.
