Since the EU passed its data protection law (GDPR) in 2018, privacy has gone from being generally desired to absolutely necessary for data-collecting companies. While a boon for compliance, privacy regulations, and the broader societal push for data accountability, has created significant costs, expansive risks, and constant headaches for companies. As over 75% of the world’s population is expected to be covered by data privacy regulations according to Gartner, the headache will not soon go away. And that’s a good thing, as the “privacy problem” is not just a regulatory painpoint but a unique opportunity to change the faulty assumptions that undergird data science in the enterprise.
Efficient and effective
Facing new limitations on data collection, organizations are also starting to rethink a paradigm of data duplication, movement, and refresh that often failed to translate bid data into business value. Many are, therefore, beginning to see that the existing way data flows in organizations is costlier, takes longer, and creates risks by default. This is where privacy, something over 65% of consumers view as critical for brand trust, can be the onus for significant improvements to the efficiency and effectiveness of data efforts. At Devron, we have seen first-hand how privacy can empower efficiency rather than just restrain it in critical contexts across government, finance, and healthcare.
So how can you improve your operation without exposing all that data? It's simple: federated machine learning…or at least that’s part of the technology that is under the hood. Devron allows you to build powerful predictive models without needing to centralize and combine the data and, more importantly in some cases, without exposing any privacy! Devron’s platform allows you to automatically deploy cloud infrastructure to train algorithms on data in public clouds, SQL databases, and legacy systems. Even without owning the data, accessing siloed sets is as simple as inviting the owner to create an account on Devron’s web app and contributing the location of their data to the project. The more data sets contributed, the faster the training and the more powerful the model!
Keeping data at the point of collection
But beyond the cost savings and privacy protection of keeping data at the point of collection, ensuring the safety of sensitive datasets is our top priority to enable confidence in data sharing and collaborative model training. That’s why Devron uses a combination of technologies deployed and trusted in the most sensitive government use cases to counter adversarial machine learning and prevent malicious actors from gaining access to your data. Further, at the local data sites, we use the advanced privacy preserving techniques to guarantee that sensitive data stays at home even when the model is applied elsewhere.
Therefore, our platform enables you to collectively derive insights from data across the enterprise, country, and world without sharing a single row of the underlying data. By increasing the quantity of accessible data, you can reduce the time to value, improve model training and accelerate model deployment for increased generalizability while preserving the privacy of data sources and methods in line with regulations like HIPAA and GDPR. This means that privacy is built-in, rather than just added on top, to ensure the safety and compliance of data sharing while making data science teams more efficient and effective in the process.
