Skip to main content

Working with proxy data in Data Factory

Tuesday, November 9, 2021

Proxy data enables organizations to collaborate. AI Sweden’s Data Factory provides the infrastructure to make it happen.

When asked about the benefits for organizations working in AI Sweden’s Data Factory, Mats Nordlund emphasizes speed and openness. 

“The datasets available in the Data Factory are already cleared. That means there is no need for agreements around data sharing, no risks around compliance, no need for complicated access control setups, etc.” says Mats Nordlund, Head of AI Sweden’s Data Factory.

He continues, “We have built the Data Factory as an open environment where it’s possible to work together on projects that are pre-competitive, sometimes close to basic research. Since nothing done in the Data Factory is protected by IP rights, it makes it possible for competitors to join forces and learn together.”

With open datasets shared by AI Sweden’s partners, sometimes functioning as proxy data for various kinds of applications, the Data Factory paves the way for fast innovation. 

Ebba Josefson Lindqvist, project manager at Data Factory, explains, “A lot is going on around data and data regulation at the moment. There are a lot of uncertainties surrounding GDPR. One option is to work with the legal aspects first and then, when that part is settled, start the actual technical development. But thanks to proxy data, you can work in parallel. While your legal team is looking into the regulatory questions, the developers can start their part by using proxy data.”

What makes the Data Factory such a strong offering?

“The mix,” says Ebba Josefson Lindqvist. “The Data Factory isn’t technology or data. It is technology and data. Also, the cumulative power of all the organizations working in the Data Factory sharing their knowledge openly and freely. From partner to partner, and from AI Sweden to our partners.”


This is AI Sweden’s Data Factory
The Data Factory consists of three pillars:
1. Training environment
A world-class, high-impact scientific center of excellence extending the forefront of distributed solutions to train AI algorithms.

2. Testbed
Where it’s possible to test new frameworks, new hardware, etc. An important part is the Edge Lab, a physical infrastructure for the development of solutions based on federated learning and edge computing.

3. Competence and compliance
Focusing on infrastructure, training, learning and legal questions.


Do you have proxy data to share with AI Sweden’s Data Factory?

If you think your organization may have datasets that could possibly function as good proxy data for different kinds of applications, please get in touch with Mats Nordlund or Ebba Josefson Lindqvist to discuss further. 

Read about how AI Sweden’s partner Zenseact used seabird data from Swedish University of Agricultural Sciences inside the Data Factory.