OPENMINED + APHERIS AI PARTNERSHIP FOR PYTORCH MOBILE FEDERATED LEARNING
This blog post is authored by Andrew Trask, Creator & Leader of OpenMined — PhD Student at the University of Oxford.
Our mission at OpenMined is to make the world more privacy-preserving by lowering the barrier-to-entry to privacy-preserving technologies through free, open-source software and education.
Today, we’re very excited to announce our Use Case partnership with apheris AI to deploy the very first open-source system for private federated learning on server, web, and mobile at scale.
In 2020, our goal is to solve real privacy problems in production and lower the barrier-to-entry for others to do so by providing both software and education. However, it’s impossible to truly know whether we are creating the right software without actively seeking to use it to solve someone’s problem.
This is why we’ve stated 7 deployments in our Roadmap for 2020, under the categories Enterprise, Smartphone, and Open Research platforms. Each deployment has a Use Case Partner and a Funding Partner — the Use Case partner is the organization within which we will actually be deploying our software. For smartphone applications, this is the organization which has a smartphone app; for enterprise applications, this is the enterprise.
Our Federated Learning with Secure Aggregation across Web & Mobile use case, funded by PyTorch, will be the very first open-source system for private federated learning on server, web, and mobile. Today, we’re very excited to announce our Use Case partnership with apheris AI to deploy this system at scale.
The mission of apheris AI is to empower companies to unlock value from distributed datasets while preserving data privacy, and our goal in this partnership is to help them deploy this technology to millions of devices in the coming years with the assistance of our libraries and community.
Specifically, we will take on a role as a supporting team behind Apheris’s efforts when delivering federated learning solutions on mobile and edge devices. We will meet with them regularly, improve our codebase based on their requests, introduce them to potential employees, contractors, partners, and customers within our community, and of course, raise awareness of achievements we make together.
We are actively looking for Use Case Partners for our Cross-Organization Model Evaluation, Cross-Organization Federated Learning, Consumer-Selected Third-party On-Device ML, and Scalable, Debuggable Compute use cases listed on our Roadmap. If you are interested in applying to be a Use Case Partner for this or another use case — you can email partnerships@openmined.org.
