Apheris Unveils Compute Gateway 3.0 - Advancing Global Data Collaboration for Artificial Intelligence
To be effective, AI models have to be trained on large amounts of data - but regulation and confidentiality requirements often put such data out of reach. Apheris enables data owners and consumers to safely form collaborations while staying fully secure, private and in control.
Today marks a significant milestone as Apheris proudly announces the general availability of Apheris Compute Gateway 3.0, a groundbreaking release introducing an end-to-end federated analytics and machine learning solution. Our 3.0 release empowers organizations to collaborate securely on highly sensitive data.
“Our mission is to power the world’s most impactful collaborative data ecosystems. With the new release, Apheris provides the technology making data collaborations secure, private and governed for ML and statistics. We are dedicated to empowering industries to collaborate seamlessly by eliminating hurdles to computational access and simplifying the complexities, risks, and obstacles that have historically made collaboration nearly unfeasible. I’m very proud of the whole team and the collaborative spirit we feel with our investors and customers.” said Robin Röhm, CEO and co-founder Apheris
What's New in Apheris Compute Gateway 3.0?
Enhanced Compute Gateway: At the core of Apheris's offering, the Compute Gateway has undergone significant upgrades. This lightweight software agent, deployable anywhere, operates within a Data Custodian’s environment to oversee federated computations at the algorithmic level, ensuring sensitive data never moves. The Compute Gateway is an integral part of the federated architecture, returning only privacy-preserving results such as model weights and aggregates while maintaining raw data confidentiality.
Integration of NVIDIA FLARE: We have integrated NVIDIA FLARE, renowned for its reliability and scalability in high-risk projects, as a federation backbone. ML Engineers can easily create custom models in standard ML tooling as well as access out-of-the box model implementations and other integrations.
Governance Portal: The newly improved user interface, the Governance Portal, allows Data Custodians comprehensive control over their datasets and computational policies. Features include registering and managing datasets, controlling computations through Asset Policy configurations, and a detailed review and approval process for compute jobs, ensuring systematic auditability and adherence to strict privacy and security standards.
Streamlined Workflow: The Compute Gateway 3.0’s workflow is structured into three straightforward steps: Prepare, Agree, and Use. This setup allows for a computational handshake between data custodians and ML engineers before execution, ensuring compliance with the highest standards of data privacy and security.
A Computational Handshake
Apheris 3.0 represents a significant leap forward in operationalizing internal policies and regulatory requirements, ensuring a secure, privacy-preserving environment for data collaboration. By keeping data localized and bringing computations to the data, Apheris facilitates a secure exchange that meets the most stringent data privacy requirements.
This release not only enhances our product's functionality but also reaffirms Apheris's commitment to advancing secure, federated computing, establishing us as a leader in the field of governed and privacy-preserving federated learning.
“Safeguarding sensitive hospital data is critical to ensure patient privacy and data security,” said Holger Roth, Principal Federated Learning Scientist, NVIDIA. “Integrating NVIDIA FLARE with the Compute Gateway 3.0’s governance, security and privacy capabilities unlocks new possibilities to make a difference in healthcare and beyond.”
For more information about Apheris 3.0 and to see how our solutions can empower your data-driven initiatives, visit our website or contact our press office.
Press Contact:
apheris AI GmbH
Kolonnenstr. 8
10827 Berlin
E-Mail: info@apheris.com
www.apheris.com
