Scaleout Systems: Advancing Distributed Machine Learning
Scaleout Systems, a leading player in distributed machine learning, drives innovation with a focus on data integrity and security. Through participation in NATO’s DIANA programme and the development of technologies such as FEDn, the company addresses the challenges of edge computing. Recently, Sweden’s Defence Minister Pål Jonson visited Scaleout to discuss the company’s latest projects and their impact on both commercial and defence applications.
The team behind Scaleout Systems, leaders in federated learning. Photo: private
Scaleout, founded in 2017 as a spin-off from Uppsala University's Department of Information Technology, develops solutions for distributed machine learning with a focus on data privacy and security. Our team of 15, including five PhDs, specializes in federated learning technologies that enable machine learning across distributed data sources without centralizing sensitive information. With strong backing from our investors (Navigare, Fairpoint, Almi Invest, and UU Invest), we are driving innovation to make AI adoption more privacy aware and reliable for society.
Our framework, FEDn, addresses these challenges through federated learning - an approach that enables model training across distributed devices while keeping data local. This methodology is particularly relevant for applications requiring real-time processing capabilities and strong data privacy guarantees.
Addressing Edge Computing Challenges
The rapid growth of edge computing presents new challenges for machine learning implementations. Organizations across sectors are generating substantial amounts of data at the edge through various devices and sensors. Traditional centralized approaches to machine learning, which require data aggregation at a central location, face significant limitations in terms of latency, network efficiency, and data privacy.
Scaleout Joins NATO's DIANA Programme
Scaleout has been selected to participate in NATO's Defence Innovation Accelerator for the North Atlantic (DIANA) Challenge Programme - a NATO organization with a mission to accelerate dual-use innovation across the Alliance. Through our FEDAIR project (Federated Aerial Intelligence for Recon), we are investigating the application of federated learning in networks of distributed drones and sensors. The project focuses on developing efficient mechanisms for secure, decentralized machine learning in scenarios where data sovereignty and network efficiency are crucial.
The research addresses several key technical challenges:
- Secure computation and data exchange at the edge
- Network resource optimization
- System resilience and security
- Adaptability across diverse operational environments
Recently, Swedish Defense Minister Pål Jonson visited our office to discuss the DIANA project and its potential impact. While our primary research focus remains on commercial applications, our participation in DIANA enables exploration of the technology's broader implications for security and defense applications.
Sweden’s Defence Minister Pål Jonson visited Scaleout to discuss their work on federated learning within the NATO DIANA programme, focusing on secure and adaptable ML technology for defence applications.
Salman Toor