Distributed Machine Learning for Privacy-preserving Knowledge Gathering and Model Optimization (DIPLOMA)

Partner Call open until: October 2024

The project concerns the development of distributed and/or federated machine-learning methods, in the context of process control, automotive industry and/or consumer electronics.

The Embedded Systems Research Division is actively seeking collaboration for a project focused on developing innovative methods and algorithms for decentralized analysis of online data streams. This project aims to facilitate edge-based prediction of critical technological process outcomes and parameters.

By utilizing distributed and federated learning techniques, we intend to refine a global data-driven model locally, using data situated at the network's edge. This approach eliminates the necessity of global data aggregation. Instead, worker nodes positioned within edge computing infrastructures will update their local models using the data they gather. Subsequently, anonymized model parameters will be shared to facilitate a comprehensive global learning process. Such a strategy not only preserves data privacy by avoiding the need for raw data exchange but also may enhance communication efficiency through adjustable transmission intervals.

Main challenges and topics of investigation include:

• Establish a use-case and the feasibility of distributed and/or federated learning methods.
• Development, deployment, and testing of functional software demonstrators, implementing the developed distributed and/or federated machine-learning method.
• Optimization of on-site model refinement methods to the restrictions of Edge environment, including performance, power, and network limitations.
• Investigating the effectiveness of data privacy preservation using distributed and federated learning approaches.

Important expected results of the project are

• A distributed and/or federated learning method for data-based modelling in the context of process control, automotive industry and/or consumer electronics
• A distributed and/or federated learning system demonstrator, supporting local (on-site) model tuning and global model optimization.

International consortium comprising 1-3 industrial partners.

• 1-3 partners
• 2-3 years runtime
• Universities and academic partners are eligible to participate (special conditions apply)
• PhD students within the project possible - supervised by academic partners

• Attractive cofinancing: SAL covers 50% of the project volume, you only pay 25% in cash and 25% in-kind.
• Competitive advantage: easier and quicker exploitation of upcoming products, by utilization of generated IP
• Minimize Risk: early start of work on technical challenges for an upcoming market
• Innovation: future technologies and exploration of emerging trends
• Partnership: cooperation in a pioneering and efficient eco-system