Masters Thesis: Continuous, Active Federated Learning for Data Streams

AI Sweden • Gothenburg, NE

2d • Hybrid

About The Position

As Sweden's national center for applied AI, AI Sweden is seeking an ambitious Master’s thesis student to work on a unique mix of AI research topics for an automotive use case. This project pioneers a new approach to real-time federated learning, enabling continuous AI adaptation while fully complying with increasing regulations. The increasing use of edge devices in vehicles has led to massive data generation, creating challenges for real-time AI adaptation, data privacy, and security. Traditional Federated Learning (FL) methods are unsuitable for real-time data-streaming environments. This project aims to advance safe automated vehicles by enhancing AI-driven perception, situational awareness, and decision-making while mitigating privacy breaches and national security risks. We introduce continuous, active federated learning for data streams to equip FL to work with high amounts of constantly incoming real-world data at the edge. The results will inform regulatory discussions on AI-driven traffic safety, supporting the development of policies for secure, automated vehicles. Our proposed approach enables data-efficient continuous model adaptation without requiring stored data, reducing cybersecurity risks and misuse of collected images or sensitive location data. The overall project is a joint collaboration between AI Sweden (coordinator), Zenseact (autonomous driving use case), Scaleout Systems (FL frameworks), and Ekkono Solutions (edge learning).

Requirements

• Ongoing Master’s studies in Computer Science, Data Science, Engineering Physics, Complex Adaptive Systems, Machine Learning, or similar.
• Comfortable with Python and deep learning.
• Comfortable with the idea that an experiment might tell you something you did not expect.

Nice To Haves

• Prior experience with active learning, continuous / incremental learning, self-supervised learning is useful but not required; an interest to pursue these topics is.

Responsibilities

• Design controlled experiments.
• Run experiments on open-weight reasoning models.
• Measure results carefully.
• Focus on one or combine several of the following research topics: Single-Pass Active Learning, Self-Supervised Learning for Data Streams, Managing catastrophic forgetting, Federated learning for data streams.
• Decide the exact research question and methods together with the student, depending on the student's background and interests, previous work, and practical feasibility.

Benefits

• Access to substantial compute.
• Working on questions that are genuinely open, in a frontier area of AI research.
• Opportunity for personal development and achievements.
• A place to grow.

What This Job Offers

• Job Type: Full-time
• Career Level: Entry Level