facebook tracking

Master Thesis - Robust federated learning on heterogeneous edge devices

Background

Today, machine learning (ML) applications are demanding data and computing resources to fulfill our needs. However, these resources are typically dispersed over various devices, regions, or organizations and often under various data-sharing regulations and policies. This makes direct data sharing complex and subsequently training ML algorithms difficult.

Federated learning (FL) is a promising technique that efficiently utilizes distributed data and computing resources. It facilitates learning shared ML models among multiple actors while complying with regulations and ensuring data privacy. FL eradicates the need to transfer vast amounts of data out of the local devices, resulting in optimized bandwidth and storage use while the acquired insights from data remain intact. FL could improve ML applications in a shorter time by utilizing more diverse data collected from various edge devices. However, the deployment of FL on edge devices like autonomous cars, in real-world setups is not well-studied. Moreover, edge devices including autonomous cars are usually heterogeneous from different aspects such as their software stack, sensor configuration, network connectivity, compute, and memory capabilities.

Hence, it is valuable to study the resulting impact of devices’ heterogeneity on FL systems to achieve desirable performance and efficient training/communication time while maximizing the utilization of insights from all cars.

Project Description

In this master thesis project, you will focus on:

  • Assessing FL algorithms that incorporate the heterogeneity assumption.
  • Studying the impact of the software stack (e.g., different ML libraries, model precisions) and sensor configuration (e.g., sensor degradation, upgrades) heterogeneity on FL systems through running experiments on various edge devices or the cloud using the real-world dataset provided by Zenseact. We will utilize open-source FL frameworks during the project.
  • Improving solutions to have robust FL for heterogeneous autonomous vehicles.

Qualifications

We are looking for two highly motivated students with a good general background in machine learning and computer vision. The following skills would be precious:

  • Deep learning
  • Federated learning
  • Python programming
  • Comfortable working with Docker, Kubernetes, edge devices, and complex systems

Further information

Please send in individual applications with CV, motivational letter, and grade transcripts.

Planned start: January 2022, with some flexibility.

Final application date: 15 of November 2021

Duration: 30 ECTS

For questions regarding the project, please contact: mina.alibeigi@zenseact.com

Additional information

  • Remote status

    Flexible remote

Or, know someone who would be a perfect fit? Let them know!

Gothenburg, Sweden

Lindholmspiren 2
417 56 Göteborg Directions View page

Making safe and intelligent mobility real.

At Zenseact, we lead the global movement of crafting tomorrow's mobility with the software platform of choice. Our mission is to “Make safe and intelligent mobility real, for everyone, everywhere”. This statement marks our conviction and dedication to bring autonomous driving out on the streets for real and is at the center of everything we do.

We could not dream of achieving this without our great teams of very talented people. We are on this journey together and our agile way of working is reflected throughout our entire organization; it is part of our culture and how we work, develop and grow together.

Teamtailor

Applicant tracking system by Teamtailor