PhD student in Computing Science with focus on federated learning - Befattningen är under tillsättning eller redan tillsatt, och det är inte längre möjligt att lämna in en ansökan

To our department, characterized by world-leading research in several scientific fields and a multitude of educations ranked highly in international comparison, we now look for a doctoral student in robust and scalable federated learning for robotic manipulation in edge-cloud settings.

The Department of Computing science has been growing rapidly in recent years where focus on an inclusive and bottom-up driven environment are key elements in our sustainable growth. The 50 Doctoral students within the department consists of a diverse group from different nationalities, background and fields. If you work as a Doctoral student with us you receive the benefits of support in career development, networking, administrative and technical support functions along with good employment conditions. See more information at:
https://www.umu.se/en/department-of-computing-science/

Is this interesting for you? Welcome with your application before May 2, 2022.

Project description

Federated learning (FL) poses multiple challenges when applied to distributed, real-time robotic manipulation in the cloud. Multi-centric data heterogeneity observed during sensing, actuation, control and manipulation is a prime concern because data in classical FL settings is typically assumed to be independent and identically distributed (iid). In contrast, data from distributed robotic nodes operating under variations in environment, kinematics and sensing modality is not necessarily iid and raises questions concerning global model convergence for performance guarantees. This project aims to address the following challenges: (i) Developing high-quality local and global representation learning models cause multiple issues, including inconsistency of representation spaces, misalignment of representations, and incomplete representations. (ii) Achieving faster model adaptation through continual or lifelong learning, (iii) Preventing inter-nodes interference and inter-nodes knowledge transfer for irrelevant knowledge assimilation across models, (iii) Selective relearning of policies for unseen environments in cloud robotics systems.

This project investigates and develops federated contrastive, transformer, and deep learning methods for learning suitable representation and task-based robot manipulation policies at scale. Each robotic node will train on varied representations to perform multiple downstream tasks, e.g., faster model and unseen environment adaption. These uniform representations motivate to develop continual, peer-assisted, deep, and imitation federated learning methods for fusing knowledge from the cloud to edge nodes to achieve high precision performance and low-cost policy training. 

The position is aimed for doctoral studies in Computing Science within the autonomous distributed systems lab, but collaboration with researchers in, e.g., machine learning, mathematical statistics, optimization, cloud robotics, deep learning or artificial intelligence is expected. (For further information, see www.cloudresearch.org).

The position is funded by The Knut and Alice Wallenberg Foundation through The Wallenberg AI, Autonomous Systems and Software Program (WASP) within a new ambitious NEST project “Intelligent Cloud Robotics for Real-Time Manipulation at Scale” in collaboration with KTH, Stockholm and Lund University, Lund. WASP is Sweden’s largest individual research program ever, a major national initiative for strategically motivated basic research, education and faculty recruitment. The program addresses research on artificial intelligence and autonomous systems acting in collaboration with humans, adapting to their environment through sensors, information and knowledge, and forming intelligent systems-of-systems. The vision of WASP is excellent research and competence in artificial intelligence, autonomous systems and software for the benefit of Swedish industry. Read more: https://wasp-sweden.org/

The graduate school within WASP is dedicated to provide the skills needed to analyze, develop, and contribute to the interdisciplinary area of artificial intelligence, autonomous systems and software. Through an ambitious program with research visits, partner universities, and visiting lecturers, the graduate school actively supports forming a strong multi-disciplinary and international professional network between PhD-students, researchers and industry. Read more: https://wasp-sweden.org/graduate-school

Admission requirements

The general admission requirements for doctoral studies are a second- cycle level degree, or completed course requirements of at least 240 ECTS credits, of which at least 60 ECTS credits are at second-cycle level, or have an equivalent education from abroad, or equivalent qualifications. To fulfil the specific entry requirements for doctoral studies in computing science, the applicant is required to have completed at least 90 ECTS credits in computing science. Applicants who otherwise have acquired skills that are deemed equivalent are also eligible.

Documented knowledge and a solid background in machine learning and robotics or machine learning and distributed systems is a requirement. The research is to a large extent interdisciplinary, and a broad competence profile and experience from other relevant areas (such as machine learning, distributed learning, adversarial learning, deep learning, IoT, discrete optimization, and statistical methods) is considered a merit. Good skills in both spoken and written English are a requirement for the position. 

Important personal qualities are, besides creativity and a curious mind, the ability to work both independently and in a group and experience in scientific interaction with researchers from other disciplines and in other countries.

About the position

The position provides you with the opportunity to pursue PhD studies in Computing Science for four years, with the goal of achieving the degree of Doctor in Computing Science. While the position is mainly devoted to PhD studies (at least 80% of the time), it may include up to 20% department service (usually teaching). If so, the total time for the position is extended accordingly, resulting in a maximum of five years.

The procedure for recruitment for the position is in accordance with the Higher Education Ordinance (chapter 12, 2§) and the decision regarding the position cannot be appealed.

The expected starting date is June 2022 or as otherwise agreed.

Application

Applications must be submitted electronically using the e-recruitment system of Umeå University.

A complete application should contain the following documents:

• A cover letter including a description of your research interests, your reasons to apply for the position, and your contact information
• A curriculum vitae
• Reprints / copies of completed BSc and/or MSc theses and other relevant publications, if any
• Verified copies of degree certificates, including documentation of completed academic courses and obtained grades
• Documentation and description of other relevant experiences or competences.

The application must be written in English or Swedish. Attached documents must be in pdf format. Applications must be submitted electronically using the e-recruitment system of Umeå University, and be received no later than 15 April 2022.

The Department of Computing Science values gender diversity, and therefore particularly encourages women and those outside the gender binary to apply for the position. The focus of the research in the Autonomous Distributed Systems Lab is to design, develop, deploy distributed learning algorithms for (autonomous) resource and application management and vice-versa for different types of IoT, IIoT, edge clouds and large-scale complex systems.

For additional information, please contact Assist. Prof. Monowar Bhuyan (monowar@cs.umu.se) or Professor Erik Elmroth (elmroth@cs.umu.se).

We look forward to receiving your application!