10 Postdoctoral Fellowship at Umea University, Sweden

Trustworthiness is a crucial consideration in modern AI decision-making systems. As a fundamental aspect of human interaction, AI trustworthiness can be improved through transparency and explainability in robotic systems. This project aims to identify factors that may undermine the trustworthiness of data and models used in human-robot interaction, and further investigates common and specific trustworthiness issues related to fairness and safety in human-robot interaction applications. An AI trustworthiness model will be developed and validated ensuring that both data and models of human interaction are robust, especially in the selected industry use cases. 

The postdoc position is linked to the research group Deep Data Mining, which focuses on fusing data science and artificial intelligence and developing AI trustworthiness (e.g., fairness, privacy) models.   The project is part of the EU project XSCAVE whose ambition lies on large scale deployment of autonomous heavy mobile machines in earthmoving, forestry and urban logistics industries. The XSCAVE consortium involves eleven partners from all over Europe.  This offers excellent opportunities for international exchange and collaboration with leading research groups and companies in the field of AI robotics, large language modelling, simulation, mobile robotics and offroad heavy equipment.

In modern software systems and the organizations that run them, a substantial part of day-to-day decisions with critical impact on individual humans and society at large is either fully automated or heavily relies on automatically provided decision support. While machine learning approaches become increasingly prevalent in this context, the cores of the systems’ reasoning engines typically remain ‘symbolic’ (knowledge-based). These symbolic or neuro-symbolic software systems are of high practical complexity, which makes them difficult to reason about, for example when assessing which meta-level changes lead to the intended real-world impact. Working towards solving this problem, the theme of the broader project is ‘Automating Reasoning about Automated Reasoning’. From a fundamental research perspective, this relates to the automation of meta-reasoning pertaining to general-purpose reasoning methods. Here, the intended focus is on reasoning approaches that can revise their conclusions in face of new evidence/knowledge. The plan is to cover both nascent approaches that are considered promising facilitators of the fusion of subsymbolic, symbolic, and human intelligence, as well as industry-scale reasoning systems. Instances of the former class are different variants of formal argumentation, whereas in the latter case, the focus can lie on mainstream rule languages that are deployed in large-scale IT systems.

The causes of brain tumours (gliomas) and pancreatic cancers are not well understood, and there are no specific blood tests to discover glioma nor pancreatic cancer. We believe that if we understand how tumours arise, we can also find new opportunities for early diagnosis or new targets for treatment. More information about our publications and the research group can be found here.

The project involves analysing imaging from CT scans and MRI, together with -omics data such as quantitative metabolomics performed on blood samples, and clinical information, using advanced bioinformatics. The tasks include detailed bioinformatic analyses of imaging data and blood sample -omics data from the patients that later developed glioma or pancreatic cancer. Tasks also include literature searches, compiling results and writing manuscripts. Some administrative tasks, such as convening group meetings may be included, as we usually rotate this in the group.

The successful candidate will perform research within the externally funded project “Innovative statistical and machine learning methods for comparing performance and outcome in register data studies”, which aims to develop, evaluate, and implement innovative statistical methods, including modern machine learning methods, to draw inferences from register data. The proposed methods will be carefully evaluated with data from educational and health registers but the developed methods are not restricted to these applications but are generic and can be used with other large-scale registers. The successful candidate is expected to plan and perform independent research but is also expected to collaborate with researchers at the department and within other subject areas. The position may also, by agreement, include teaching at undergraduate and graduate levels in statistics up to 20% of full time.

The focus of the research is to understand the molecular mechanisms of DNA interstrand crosslink repair and the Fanconi Anemia DNA repair pathway, which plays a key role in genome stability in humans. The postdoctor will study our recent findings, using biochemistry, live-cell imaging, CRISPR/Cas9-mediated genome engineering and cryo-EM.

The primary project aims to investigate how humoral immunity develops and is maintained in mucosal tissues and blood during respiratory viral and bacterial infections. The study is based on longitudinal patient cohorts, with samples of blood and saliva collected during and after infection or vaccination. Our overarching goal is to map the mechanisms that lead to long-lasting antibody-based immunity in mucosal tissues, with a focus on IgA, and contribute knowledge essential for the development of future mucosal vaccines and antibody therapies. The project is part of a larger collaborative effort between Umeå University, Karolinska Institute, the Royal Institute of Technology, and Linköping University, funded by the Knut and Alice Wallenberg Foundation.

The successful candidate will contribute to the Autonomous Distributed Systems (ADS) Lab within the Department of Computing Science. The ADS Lab is an internationally leading research group with a focus from distributed AI to autonomous resource management and modern. The Lab currently comprises over 20 experienced and world-leading research colleagues from more than 10 different countries. Collaborations are performed with industries like Google, IBM, Intel, Red Hat, and Ericsson, as well as universities and institutes such as Princeton University, University of Massachusetts Amherst, Carnegie Mellon University, Princeton University, Lawrence Berkeley Lab, Nanyang Technical University in Singapore, Uppsala University, Lund University, Universidad Complutense de Madrid, Leeds University, Barcelona Supercomputer Center, TU Vienna, TU Delft, and many more.  For more information, see https://www.cloudresearch.org

The human gut microbiome plays an important role in health and disease. A small proportion of the species that inhabit our gut belong to the Archaeal domain of life. These species are poorly characterized at the molecular level, and thus their function in the microbiome is not completely understood.

In this project, you will use functional imaging and proteomics approaches to provide hints of protein function, and you will follow up those hints using biochemistry and molecular biology approaches.

The project aims to determine structural dynamics of membrane proteins using biophysical methods, such as time-resolved synchrotron X-ray solution scattering, cryo-electron microscopy, and computer simulations. The specific focus will be on the regulation of ATP-dependent ion transport. The research will be conducted in collaboration with the University of Michigan and will include research visits to synchrotrons in Sweden and France.