PhD Fellow in Machine Learning for graphs and time series data

The position

A PhD position in machine learning for graphs and time series data is available at the Department of Mathematics and Statistics, Faculty of Science and Technology.

The position is financed by the RCN (Research Council of Norway) funded project “RELAY: Relational Deep Learning for Energy Analytics”. The main focus of the research will be to advance the state-of-the-art of machine learning models for graphs and time series, such as Graph Neural Networks, State-Space Models, Recurrent Neural Networks, and Reservoir Computing.The main field of application will be Energy Analytics, but the developed methodologies will be general purpose and applicable also to other fields.

The position is for a period of three years. The objective of the position is to complete research training to the level of a doctoral degree. Admission to the PhD programme is a prerequisite for employment, and the programme period starts on commencement of the position.

The workplace is at UiT in Tromsø. The candidate must be able to start in the position in Tromsø within a reasonable time, within 6 months after receiving the offer.

The project will be framed within the activities of ARC -  Arctic Centre for Sustainable Energy and will be carried out in collaboration with three industrial partners in Northern Norway (Ishavkraft, Finnmarkkfrat, ELMEA) and the following universities:

• Sapienza University of Rome, Italy 
• University of Pisa, Italy 
• Universita’ della Svizzera italiana, Switzerland
• University of South-Eastern Norway, Norway

Field of research and the role of the PhD Fellow

The project will mainly focus on basic research in machine learning models for time series and graphs. One of the main goal will be to push the boundaries in the field of relational deep learning by:

• Creating innovative tools (novel architectures, training strategies, etc,…) for processing spatio-temporal data, e.g., multiple time series whose relationships are described by a graph.
• Enhance the capabilities of existing deep-learning models by gaining theoretical and practical insights.

The research activities are divided into five Work Packages.

Randomized architectures to handle big data

The goal is to generate informative spatio-temporal representations without the need for traditional training or supervision. By using suitable randomized techniques, it will be possible to improve the scalability of large spatio-temporal models without compromising their performance.

Multi-scale representations with graph coarsening

The objective is to create multi-scale representations with graph pooling (a procedure to generate smaller graphs that carry the original information) to manage the complexity of spatio-temporal models. New graph pooling techniques suitable for spatio-temporal data will be developed and used to enhance the performance on tasks of interest (e.g., forecasting), to identify underlying factors in the system, handle missing data, and integrate multi-resolution data from various sources.

Uncertainty quantification 

The goal is to model the uncertainty in deep learning models for spatio-temporal data by means of Bayesian and frequentist approaches. This involves modifying existing deterministic models to include probabilistic components and extending techniques for generating confidence intervals to spatio-temporal data by addressing the challenge of capturing both spatial and temporal dependencies.

Interpretability

Develop new techniques that allow for a human-understandable explanation of the model’s output, thus aiding in systematic pattern discovery within the data. Due to the complex and irregular structure of spatio-temporal data, existing interpretability tools are not suitable. The goal is to extend current approaches to spatio-temporal models and to develop new ones based on probabilistic frameworks.

Applications

The methodologies developed in the project will be mainly applied to analyze energy systems. These systems present complexities that traditional models frequently struggle to address, and the application of advanced relational deep learning techniques aims to provide more effective solutions. The main applications in energy analytics will be: enhanced load forecasting, dynamic power flow optimization, and localization of energy faults on the grid.

While energy analytics will be the main field of application, since  the methodologies in work packages 1-4 are general purpose, other application areas can also be considered.

Contact

For further information about the position, please contact Associate Professor Filippo Maria Bianchi: 

email: filippo.m.bianchi@uit.no

For more information about the project and the partners, please refer to the project website: https://en.uit.no/project/relay

Qualifications

This position requires a master’s degree or equivalent in Computer Science, Electrical Engineering, Applied Mathematics, or a closely related field, with a focus on areas such as machine learning, data science, or computational statistics. If you are near completion of your master’s degree, you may still apply.

The applicant must have:

• A demonstrated interest or background in Artificial Intelligence/Machine Learning, particularly with a vision to innovate and contribute significantly to research in deep learning for time series and graph data analysis.
• Proficiency in Python is mandatory. Familiarity with the Linux operating system and with common programming tools and environments (Git, SSH, Anaconda, VSCode/Pycharm, etc…) is also required.
• A solid understanding of deep learning and experience with Pytorch and common data analysis libraries such as Pandas, scikit-learn, Seaborn, etc...
• A proactive approach to learning and implementing new coding practices, with the ability to adapt to and utilize new frameworks and languages as dictated by evolving project demands.
• Commitment to staying informed about cutting-edge developments in deep learning, time series analysis, and graph data processing, and the ability to cultivate a robust research methodology.
• Excellent written and verbal communication abilities, with the skill to articulate complex concepts clearly and effectively.
• Willingness to communicate research results through blogs and social media and to participate in open source projects.

Applicants must document fluency of in English and be able to work in an international environment. Nordic applicants can document their English capabilities by attaching their high school diploma. 

In the assessment, the emphasis is on the applicant's potential to complete a research education based on the master's thesis or equivalent, and any other scientific work. In addition, other experience of significance for the completion of the doctoral programme may be given consideration.

We will also emphasize motivation and personal suitability for the position. We are looking for candidates who:

• Have good collaboration skills
• Have good communication and interaction with colleagues and students
• Wants to contribute to a good working environment

As many people as possible should have the opportunity to undertake organized research training. If you already hold a PhD or have equivalent competence, we will not appoint you to this position.

Admission to the PhD programme 

For employment in the PhD position, you must be qualified for admission to the PhD programme at the Faculty of Science and Technology and participate in organized doctoral studies within the employment period.

Admission normally requires:  

A bachelor's degree of 180 ECTS and a master's degree, or an integrated master's degree.

UiT normally accepts higher education from countries that are part of the Lisbon Recognition Convention.

In order to gain admission to the programme, the applicant must have a grade point average of C or better for the master’s degree and for relevant subjects of the bachelor’s degree. A more detailed description of admission requirements can be found here.

Applicants with a foreign education will be subjected to an evaluation of whether the educational background is equal to Norwegian higher education, following national guidelines from NOKUT. Depending on which country the education is from, one or two additional years of university education may be required to fulfil admission requirements, e.g. a 4-year bachelor's degree and a 2-year master's degree. 

If you are employed in the position, you will be provisionally admitted to the PhD programme. Application for final admission must be submitted no later than two months after taking up the position.

Inclusion and diversity

UiT The Arctic University of Norway is working actively to promote equality, gender balance and diversity among employees and students, and to create an inclusive and safe working environment. We believe that inclusion and diversity are a strength, and we want employees with different competencies, professional experience, life experience and perspectives.

If you have a disability, a gap in your CV or immigrant background, we encourage you to tick the box for this in your application. If there are qualified applicants, we invite at least one in each group for an interview. If you get the job, we will adapt the working conditions if you need it. Apart from selecting the right candidates, we will only use the information for anonymous statistics.

We offer

• Involvement in an interesting research project 
• Good career opportunities 
• A good academic environment with dedicated colleagues  
• Flexible working hours and a state collective pay agreement  
• Pension scheme through the state pension fund 
• PhD Fellows are normally given a salary of 532 200 NOK/year with a 3% yearly increase

Norwegian health policy aims to ensure that everyone, irrespective of their personal finances and where they live, has access to good health and care services of equal standard. As an employee you will become member of the National Insurance Scheme which also include health care services.

More practical information about working and living in Norway can be found here: https://uit.no/staffmobility

Application 

Your application must include: 

• Cover letter explaining your motivation and research interests
• CV
• Diploma for bachelor's and master's degree
• Transcript of grades/academic record for bachelor's and master's degree
• Explanation of the grading system for foreign education (Diploma Supplement if available)
• Documentation of English proficiency
• Three references with contact information, including the master thesis supervisor
• Master’s thesis, and any other academic works

Qualification with a master’s degree is required before commencement in the position. If you are near completion of your master’s degree, you may still apply and submit a draft version of the thesis and a statement from your supervisor or institution indicating when the degree will be obtained. You must still submit your transcripts for the master’s degree with your application.

All documentation to be considered must be in a Scandinavian language or English. Diplomas and transcripts must also be submitted in the original language, if not in English or Scandinavian. If English proficiency is not documented in the application, it must be documented before starting in the position. We only accept applications and documentation sent via Jobbnorge within the application deadline. 

General information 

The appointment is made in accordance with State regulations and guidelines at UiT. At our website, you will find more information for applicants. 

Remuneration for the position of PhD Fellow is in accordance with the State salary scale code 1017. A compulsory contribution of 2 % to the Norwegian Public Service Pension Fund will be deducted. You will become a member of the Norwegian Public Service Pension Fund, which gives you many benefits in addition to a lifelong pension: You may be entitled to financial support if you become ill or disabled, your family may be entitled to financial support when you die, you become insured against occupational injury or occupational disease, and you can get good terms on a mortgage. Read more about your employee benefits at: spk.no.

A shorter period of appointment may be decided when the PhD Fellow has already completed parts of their research training programme or when the appointment is based on a previous qualifying position PhD Fellow, research assistant, or the like in such a way that the total time used for research training amounts to three years. 

We process personal data given in an application or CV in accordance with the Personal Data Act (Offentleglova). According to the Personal Data Act information about the applicant may be included in the public applicant list, also in cases where the applicant has requested non-disclosure. You will receive advance notification in the event of such publication, if you have requested non-disclosure.