UiT The Arctic University of Norway, Faculty of Science and Technology / Department of Mathematics and Statistics

Position ID: 2675-PHD2 [#20117]
Position Title: Postdoctoral Research Fellow in Statistics and Deep Learning
Position Type: Postdoctoral
Position Location: TromsøNordland 9037, Norway
Subject Area: Statistics and Deep Learning
Application Deadline: 2022/08/15 11:59PMhelp popup (posted 2022/06/07, listed until 2022/08/15)
Position Description:    

The position 

At the Department of Mathematics and Statistics (DMS) a position is available for a Postdoctoral Research Fellow. This position is financed by the IKT+ program at the Norwegian Research Council through the project Transforming ocean surveying by the power of DL and statistical methods. The successful candidate will focus on analysis of heterogeneous data and Bayesian Deep Learning. The position is affiliated with the Machine Learning (ML) group and the candidate will collaborate with Department of Geosciences (DG) at UiT and the two companies Multiconsult and Argeo. The candidate will collaborate closely with researchers at the ML group, DMS, DG and the two companies. In addition, the candidate will collaborate with national and international researchers that participate in the project. 

The position is a fixed term position for a duration of three years. Appointment to the position of Postdoctoral Research Fellow is mainly intended to provide qualification for work in top academic positions. It is a prerequisite that the applicant can carry out the project over the full course of the employment period. No person may hold more than one fixed term position as a Postdoctoral Research Fellow at the same institution. 

If you receive a personal overseas research grant from NFR it is possible to apply NFR for an extension of the fellowship period corresponding to the length of the stay abroad (minimum three months, maximum 12 months). 

The workplace is at UiT in Tromsø and available from 1 October 2022. Potential candidates must be able to start in the position within a reasonable time, and not later than 3 months after receiving the offer.

The project 
The present project funded by the IKT+ program will focus on the following three applications related to environmental monitoring of the seabed with objects at quite different scales as indicated below: 

  • Recognize selected macro-scale objects, of both natural and anthropogenic origin during ocean surveying. 
  • Recognize selected marine habitats and their endemic macro-fauna and -flora at the seabed, a basis in most marine monitoring studies. 
  • Existing time-consuming procedures for classification of microparticles (diameter < 1 mm; foraminifera and micro-plastic particles), under the microscope will be replaced by novel AI algorithms. These microparticles represent important tools for studying climate and environment. 

    By joining forces using ML, statistical methods, and ICT the present project aims to: 
  • Develop groundbreaking new DL-methodology 
  • Automate information extraction from data collected in seabed monitoring aiming for enhanced quality and reduced costs.

An important part of the postdoc’s contribution in this project will be within classification where highly heterogeneous sources of information are utilized. The postdoc will develop novel directions within classification where the data input contains information of this type. In our test case for the classification of a set of objects, this is highly relevant because multiple sources of information are available. To be specific, these sources include digital 2D images, acoustic images, and video sequences. On top of this, we have auxiliary data given as e.g., contextual descriptions of sediment samples, including geographical location, selected physical and chemical properties, in addition to water column (current speed, temperature, salinity etc.) and various seabed geophysical measures such as multibeam (MBES), backscatter data, side scan sonar etc. 

Network design and uncertainty estimation in Deep Learning is the other main area where the successful candidate will contribute. This work will be done in close collaboration with PhD students and researchers in Visual Intelligence, a center for research-based innovation, led by the ML group at UiT. A crucial part of this work will be trans-dimensional Bayesian Neural Networks where trans-dimensional describes the problem and methods like Reversible Jumps is a possible solution. Computational issues are important here and both Reversible Jumps and within-model ('marginal likelihood') approaches will be pursued. A starting view for this part of the project is that MCMC methods are important scientifically even if they are too slow to use in practice since they can generate an expensive 'gold-standard' benchmark to judge approximate methods against. Separate from computational considerations, there are also wider, conceptual benefits from interpreting Trans-dimensional posteriors over Artificial Neural Networks, in understanding the roles of different aspects of network architecture. 

Contact 
For further information about the position, please contact professor Fred Godtliebsen:

Qualifications
This position requires a Norwegian PhD degree in physics, machine learning,statistics, computer science, or similar, or a corresponding foreign PhD degree recognized as equivalent to a Norwegian PhD degree. If you’re at the final stages of your PhD, you may still apply if you have submitted your PhD thesis for doctoral degree evaluation within the application deadline. You must submit the thesis with your application. You must have dissertated before the start-up date of the position.

The techniques under study in this position should be examined theoretically, but with an emphasis of practical application to data and use cases as dictated by the partnering organizations. For this reason, the research will concern a strong focus on practical implementations, requiring insight into:

  • Applied deep learning using popular deep learning libraries and frameworks, such as PyTorch, Tensorflow, or JAX The use of probabilistic programming languages (Stan, Pyro, PyMC3, etc) for Bayesian and ensemble approaches to neural network implementation, and the intersection thereof with deep learning frameworks 
Software and data engineering concepts for developing tenable, scalable implementations of deep learning training programs as well as data pipelines for online deployment of trained neural network models.

The suitable candidate should have:
  • Background in signal and image processing
  • A solid background in machine learning, statistics, mathematics and linear algebra is also required.
  • Skills in programming and English.

Experience with deep learning (through courses, research projects, or similar), including hands-on experience with software tools such as Pytorch and Tensor Flow, will be considered a strength. Knowledge of pattern recognition and big data processing is considered as an asset.

Other required qualification skills include: 
  • Independence and self-motivation 
  • Creativity and ability to think outside the box 
  • Excellent work ethics and commitment to the job 

Applicants must document fluency of in English and be able to work in an international environment. International experience is an advantage. During the assessment emphasis will be put on the candidate’s motivation, potential for research, and personal suitability for the position. At UiT we put emphasis on the quality, relevance and significance of the research work and not on where the work is published, in accordance with the principles of The San Francisco Declaration on Research Assessment (DORA). 

Inclusion and diversity 
UiT The Arctic University i 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 is 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 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 
  • An inclusive, dynamic, and inspiring work environment 
  • Good welfare arrangements for employees 
  • Flexible working hours and a state collective pay agreement 
  • Pension scheme through the state pension fund 
  • More practical information for working and living in Norway can be found here: https://uit.no/staffmobility 

Application 
Your application must include: 
  • Application letter 
  • CV 
  • Vision statement for research activities (max 2 pages) 
  • Description of your past research project and academic production that are relevant to the advertised position (max 1 page) 
  • Diplomas and transcripts (all academic degrees) 
  • Contact information to 2-3 references 
  • Academic works, up to ten. The doctoral thesis is regarded as one work. 

You must submit a vision statement that describes how you scientifically can contribute to the described field of research. All documentation to be considered must be in a Scandinavian language or English. We only accept applications and documentation sent via Jobbnorge within the application deadline. 

Assessment 
The applicants will be assessed by an expert committee. The committee’s mandate is to undertake an assessment of the applicants’ qualifications based on the application documents and the text of the announcement. The Assessment Committee will give emphasis to the potential for research, as presented in the PhD thesis or equivalent, or in any other academic works. In the assessment, consideration may also be given to professional experience and any other activity which may be relevant for the project. 

The applicants who are assessed as best qualified will be called to an interview. The interview shall aim to clarify the applicant’s personal suitability for the position and motivations. During the interview, the candidate will be asked to give a short presentation of their past research activities. 

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. 

The remuneration for Postdoctoral research fellow is in accordance with the State salary scale code 1352. A compulsory contribution of 2 % to the Norwegian Public Service Pension Fund will be deducted. 

The successful candidate must be willing to get involved in the ongoing development of their department and the university as a whole. 

UiT wishes to increase the proportion of females in academic positions. In cases where two or more applicants are found to be approximately equally qualified, female applicants will be given priority. According to the Norwegian Freedom and Information Act (Offentleglova) information about the applicant may be included in the public applicant list, also in cases where the applicant has requested non-disclosure.

We are not accepting applications for this job through Mathjobs.Org right now. Please apply at https://www.jobbnorge.no/en/available-jobs/job/227795/postdoctoral-research-fellow-in-statistics-and-deep-learning#?p=1 .
Contact: professor Fred Godtliebsen, +4777644019
Email:
Postal Mail:
N-9037 Tromsø, Norway
Web Page: uit.no