PhD position in Aarhus University on Novel methods for explainable artificial intelligence (Deadline November 1st)

[iosifidi...@gmail.com]

Dear all, 

Applications are invited for a PhD fellowship/scholarship at Graduate School of Natural Sciences, Aarhus University, Denmark, within the Computer Science programme. The position is available from February 2023 or later.

Title:
Novel methods for explainable artificial intelligence  

Research area and project description:

Explainable AI (XAI) is the research field of making decisions of artificial intelligence (AI) models understandable to people. While multiple approaches, such as feature importance or counterfactual examples, have been proposed, many challenges remain unsolved: What types of explanations are suitable for understanding the inner-workings of the AI model's prediction? How can we provide explanations on the dynamics of the input (e.g., spatial and temporal dynamics in videos) and/or different types of information (e.g., color and depth)? Should we provide post-hoc explanations of high-performing but complex Deep Learning models, or should we design inherently explainable Deep Learning models with potentially inferior performance? What are the best metrics for evaluating the multi-objective performance of model explanability and accuracy? How can domain knowledge assist in deriving explanations or in designing explainable models? With this PhD project, we would like to study some of these questions for bringing XAI one step closer to deployment in practice. The successful candidate will investigate how to build new techniques and methodologies for interpretable deep learning models in a project jointly supervised by Professor Ira Assent, Department of Computer Science, Aarhus University, and Professor Alexandros Iosifidis, Department of Electrical and Computer Engineering, Aarhus University, as part of the Pioneer Center for AI, Denmark.

The project is part of The Pioneer Centre for Artificial Intelligence (AICentre.dk), a multi-university research centre led by Professor Serge Belongie at the University of Copenhagen, including Aarhus University, the IT University of Copenhagen, the Technical University of Denmark, and Aalborg University. The centre focuses on fundamental AI research, and within an interdisciplinary framework, develops platforms, methods, and practices addressing society’s greatest challenges.

While PhD students in the centre have their primary affiliation in their supervisor's research group, they also benefit from a network of students, researchers, and courses from the 5 universities. Researchers in the centre have extraordinary access to computing support.

Please upload a project description (½-4 pages). This document should describe your ideas and research plans for this specific project. If you wish to, you can indicate an URL where further information can be found.

Qualifications and specific competences:
We are looking for candidates with a Master’s degree in computer science or closely related field, with strong mathematical and algorithmic skills and interests. Strong students with a Bachelor’s degree are also considered.

Place of employment and place of work:
The place of employment is Aarhus University, and the place of work is Aabogade 34, 8200 Aarhus N, Denmark. 

Contacts:
Applicants seeking further information for this project are invited to contact: Professor Ira Assent, i...@cs.au.dk

How to apply:

For information about application requirements and mandatory attachments, please see the Application guide. Please read the Application guide thoroughly before applying.

When ready to apply, go to https://phd.nat.au.dk/for-applicants/apply-here/ (Note, the online application system opens 1 September 2022)

1. Choose November 2022 Call with deadline 1 November 2022 at 23:59 CET.
2. You will be directed to the call and must choose the programme “Computer Science”.
3. In the boxed named “Study”: In the dropdown menu, please choose: “Novel methods for explainable artificial intelligence (Nmeari)”

Please note:

• The programme committee may request further information or invite the applicant to attend an interview.

Best regards,

Alexandros Iosifidis, Professor

Department of Electrical and Computer Engineering

Aarhus University