PhD Stipend in Safe Machine Intelligence for Optimal Decision Making

In line with the commitment to building a theoretical foundation of safe reinforcement learning, we are looking for a highly motivated and talented PhD student to join our team. The position is a part of the SAFEMig Project. The project aims to develop efficient and safe RL algorithms with provable guarantees.

An autonomous system uses its senses(physical measurements) to learn about a physical environment. These decisions are taken to reach a specific goal based on the observed environment. At each time, an action is taken. It influences possible future states which are uncertain due to stochastic disturbances.  To take an optimal decision means to take an action that gives the best expected outcome. This is the so-called stochastic optimization. Some decisions may bring the autonomous system into unsafe or hazardous conditions, where the people’s health and equipment are endangered. Specifically, the notion of safety to be developed in the project will characterize the risk of entering an unsafe situation. We examine two concepts of safety: probabilistic safety and fatigue safety. Probabilistic safety is the probability of reaching hazardous states at most once, whereas fatigue safety allows rare visits to hazardous states. To characterize the level of safety, we define safety metrics. Subsequently, we develop numerically tractable methods for computing safety. The level of safety can be managed by taking proper decisions. The trade-off between safety and greediness are assessed using stochastic constrained optimization.

At the outset, it is assumed that the mathematical model of the system is known. Afterwards, leaning upon the developed theory of model-based decision, the project develops data-driven algorithms where the actual model is unavailable. As a proof of concept, the theoretical findings of the SAFEmig project are evaluated in commercial refrigeration and wastewater management case studies. The first case study is proposed to verify the SAFEmig algorithms for data-driven safety. In the second case study, the decision learning algorithms are designed to control the wastewater with the measurement data.

Specifically, the goals of the SAFEmig project are:

• Characterization of two forms of safety: probabilistic safety and fatigue safety.
• Assuming the perfect knowledge of the model, establishing stochastic optimization with probabilistic and fatigue safety guarantees.
• Developing the theory of decision learning with safety constraints.

 Qualification requirements: 

• Master's degree in Electrical Engineering or a related field.
• Excellent understanding of dynamical systems and optimal control.
• Proficiency in programming languages such as Matlab and Python.
• Excellent problem-solving and analytical skills.
• Strong written and verbal communication skills.
• Prior experience in reinforcement learning is a plus.

The application must contain the following: 

• Cover Letter(a motivated text up to two pages wherein the reasons for applying, qualifications in relation to the position, and intentions and visions for the position are stated).
• Current CV(including personal data).
• Copy of relevant diplomas(Master of Science). On request you could be asked for an official English translation.
• Scientific qualifications. You may attach up to 5 publications(optional).
• References/recommendations.

About Us:

The Learning and Decision Lab is part of the Automation and Control Section in the Department of Electronic Systems, Aalborg University. In electronic engineering, Aalborg University is known worldwide for its high academic quality and societal impact. The Department of Electronic Systems employs more than 200 people, of which about 90 are PhD students, and about 40% of all employees are internationals. In total, it has more than 600 students in its BSc and MSc programs, which are based on AAU's problem-based learning model. The department leverages its unique research infrastructure and lab facilities to conduct world-leading fundamental and applied research within communication, networks, control systems, AI, sound, cyber security, and robotics. The department plays an active role in transferring inventions and results into applications in close collaboration with industrial partners worldwide. You can read more about the department at www.es.aau.dk.

We at the Learning and Decision Lab are comprised of 14 international junior and senior researchers. It is headed by Professor Rafael Wisniewski. We're a cutting-edge research lab pioneering developments in safety, reinforcement learning, and game theory, to name a few. Our commitment is to drive progress in safe reinforcement learning, particularly emphasizing novel algorithm development with provable guarantees. We strive to publish our research findings in top-tier journals and conferences such as IEEE Transactions of Automatic Control, Automatica, IEEE CDC, L4DC, ICML, and NeurIPS.

You may obtain further information from Prof. Rafal Wisniewski, Department of Electronic Systems, email:[email protected] , concerning the scientific aspects of the stipend.

PhD stipends are allocated to individuals who hold a Master's degree. PhD stipends are normally for a period of 3 years. It is a prerequisite for allocation of the stipend that the candidate will be enrolled as a PhD student at the Technical Doctoral School of IT and Design in accordance with the regulations of Ministerial Order No. 1039 of August 27, 2013 on the PhD Programme at the Universities and Certain Higher Artistic Educational Institutions. According to the Ministerial Order, the progress of the PhD student shall be assessed at regular points in time.

Shortlisting will be applied. This means that subsequent to the deadline for applications the head of department supported by the chair of the assessment committee will select candidates for assessment. All applicants will be informed whether they will be assessed or not.

For further information about stipends and salary as well as practical issues concerning the application procedure contact Ms. Lisbeth Diinhoff, The Doctoral School at The Technical Faculty of IT and Design, email:[email protected] , phone:+45 9940 9589.

For more information of The Technical Doctoral School of IT and Design:www.phd.tech.aau.dk

The application is only to be submitted online by using the"Apply online" button below.

AAU wishes to reflect the diversity of society and welcomes applications from all qualified candidates regardless of personal background or belief.

Appointment and salary as a PhD fellow are according to the Ministry of Finance Circular of 15 December 2021 on the Collective Agreement for Academics in Denmark, Appendix 5, regarding PhD fellows, and with the current Circular of 11 December 2019 on the employment structure at Danish universities.