Postdoc in Neuro-Symbolic Reasoning and Reinforcement Learning

In this project we study a new approach to synthesis of efficient communication schemes in multi-agent systems, trained via reinforcement learning. We combine symbolic methods with machine learning, in what is referred to as a neuro-symbolic system, where a neural network learns to produce programs in a symbolic language to solve a task at hand. We believe this combination of neural and symbolic methods will be an important next step in the development of AI beyond todays capabilities.

This is a collaborative project with Dr. Vikas Garg at Aaalto University in Finland.

Information about the project
This project touches on several fields, not only AI and computer science but also linguistics and cognitive science. We place our research in a multi-agent context as we consider language learning a collaborative task rather the something that happens entirely in isolation, and take inspiration from studies on human communication in collaborative settings. Agents will not be restricted to use only primitives provided as input, but interactions will be interleaved with (symbolic) steps to extend and revise their current language with novel higher-level concepts, allowing for generalisation and more informative messages.

As a reference, to get an idea of the project, the applicant might want to read this recent paper, which reports on a pilot study for what we will now develop further: 
Towards Learning Abstractions via Reinforcement Learning. E. Jergéus, L. Karlsson Oinonen, E. Carlsson and M. Johansson. 8th Int. Workshop on Artificial Intelligence and Cognitive Science (2022).

The post-doc will join Moa Johansson's vibrant cross-diciplinary research group, including computer scientists, data scientists and a political scientist. We span both machine learning and symbolic AI, and currently study, for example, the language of political parties in Sweden using techniques from NLP, as well as automated reasoning and automated discovery of mathematical conjectures. In addition, we interact and collaborate with many other researchers at Chalmers in both AI, Data Science as well as formal methods, reasoning and programming languages. Chalmers is one of the leading technical universities in Sweden, and its Computer Science department is one of the best and largest in the country.

The project is funded by WASP, which aims to strengthen AI research in Sweden, which also provides many opportunities for networking with other WASP-funded researchers at other universities.

Major responsibilities
We are looking for a Post-doc with an interest in AI (machine learning and/or symbolic methods) to work as a researcher on this project, which will include implementation work. The Chalmers post-doc will work in collaboration with a post-doc or PhD student employed at Aalto University under supervision of Vikas Garg. The initial focus of the Chalmers post-doc will be neuro-symbolic methods, and the development of a working prototype for multi-agent reinforcement learning driving the introduction of new concepts in the communication. The Aalto candidate will focus more on e.g. multi-agent credit assignment. However, we encourage and welcome the post-doc to also develop their own ideas of directions suitable for this project to develop also as an independent researcher.

The position is funded for two years of full time research work. However, the post-doc may also do some teaching, most likely in the areas of Data Science and AI. This will provide an opportunity to gain valuable teaching experience for a future academic career, and employment may be extended with the corresponding time.

Qualifications
To qualify for the position of postdoc, you must hold a doctoral degree awarded no more than three years prior to the application deadline (according to the current agreement with the Swedish Agency for Government Employers). The candidate should have a PhD degree in e.g. Computer Science, AI, computational linguistics, computational cognitive science, mathematics or other relevant fields.

The position requires sound verbal and written communication skills in English. Swedish is not a requirement but Chalmers offers Swedish courses.

Contract terms
This postdoc position is a full-time temporary employment for two years.

We offer
Chalmers offers a cultivating and inspiring working environment in the coastal city of Gothenburg. 
Read more about working at Chalmers and our benefits for employees.

Chalmers aims to actively improve our gender balance. We work broadly with equality projects, for example the GENIE Initiative on gender equality for excellence. Equality and diversity are substantial foundations in all activities at Chalmers.

Read more and apply here

Application deadline: 31st October 2022, or until a suitable candidate has been found.

*** Chalmers declines to consider all offers of further announcement publishing or other types of support for the recruiting process in connection with this position. ***