Interpretable Reinforcement Learning

Reinforcement learning (RL) enables agents to learn optimal policies by interacting with the environment. The agent collects experience from trial-and-error and optimises its action rules from the environment feedback. RL is getting popular in robotic systems for planning and accomplishment of complex tasks, such as robot navigation and manipulation. However, for these systems to work effectively and safely in the real world, the behaviours of RL agents need to be reliable, and interpretable.

To develop such RL agents, an in-depth understanding of the learning process, i.e., how the agent’s behaviours are developed, is the key. However, it is very challenging due to the random exploration in RL and the long-time learning process. On the other hand, visual analytics researches on providing interactive visual interfaces to facilitate analytical reasoning and has contributed to the interpretability of black-box models.

In this project, we will explore how to integrate visual analytics into the development of reliable and interpretable RL agents. Research questions include: 1) interactive visualization of the learning process to facilitate understanding of and control over RL processes; 2) visual analysis of agent’s interaction with the environment for reasoning of its behaviours; and 3) interactive integration of expert knowledge to regulate the learning process.

Keywords: reinforcement learning, visual analytics, explainable AI

For more information about this project, please contact Dr Jing Wu [Email Address Removed]

Academic criteria: A 2:1 Honours undergraduate degree or a master's degree, in computing or a related subject. Applicants with appropriate professional experience are also considered. Degree-level mathematics (or equivalent) is required for research in some project areas.

Applicants for whom English is not their first language must demonstrate proficiency by obtaining an IELTS score of at least 6.5 overall, with a minimum of 6.0 in each skills component.

How to apply: If you would like to be considered for the School Funded Studentships, please submit your application before the application deadline 29th April 2022 via Computer Science and Informatics - Study - Cardiff University

In order to be considered candidates must submit the following information: 

• Supporting statement 
• CV 
• In the ‘Research Proposal’ section of the application enter the name of the project you are applying to.
• In the funding field of your application, insert “I am applying for 2022 PhD Scholarship in Computer Science and Informatics”, and specify the project title and supervisors of this project in the text box provided.
• Qualification certificates and Transcripts
• References x 2 
• Proof of English language (if applicable)

If you have any questions or need more information, please contact [Email Address Removed]