Centre for Informatics and Applied Optimization (CIAO)

Faculty of Science and Technology

Federation learning agents (FLAG)


Associate Professor Peter VamplewPeter Vamplew 

Dr Richard Dazeley

Richard Dazeley 
Dr Cameron Foale 
Dr Dean Webb 
Adam Bignold 
Evan DekkerEvan Dekker 
Anthony Rawlins 
Charlotte Young 

FLAG’s research lies primarily in the development and application of reinforcement learning (RL) algorithms. These algorithms allow intelligent software agents to learn to carry out near-optimal decision making in the context of sequential decision-making tasks, based on a reward signal. RL methods can learn to perform at high levels even on tasks where human knowledge is limited.

Our current research is focused on four main areas:

  • The extension of RL methods to problems with multiple conflicting objectives. We have been among the world leaders in establishing multiobjective reinforcement learning (MORL) as a distinct and growing sub-discipline of RL.
  • Designing RL methods to operate effectively in domains with coarse state space discretisation.
  • Development of assisted RL methods which can learn effectively both when provided with advice by a human or other advisor, or when learning independently.
  • The use of MORL methods to implement safe, trusted and ethical autonomous artificial intelligence.

Our work sometimes crosses over into related areas such as broader machine learning, and specific areas of application including cybersecurity and digital forensics.