Here are some of my recent research projects. For potential students: Please contact me if you are interested in doing a PhD, Master thesis, Bachelor thesis, or any other kind of projects with me. 

• Learning to communicate in MARL: 
  • There is a completing PhD project conducted by Changxi Zhu where I am the daily supervisor.  See our survey paper on JAAMAS and recent RLC paper on multi-task Comm_MARL. Attention: Changxi will defend his thesis on Oct 29th! 
  • There is an ongoing PhD project conducted by Thomas Wessels, where I am the daily supervisor. This project will focus on learning the causal effects of communication in MARL. It is funded by Hybrid Intelligence, and is co-supervised by Prof. Sara Magliacane. 
• Neuro-symbolic (MA)RL: 
  • There is a recently completed PhD project conducted by dr. Shuai Han, where I was the daily supervisor. The project focuses on the structured (MA)RL with sparse reward and neuro-symbolic knowledge. See our recent paper on AAMAS. 
  • There is an ongoing PhD project on neuro-symbolic MARL, conducted by Satchit Chatterji at UvA, where I co-supervise (with dr. Erman Acar and dr. Giovanni Sileno)
  • Check our last workshop on ‘Inductive Biases in Reinforcement Learning’ in Reinforcement Learning Conference 2025. 
• RL with natural language processing & LLMs:
  • Use deep RL for dialogue systems policy optimization: A completed PhD project and ongoing collaborative project with dr. Yangyang Zhao.  See our recent papers on TACL and EMNLP. 
  • Specialize NLP models through reinforcement learning: This is an ongoing PhD project conducted by Leon Eshuijs at VU, where I co-supervise (with Prof. Antske Fokkens). See our recent CoNLL paper and EMNLP paper. 
• Use RL to support real-time human (healthcare) behaviors:
  • Following the success of our project 'Wie zorgt?', the Wie Zorgt 2 project (i.e. Intelligent Home Care for Dementia) has recently been funded by ClickNL. dr. João M. G. Ribeiro is working on this project as a Postdoc. In this project, we aim to develop an intelligent RL-based conversational system to decrease the pressure on healthcare professionals and improve the quality of care for the elderly with dementia. See our recent paper on the topic.  
  • There is an ongoing PhD project on (RL-based) social robot for group discussion support, conducted by Ruben Weijers in TU Delft. I co-supervise the project with dr. Catharine Oertel, dr. Mike Ligthart and Prof. Catholijn Jonker. 
• Understand public sentiment and communication from social media data during the COVID-19 outbreak period: This research started with NLP analysis in the PuReGoMe project and now we extend it to analyze the dynamics of communication networks. We have a great number of Dutch social media data and are open to collaborations. 

Graduated PhD projects:

• dr. Yangyang Zhao (2022.12): Task-Oriented Dialogue Policies Based on the Deep Reinforcement Learning. Now lecturer in Changsha University of Science and Technology
• dr. Selene Baez Santamaria (2025.10):  Knowledge-centered conversational agents with a drive to learn. Now postdoc at University of Zurich
• dr. Shuai Han (2026.03): Improving sample efficiency of reinforcement learning: Exploiting structural knowledge for decision making