The University of Southampton
Email:
D.M.Bossens@soton.ac.uk

Dr David Bossens 

In a quest do understand the mind, David started the study of psychology, in which he obtained an MSc in Psychology (Theory and Research Option) from the Catholic University of Leuven (KUL) in the year 2014 (cum laude). Beyond the neuroscience and principles of cognition that inspired the quest, these years were also formative in terms of research methodology, with a strong emphasis on the scientific method, modelling techniques and statistics. While studying psychology, David found that he was more interested in designing systems that following intelligent principles (Artificial Intelligence), rather than trying to copy the -- potentially sub-optimal -- way in which humans reason (Psychology). Thus he went on to do an MSc in Artificial Intelligence (Engineering and Computer Science Option) also at the KUL, a study from which he graduated in the year 2015 (magna cum laude).

From late 2015 until 2019, David joined the University of Southampton for a PhD programme in artificial intelligence. During this project, he studied reinforcement learning in long-term unknown environments, where only few assumptions could be made on the learning environment: sparse reward environments may hamper the ability to learn any rewarding patterns; partial observability and limited knowledge hamper the ability to investigate the environment; many unknown tasks may be presented in sequence (lifelong learning).

Since 2019, David has started his postdoc to improve the robustness of robots to sudden faults or other types of changes in the environment. One current line of investigation is how to evolve a suitable behavioural repertoire, such that searching across this repertoire allows a rapid adaptation to unforeseen changes. Another line of investigation is how to perform realistic, efficient, online adaption based on such behavioural repertoires, despite the many challenges that may occur, such as the gaps between simulation and reality, the different faults affecting different robots, etc.

To get on idea of David's publications, you can visit his Google scholar page. David has also been covered in an Elsevier press article on the topic of active adaptive perception, a unique blend of the universalist and sub-symbolic approach towards artificial general intelligence.

Publications

Bossens, David, Townsend, Nicholas and Sobey, Adam (2019) Learning to learn with active adaptive perception. Neural Networks, 115, 30-49. (doi:10.1016/j.neunet.2019.03.006).

Bossens, David, Mouret, Jean-Baptiste and Tarapore, Danesh (2020) Learning behaviour-performance maps with meta-evolution. In Proceedings of the Genetic and Evolutionary Computation Conference (GECCO) 2020. 10 pp .

Contact

Share this profile FacebookTwitterWeibo