The Department of Computing Science at Umeå University seeks a postdoctor who will work in the topic of “Neurosymbolic-AI and cognitive architectures” in the WASP-project directed by Assoc. Prof. Zoe Falomir. The employment is full-time for two years with access 31 January 2025 or by agreement.
Project description and working tasks
Neurosymbolic-AI approaches, that is neural symbolic reasoning and learning, try to imitate the fundamental cognitive abilities of intelligent agents: learn from their environment, and reasoning about what has been learned. In these systems, neural networks can learn patterns of spatial relations from the environment that allow reasoning (i.e. learning to reason). Aligning semantic models of expert knowledge to the learning occurred from observing the environment can be used to explain the learning (i.e. reasoning about the learnt). Moreover, cognitive architectures store agents’ experience in their episodic memory which can be queried so that reasoning about past events can be carried out in order to infer the future (i.e. prospection).