A fully funded 3.5-year PhD position in the School of Computing and Information Systems at the University of Melbourne, focused on helping AI agents learn strategic knowledge from observed behaviour and transfer it to new tasks with different representations.
Institution: University of Melbourne, School of Computing and Information Systems
Field: Computer science; strategy extraction and generalisation for AI agents, machine learning, knowledge representation and reasoning, sequential decision-making
Level: PhD
Funding: Fully funded scholarship, 3.5 years. The scholarship is open to both domestic and international candidates.
Project: Building on prior work that used sequence alignment to extract strategies from agent behaviour and ontological reasoning to generalise them for learning policies in new tasks, the PhD will explore three directions: reducing the amount of hand-crafted modelling of source and target task domains; extracting richer strategies, such as conditional, hierarchical or temporally extended patterns; and finding strategically relevant similarities between contexts, including contexts with substantially different representations. The candidate will also study how transferred strategies can guide policy learning and when transfer improves learning efficiency and performance.
Eligibility: Strong background in computer science, mathematics or a related field, with interest in machine learning, knowledge representation and reasoning, or sequential decision-making.
Deadline: Nov 20, 2026 for expressions of interest, for an early 2027 start.
How to apply: Complete the expression of interest form at https://lnkd.in/gPDgc2Yg and provide a one-page CV and academic transcripts. The contact for this position is Senior Lecturer Michelle Blom.
