Let's fail elegantly together

This seminar will discuss how we can design agents that act in such a way that they select collective strategies to avoid more critical failures (norm violations) and mitigate the effects of violations that do occur.

Location: Room C/2.07, Queen's Buildings, School of Computer Science and Informatics.
Date: 8 November 2017 at 14:00

The Chair for this seminar is Federico Cerutti.

Abstract

In making practical decisions, agents are expected to comply with ideals of behaviour or norms. In reality, it may not be possible for an individual or a team of agents to be fully compliant. Actual behaviour often differs from the ideal.

We model the normative requirements of a system through contrary-to-duty obligations and violation severity levels, and propose a novel multi-agent planning mechanism based on decentralised partially observable Markov decision process (POMDPs) that uses a qualitative reward function to capture levels of compliance: N-Dec-POMDPs (N-decentralised-partially observable Markov decision process).

We develop mechanisms for solving this type of multi-agent planning problem and show, through empirical analysis, that joint policies generated are equally as good as those produced through existing methods but with significant reductions in execution time.

Biography

Tim Norman is Professor of Computer Science and Head of the Agents, Interaction and Complexity Group at the University of Southampton.

He read Electronic and Electrical Engineering at University of Wales, Swansea, then graduated in 1997 with a PhD in Computer Science from University College London in the area of AI planning and scheduling.

After working as a postdoc at Queen Mary University of London, he moved to the University of Aberdeen in 1999 where he was promoted to Professor in 2009. He joined the Agents Interaction and Complexity Group at ECS (Electronic and Computer Science) Southampton in 2016.

He is an expert in artificial intelligence, logic and automated reasoning. His most significant research contributions lie in autonomous agents and multi-agent systems, argumentation, logics of norms, action and imperatives, and in trust assessment and trust-informed decision making.

Tim is currently collaborating with Alun Preece, Federico Cerutti and others in the University on the International Technology Alliance in Distributed Analytics and Information Sciences (DAIS ITA) project, and with Federico commercialising prior research on argumentation-based analytics.

Further reading

Paper: Gasparini, Luca, Norman, Timothy and Kollingbaum, Martin J. (2017) Severity-sensitive norm-governed multi-agent planning

Code (BSD license): Gasparini, Luca, Norman, Timothy and Kollingbaum, Martin J. (2017) Norm analysis and norm-governed planning University of Southampton