Workshop on Multi-Objective Decision Making

In the modern world, many problems have more than one objective. For example, imagine an office in which the meeting scheduling controlled by an autonomous system needs to maximise the comfort levels of the building's occupants while minimising the energy consumption, or an agent that advises a medical treatment plan for schizophrenia that aims to minimise the severity of the symptoms, while minimising weight gain and maximising quality of life. When a priori scalarisation of such a multi-objective problem is not possible, explicitly multi-objective methods are necessary to enable (multi-)agent systems for these environments.

The Multi-Objective Decision Making (MODeM) workshop aims to bring together people from across the agents community and beyond. Inside of the agents community (and the operations research, control theory and robotics communities), researchers have recently been working on multi-objective: decision-theoretic planning, reinforcement learning, multi-agent coordination, constraint optimisation problems, path planning and game theory. In adjacent communities, multi-objective evolutionary algorithms and multi-objective (heuristic) optimisation, and multi-criteria decision-making and multi-attribute utility theory, are large and long-established (sub)fields. Another highly related problem is that of preference elicitation with respect to different objectives, which is studied in the field of computational social choice. The goal of this workshop is to bring together ideas from all these (sub)fields and communities, leading to cross-pollination, and hopefully interesting new collaborations.