Abstract: In this talk I'd like to go through a few "advanced" topics in agent-based modelling and apply them in turn to a simple model of fisheries.
First we will examine control in agent-based models. I will first introduce policy optimization in the context of open loop control, then closed loop control and move finally into policy discovery through reinforcement learning.
Second I will describe the techniques we are currently using to take the model to the data and in particular to fit behavioural parameters of the agent-based model with respect to macro and micro-patterns by indirect inference.


Research Programmes