Agent-Based Modelling (ABM) has long been applied to the study of complex phenomena. Over the past decade, it has become increasingly popular across many scientific fields including the study of biological, ecological and agricultural systems. Agent-Based models provide a rich environment for the experimentation and analysis of these systems at different levels of complexity by conceptually breaking them down into individual interacting components. However, the likelihoods, the functions that describe the probability of the observed data given parameter values, for these models are not analytically or computationally tractable. This makes statistical inference for these models challenging.
This session provides a forum for the dissemination of advances in applying ABM approaches to the development of simulation systems in biology, ecology and agriculture. Statistical inferential methods for ABM will also be discussed. Submissions addressing the scientific challenges of modelling these systems using the ABM paradigm and novel approaches for dealing with statistical inference of these models are especially welcomed. Areas of interests, not necessarily exhaustive, include geo-spatial ABMs, hybrid models, model validation and verification, integration with commercial off-the-shelf or open source Geographical Information Systems, optimisation techniques, ABM-enhanced Decision Support Systems, mobile applications, likelihood-free techniques for ABMs such as Approximate Bayesian Computation, Bayesian indirect inference and Bayesian synthetic likelihood.
Key topics: Likelihood-free techniques, Simulation systems in biology, Statistical inferential methods, Optimisation techniques