Dr Joel Dyer
Computer Science, University of Oxford
Joel is a postdoctoral researcher developing methodology and tooling for agent-based simulation models. He recently completed a DPhil in computational statistics and machine learning at the University of Oxford’s Mathematical Institute and Institute for New Economic Thinking under the supervision of Prof. J. Doyne Farmer, where his research focus was on likelihood-free parameter inference methods for simulation models in the social sciences. Previously, Joel has worked as a Research Scientist at Improbable, and has held visiting research positions at The Alan Turing Institute and University of Bristol through The Alan Turing Institute’s Enrichment Scheme.
Bayesian calibration of differentiable agent-based models
30 May 23
We discuss how generalised variational inference procedures may be employed to provide misspecificat...
No. 2022-30 - Calibrating agent-based models to microdata with graph neural networks
29 Jun 22
Calibrating agent-based models (ABMs) to data is among the most fundamental requirements to ensure t...
Amortised likelihood-free inference for expensive time-series simulators with signatured ratio estimation
24 Feb 22
We propose a kernel classifier for sequential data using path signatures based on the recently intro...
No. 2022-05 - Black-box Bayesian inference for economic agent-based models
01 Feb 22
In this paper, we investigate the efficacy of two classes of simulation-efficient black-box approxim...