Skip to main content

Dr. Stephen Kelly

Assistant Professor

Department of Computing and Software

Expertise:
Genetic Programming; Evolutionary Reinforcement Learning; Artificial Life; Nature-Inspired Computing in Robotics, Art, and Games
Research Clusters:
Email:
Office:
ABB 533

Biography

Stephen Kelly is an Assistant Professor in the Department of Computing and Software at McMaster University and a visiting researcher at Google Brain. He received his PhD in computer science from Dalhousie University and completed an NSERC post-doctoral fellowship at the BEACON Center for the study of Evolution in Action at Michigan State University. Stephen’s work focuses on genetic programming in predictive control environments, with a particular interest in how emergent forms of memory and hierarchy allow digital evolution to build programs in dynamic, partially-observable, and multi-task environments. His work is published in international journals and conference proceedings, and has won best paper awards at EuroGP 2017 and GECCO 2017, and a 2018 Silver “Humie” Award for Human-Competitive Results Produced by Genetic and Evolutionary Computation.

Prior to studying computer science, Dr. Kelly completed a Bachelor of Fine Arts at the Nova Scotia College of Art and Design. His interdisciplinary research-creation practice explores bio-inspired computing as raw material for storytelling, activism, and public engagement. The results are mechatronic art/science hybrids that have been exhibited at major international venues such as LABoral Art and Industrial Creation Centre (Gijón, Spain), MUTEK Festival of Digital Creativity and Electronic Music (Montreal, Canada), Nuit Blanche (Paris, France), and the Art Gallery of Nova Scotia.

Publications

Selected

Stephen Kelly, Tatiana Voegerl, Wolfgang Banzhaf, and Cedric Gondro

Evolving Hierarchical Memory-Prediction Machines in Multi-Task Reinforcement Learning

Genetic Programming and Evolvable Machines

Stephen Kelly and Malcolm I. Heywood

Emergent Solutions to High-Dimensional Multi-Task Reinforcement Learning

Evolutionary Computation, 26(3):347-380, 2018. MIT Press

Stephen Kelly, Robert J. Smith, Malcolm I. Heywood, and Wolfgang Banzhaf

Emergent Tangled Program Graphs in Partially Observable Recursive Forecasting and ViZDoom Navigation Tasks

ACM Transactions on Evolutionary Learning and Optimization, 1(3), 2021

View more Publications