Dr. Iain Dunning

Researcher and engineer, with a focus on applying techniques from machine learning and optimization to solve difficult decision problems - plan a logistics network, trade futures, run a power grid, or solve a video game.
Currently living in New York City, 🇺🇸.
Contact: iaindunning 📧 gmail.

Recent Work History

Hudson River Trading

Team Lead & Researcher (August '18 —)
I run "HAIL" - HRT AI Labs. We make deep learning work for systematic trading by integrating and translating techniques from other domains and developing proprietary extensions for our unique needs.

DeepMind Technologies

Senior Research Engineer (July '16 — July '18)
Applying large-scale artificial intelligence techniques like deep reinforcement learning to complex environments, using cutting-edge deep learning hardware. Team lead for five engineers, tech lead for multiagent research engineering.

Massachusetts Institute of Technology

Teaching & Research Assistant (Sep. '11 — May '16)
See research below. Co-created XX,000-person EdX class The Analytics Edge. Taught MBA and executive MBA residential versions of the class at MIT Sloan School of Management.


Intern (June 13 — Aug. '13)
Worked on search engine crawler. Designed & implemented algorithms to improve the crawl prioritization, analyzed impacts on very large (O(1010) rows, O(103) TB) datasets with MapReduce/Flume.


Human-level performance in 3D multiplayer games with population-based reinforcement learning
M. Jaderberg*, W. Czarnecki*, I. Dunning*, L. Marris, G. Lever, A. García Castañeda, C. Beattie, N. Rabinowitz, A. Morcos, A. Ruderman, N. Sonnerat, T. Green, L. Deason, J. Z. Leibo, D. Silver, D. Hassabis, K. Kavukcuoglu, T. Graepel.
Science, 2019.
[Science] [arXiv] [video] [DeepMind blog]

The Hanabi Challenge: A New Frontier in AI Research
N. Bard*, J. Foerster*, S. Chandar, N. Burch, M. Lanctot, F. Song, E. Parisotto, V. Dumoulin, S. Moitra, E. Hughes, I. Dunning, S. Mourad, H. Larochelle, M. Bellemare, M. Bowling.
Artificial Intelligence, 2020.

Malthusian Reinforcement Learning
J. Z. Leibo, J. Perolat, E. Hughes, S. Wheelwright, A. Marblestone, E. Duéñez-Guzmánn, P. Sunehag, I. Dunning, T. Graepel.
AAMAS, 2019.

Bayesian Action Decoder for Deep Multi-Agent Reinforcement Learning
J. Foerster, F. Song, E. Hughes, N. Burch, I. Dunning, S. Whiteson, M. Botvinick, M. Bowling.
Submitted, 2019.

IMPALA: Scalable distributed deep-RL with importance weighted actor-learner architectures
L. Espeholt*, H. Soyer*, R. Munos*, K. Simonyan, V. Mnih, T. Ward, Y. Doron, V. Firoiu, T. Harley, I. Dunning, S. Legg, K. Kavukcuoglu.
ICML, 2018.
[arXiv] [DeepMind blog]

What Works Best When? A Systematic Evaluation of Heuristics for Max-Cut and QUBO
I. Dunning, S. Gupta, J. Silberholz.
INFORMS Journal on Computing, 2018.

Inequity aversion improves cooperation intertemporal social dilemmas
E. Hughes*, J. Z. Leibo*, M. Philips, K. Tuyls, E. Duéñez-Guzmán, A. García Castañeda, I. Dunning, T. Zhu, K. McKee, R. Koster, H. Roff, T. Graepel.
NeurIPS, 2018.

Learning Fast Optimizers for Contextual Stochastic Integer Programs
V. Nair, K. Dvijotham, I. Dunning, O. Vinyals.
UAI, 2018.

Population based training of neural networks
M. Jaderberg, V. Dalibard, S. Osindero, W. Czarnecki, J. Donahue, A. Razavi, O. Vinyals, T. Green, I. Dunning, K. Simonyan, C. Fernando, K. Kavukcuoglu.
DeepMind tech report, 2017.
[arXiv] [DeepMind blog]

Extended formulations in mixed integer conic quadratic programming
J. P. Vielma*, I. Dunning, J. Huchette, M. Lubin.
Mathematical Programming Computation, 2017.

JuMP: A modeling language for mathematical optimization
I. Dunning*, J. Huchette*, M. Lubin*.
SIAM Review, 2017.
[arXiv] [JuMP package]

Multistage robust mixed-integer optimization with adaptive partitions
D. Bertsimas, I. Dunning*.
Operations Research, 2016.
[Optimization Online]

Reformulation versus cutting-planes for robust optimization
D. Bertsimas, I. Dunning*, M. Lubin*.
Computational Management Science, 2016.
[Optimization Online]

Computing in operations research using Julia
M. Lubin, I. Dunning.
INFORMS Journal on Computing, 2015.

See Google Scholar for other references.
Some papers use alphabetical ordering - asterix indicates "first" author.


Massachusetts Institute of Technology

Ph.D., Operations Research (Sep. '11 — May '16)
At the MIT Operations Research Center, supervised by Prof. Dimitris Bertsimas

University of Auckland

B.E.(Hons), Engineering Science (Mar. '07 — Dec. '10)