Research
The connection between cause and effect is how we first learn about the world. But, surprisingly, modern machine learning is blind to cause and effect, relying on correlations extracted from data rather than causality.
My research has focused on using concepts and tools from fundamental physics to develop algorithms for extracting cause and effect from correlations. My main achievement has been to apply these algorithms in diverse areas: privacy, healthcare and decision-making. I’ve been able to dramatically reduce the rate of medical misdiagnoses, was the first to develop quantum cryptography for large-scale quantum networks, and showed how to combine counterfactual reasoning with deep learning to make automated decision making more efficient and trustworthy.
My research also focuses on the foundations of physics. I have derived connections between the laws of physics and fundamental limits on computation, and set bounds on the structure of post-quantum physics.
My work has been called a “breakthrough” by Newsweek, and MIT Technology Review said it has the potential to “supercharge medical AI” and that it “is set to improve automated decision making in finance, health care, ad targeting, and more.“ My research has also been featured in media outlets including New Scientist, The Times, The Telegraph, The London Review of Books, & Gizmodo. My research papers on causality were highlighted in the 2020 State of AI Report and selected as an Editors’ Highlight by Nature Communications in both 2020 and 2022.
I’ve listed some of my research papers by topic on the right-hand side of this page. See my Google Scholar profile for a full, up-to-date list of all my research papers.
Causal Inference and Machine Learning
Estimating categorical counterfactuals via deep twin networks, Nature Machine Intelligence 2023. See coverage in MIT Technology Review.
Disentangling Causal Effects from Sets of Interventions in the Presence of Unobserved Confounders, Conference on Neural Information Processing Systems 2022 (NeurIPS-2022). See coverage from Spotify Research.
Technology readiness levels for machine learning systems, Nature Communications 2022. Selected as Editors Highlight. See coverage from MIT Media Lab.
Integrating overlapping datasets using bivariate causal discovery, Association for the Advancement of Artificial Intelligence 2020 (AAAI-20). See coverage in MIT Technology Review and Karma.
Improving the accuracy of medical diagnosis with causal machine learning, Nature Communications 2020. Selected as Editors Highlight. See coverage in New Scientist, The Times, The Telegraph, London Review of Books, and on the New Scientist podcast.
Quantum Cryptography and Quantum Causality
Towards device-independent information processing on general quantum networks, Physical Review Letters 2018. See coverage in New Scientist, Newsweek, and Gizmodo.
Device-independent certification of non-classical joint measurements via causal models, Nature Quantum Information 2019. See coverage in Gizmodo.
Quantum common causes and quantum causal models, Physical Review X 2017. See coverage in American Physical Society Viewpoint article and MinutePhysics video.
Quantum Computing
The computational landscape of general physical theories, Nature Quantum Information 2019.
Deriving Grover’s lower bound from physical principles, New Journal of Physics 2016.
Generalised phase kick-back: The structure of computational algorithms from physical principles, New Journal of Physics 2016.
Post-Quantum Physics and Quantum Foundations
A no-go theorem for theories that decohere to quantum mechanics, Proceedings of the Royal Society A 2018. See coverage in New Scientist, and in a special feature issue of New Scientist.
Higher-order interference in extensions of quantum theory, Foundations of Physics 2017.
Ruling out higher-order interference from purity principles, Entropy 2017. See coverage in New Scientist.