Physics-AI opportunities at MIT

Below are examples of our research at the interface of physics and AI, both using AI to do physics better and using physics to build better AI.

Using quantum computers to improve machine learning and optimization algorithms
Christoph Paus
With my students and postdocs, we use various machine learning techniques to push our CMS data analysis and squeeze the most out of the data. Specifically, we work on lepton identification, boosted particle reconstruction including subjet structure information, improved mass and energy reconstruction.
Eddie Farhi
Quantum computing and machine learning
Understanding the physics of machine learning, including natural realizations in quantum systems and noisy classical systems
Jackie Hewitt
Analysis and interpretation of 21 cm cosmology data from HERA
Jeremy England
The non-equilibrium statistical mechanics of neural networks
Jesse Thaler
Studying LHC events using techniques from weak supervision & topicmodeling. In the process, discovered linear basis for collider observables, allowing linear regression to match the classification performance of deep neural networks.
Liang Fu
Modeling many-body systems near phase transitions with machine learning-accelerated Monte Carlo simulation
Novel AI techniques to help with physics research. Physics-inspired algorithms for better machine learning. Novel hardware for AI.
Machine learning tools for reconstructing, monitoring, and analyzing experimental particle physics data CMS
Matt Evans
LIGO event detecton and analysis with machine learning
Modeling phase transitions with deep learning. Using physics-inspired techniquest to make deep learning algorithms more efficient, transparent and trustworthy.
Mike Williams
Machine learning tools for analyzing experimental particle physics data from LHCb & CMS
Philip Harris
Real-time LHC data reconstruction using FPGA-based DNN inference; Higgs coupling extraction via adversarial NNs for automated tuning of MC QCD simulations
Sarah Seager
Machine learning for finding exoplanets in large data sets
Seth Lloyd
Quantum machine learning
Use machine learning to speed up lattice QCD calculations for particle and nuclear physics (e.g. arXiv:1801.05784)


We’re using machinelearning tools to analyze particle physics data from the Large Hadron Collider.

We’re developing technology for faster and more energy-efficient deep learning using optical chips, that compute using photons instead of electrons.

We’re using techniques from condensed matter physics to help understand how our brains process information,

We use machine-learning techniques to detect extrasolar planets and gravitational waves from colliding black holes.