Machine Learning, AI and HPC research at the IFT:
Higgs physics, Beyond the Standard Model (BSM) physics and LHC
We use ML to study the LHC sensitivity to the Higgs self-coupling in VHH production with DNNs. We also use supervised / unsupervised / weakly supervised taggers for complex objects: multi-pronged jets, jets containing prompt non-hadronic objects, jets containing non-hadronic objects with displaced vertices, collimated non-hadronic objects. We also perform automated searches for BSM physics via direct searches for new physics with unsupervised methods.
Cosmology, Dark Energy, Gravitational Waves
We use Genetic Algorithms in order to make model independent reconstructions of the expansion history of the Universe and the evolution of matter perturbations, along with null (=consistency) tests of the cosmological constant model. We also train Neural Networks to detect gravitational wave events and extract cosmological parameters, in particular in the case of hyperbolic encounters which have unique signatures for Primordial Black Holes.
Astroparticle Physics and Dark Matter (DM)
We use ML to perform signal/background discrimination in direct DM detection experiments (evolving from early boosted decision trees) and at the LHC (monojet, monophoton). We also use of deep learning to obtain the DM density profile (useful mainly for the inner parts of the galaxies) and ML algorithms to disentangle between standard astrophysical sources and dark matter subhalos.
Theoretical Condensed Matter and Quantum Information
We apply ML techniques to mitigate the error in the present noisy quantum computers. With a combination of ML and tensor network techniques we can design more efficient algorithms. We also work to incorporate quantum logic into ML, in what is known as quantum ML.
Top Physics
We use CNNs feed with jet images to create hadronic and leptonic top taggers.
High Energy QCD
We use ML and AI techniques to detect and classify multi-particle final states at hadron colliders.
Some selected publications:
Taggers for multi-pronged jets for Atlas and CMS:
https://arxiv.org/abs/2002.12320
Reinterpretation of LHC Results and Machine learning:
https://arxiv.org/abs/2003.07868
Higgs flavors at the LHC with Neural Networks:
https://arxiv.org/abs/2008.12538
Dark matter constraints from dwarf galaxies: a data-driven analysis:
https://arxiv.org/abs/1803.05508
Search for solar electron anti-neutrinos with Super-Kamiokande and machine learning:
https://arxiv.org/abs/2012.03807
Neutron-Antineutron Oscillation Search with Super-Kamiokande and machine learning
https://arxiv.org/pdf/2012.02607.pdf
Search for proton decay with Super-Kamiokande and machine learning:
https://arxiv.org/abs/2010.16098
Genetic Algorithms and GWs with lensing or standard sirens:
https://arxiv.org/abs/2011.02718
https://arxiv.org/abs/2007.14335
Cosmology, Genetic Algorithms and BSM physics with Euclid:
https://arxiv.org/abs/2007.16153
Dark Energy and Genetic Algorithms:
https://arxiv.org/abs/2010.04155
https://arxiv.org/abs/2001.11420
https://arxiv.org/abs/1910.01529
https://arxiv.org/abs/1205.0364
Swampland conjectures and ML:
https://arxiv.org/abs/2012.12202
Tensor Networks and ML:
https://arxiv.org/abs/1902.02362
Application of ML techniques to error mitigation in the present noisy quantum computers:
https://arxiv.org/abs/2012.00831
Unified approach to data-driven quantum error mitigation:
https://arxiv.org/abs/2011.01157
Boltzmann Machines and 1D Quantum Many-Body Wave Functions:
https://arxiv.org/abs/1910.13454