Scientists at the International Islamic University Malaysia through a collaboration with CERN or the European Council for Nuclear Research (Conseil Européen pour la Recherche Nucléaire) and Malaysia National Center for Particle Physics (NCPP) are using a unique artificial intelligence approach to discover new particles. Discovery of these new particles can clarify physical phenomena that can’t be explained by existing theories.
All elements are made of particles, like protons, neutrons, and electrons. The nucleus of the helium atom, for example, contains two protons and two neutrons, with two electrons moving around it.
A Helium atom has two protons and two neutrons in its nucleus that can each be broken down into three quarks.
Particles such as electrons are fundamental particles that cannot be broken down into smaller components. But protons and neutrons are composed of three quarks each. Particle physics studies the properties of fundamental particles and how they interact with each other.
We now know of 17 fundamental particles, including the Higgs boson discovered in 2012 at CERN, Geneva. Physicists are confident there are more fundamental particles out there, as several known phenomena cannot be explained by the current set of particles. These phenomena include neutrino oscillations, dark matter, matter-antimatter asymmetry, and the hierarchy problem. Many new fundamental particles have been proposed by particle physicists as a way of explaining why this phenomena happens, this new particle includes the supersymmetry particles, axions, and quantum black holes. Collectively, this particle are call Beyond Standard Model (BSM) particle.
The Xenada Lab at IIUM is developing several artificial intelligence (AI) methods to help physicists discover BSM particles, if they exist. Using AI to classify particles is not new. But the IIUM approach is unique in that it uses a semi-supervised learning model rather than the popular fully-supervised one.
A fully-supervised AI model learns from both the standard particle model and the BSM at the same time. In contrast, IIUM's semi-supervised learning model, learns only from the standard particle model without using BSM. This AI model only knows standard model particles, and flags anything that is deemed abnormal to be a beyond standard model particle.
In addition to AI models, the Xenada Lab also develops innovative methods for visualizing the high dimensional data that is common in particle physics research. Visualization methods that we have developed include using graph theory, principle component analysis, and SOM U-Matrix. Visualization of data helps to identify clusters in the sample.
The Xenada lab in IIUM Kuantan focus on innovating innovative AI, breaking away from mainstream approaches, contact can be made through email [email protected]. Information about IIUM can be found at https://www.iium.edu.my/
Mapping of Dimuon Sample on SOM U-Matrix shows a small cluster of Z Boson
