Solving a Higgs optimization problem with quantum annealing for machine learning.

Area of Science:

• High-energy physics
• Machine learning applications in particle physics
• Quantum computing and annealing

Background:

• Machine learning is crucial for identifying Higgs-boson decays amidst standard-model processes.
• Current methods rely on simulations that can introduce label noise and systematic errors.
• Overtraining and errors in training data correlations are significant challenges.

Purpose of the Study:

• To apply quantum and classical annealing to a Higgs-signal-versus-background machine learning optimization problem.
• To develop a robust classifier resilient to simulation imperfections.
• To compare the performance of annealing-based classifiers with state-of-the-art methods.

Main Methods:

• Mapped the machine learning problem to finding the ground state of an Ising spin model.
• Constructed a strong classifier using weak classifiers based on Higgs decay photon kinematic observables.
• Utilized quantum and classical annealing techniques for optimization.

Main Results:

• Annealing-based classifiers performed comparably to current state-of-the-art machine learning methods.
• Classifiers are simple functions of interpretable experimental parameters.
• Demonstrated an advantage over traditional methods for small training datasets.

Conclusions:

• Quantum and classical annealing offer a robust and interpretable alternative for particle physics classification.
• The technique's simplicity and error resilience suggest broad applicability in experimental particle physics.
• Potential applications include real-time event selection and neutrino physics classification.