Fitting a collider in a quantum computer: tackling the challenges of quantum machine learning for big datasets

Area of Science:

• High Energy Physics
• Quantum Computing
• Machine Learning

Background:

• Quantum systems face challenges processing large, high-dimensional datasets common in high energy physics.
• Feature and data prototype selection are key to overcoming these limitations.

Purpose of the Study:

• To investigate feature and data prototype selection techniques for quantum machine learning in high energy physics.
• To benchmark quantum algorithms against classical methods on large datasets.

Main Methods:

• Grid search and training of quantum machine learning models.
• Benchmarking against classical shallow machine learning methods.
• Application of Sequential Backward Selection and Principal Component Analysis for feature selection.

Main Results:

• Quantum algorithms achieved performance comparable to classical algorithms, even on large datasets.
• Principal Component Analysis demonstrated more stable results than Sequential Backward Selection for feature selection.
• Variability in quantum model performance is linked to the use of discrete variables.

Conclusions:

• Principal Component Analysis-transformed data is suitable for quantum machine learning in high energy physics.
• Careful feature selection is essential for reliable quantum algorithm performance.
• Quantum machine learning shows promise for analyzing complex physics datasets.