Quantum Convolutional Neural Networks for High-Energy Physics Analysis at the LHC

• Organization
  • Machine Learning for Science (ML4Sci)
• Student
  • Eraraya Ricardo Muten
• Mentors
  • Prof. Sergei V. Gleyzer, Dr. Emanuele Usai, and Raphael Koh
• Project Details
  • Project Idea
  • Accepted Proposal

Introduction

One of the challenges in High-Energy Physics (HEP) is events classification, which is to predict whether an image of particle jets belongs to events being sought after or just background signals. Classical Convolutional Neural Network (CNN) has been proven a powerful algorithm in image classification, including jets image. As quantum computers promise many advantages over classical computing, comes a question on whether quantum machine learning (QML) can give any improvement in solving the problem. This project aims to demonstrate quantum machine learning’s potential, specifically Quantum Convolutional Neural Network (QCNN), in HEP events classification from image data. The code used in the research is wrapped as an open-source package to ease future research in this field.

How to Use

Package Description

This package is a TensorFlow Quantum implementation of quantum convolution and classifier with Data Re-uploading ansatz. Both are wrapped as Keras layers that can easily be integrated into other Keras layers (classical and/or quantum), acting as building blocks for Quantum Convolutional Neural Networks (both hybrid and fully quantum). The model can be trained using Keras API.

Installation

git clone https://github.com/eraraya-ricardo/qcnn-hep.git
cd qcnn-hep
python -m pip install -r requirements.txt
python setup.py

For a more detail step-by-step installation, please refer to Docs and Tutorial.