Read the original document by opening this link in a new tab.
Table of Contents
Introduction and Motivation
QCNN Circuit Model
Sample Complexity
MERA and QEC
Training Procedure
Applications
Summary
This paper introduces and analyzes a novel quantum machine learning model known as Quantum Convolutional Neural Networks (QCNN). The QCNN architecture combines the multi-scale entanglement renormalization ansatz and quantum error correction to efficiently train and implement on near-term quantum devices. The paper demonstrates the potential of QCNN in recognizing quantum states associated with 1D symmetry-protected topological phases and devising quantum error correction schemes. The QCNN model is compared to conventional methods in terms of sample complexity, showing superior performance near criticality. Additionally, the paper discusses the theoretical insights gained by interpreting QCNN in terms of MERA and QEC. The training procedure for obtaining a QCNN for a specific phase is also described, showcasing the iterative process of updating unitaries to optimize the model. Applications of QCNN in quantum phase recognition and quantum error correction optimization are presented, highlighting the advantages of using QCNN in these contexts.