Workshop on Quantum computing: Quantum Machine Learning
Abstract: Quantum convolutional neural networks (QCNNs) have been proposed as a method to classify quantum state inputs into their different phases of matter. This is in contrast to traditional methods based on order parameters, which raises the question about the accuracy of the QCNN phase recognition algorithm.
In this talk I will introduce these QCNNs and provide some intuition regarding how they were designed. Then, by employing the adiabatic state preparation algorithm as a tool, analyse the accuracy of the QCNN for classifying the Hamiltonians of the cluster-Ising model into their respective phases. By studying the distribution of how the QCNN output changes with Hamiltonian parameters, we show properties of phases that imply correct classification. Using these results, we construct an infinite family of QCNNs with similar classification abilities.
About the speaker: Nathan McMahon obtained his PhD in 2018 in quantum information theory from the University of Queensland, Australia. Since then, he has held postdoctoral positions at the University of Queensland, Friedrich-Alexander University Erlangen-Nurnberg, and most recently, Leiden University. He has been awarded an Alexander von Humboldt fellowship and is currently interested in fundamental aspects of quantum machine learning.