Principal Investigator: Adán Garriga
Email: adan.garriga@eurecat.org
Web: https://eurecat.org/en/field-of-knowledge/quantum-computing/
Brief Theme Description:
The proposed research theme aims to advance quantum machine learning (QML) algorithms tailored to near-term, noisy intermediate-scale quantum (NISQ) devices, with a particular focus on generative models. Building on our recent work on shallow instantaneous quantum polynomial-time (IQP) circuits as quantum circuit Born machines for random graph generation, we will further develop this line into a general framework for scalable quantum generative modeling.
Methodologically, the project will explore hybrid QML architectures that (i) train predominantly on classical hardware via statistically motivated loss functions, while (ii) delegating sampling and certain subroutines to quantum processors and high-fidelity quantum simulators (up to 34 qubits available at EURECAT). We will investigate encoding schemes and circuit designs that are both expressive and hardware-efficient, studying their robustness to noise, their ability to capture higher-order correlations in complex networks, and their practical performance against strong classical baselines. Collaboration and secondments with the QML-CVC group will ensure a strong link to computer vision and representation learning, enabling cross-fertilization between quantum generative models for graphs, images, and spatio-temporal data, and paving the way for realistic, industry-oriented use cases of QML in the NISQ era.
Available Infrastructures: 34 qubits Quantum Simulator. Digital Annealer. GPUs clusters at Eurecat and CVC.
Possible Secondments: 1) QML-CVC (https://qml.cvc.uab.es/). 2) Quantum Information Theory Group at ICFO (https://www.icfo.eu/research-group/7/quantum-information/home/437/).
Keywords: Quantum Computing, Quantum Machine Learning, Variational Algorithms, Hybrid Algorithms, Generative Quantum Machine Learning.
