From classical to quantum machine learning through tensor networks
Description
This project explores the intersection of machine learning and quantum mechanics to address computational challenges in both fields. Machine learning, a key driver of computational demand, is applied to solve complex quantum problems, while quantum computation holds the promise of accelerating machine learning tasks. The research focuses on three objectives:
1. **Neural Quantum States (NQS):** Leverage machine learning models, such as neural quantum states, to analyze strongly correlated many-body quantum systems where conventional methods fail, tackling issues like dynamics, localization, and phase transitions.
2. **Tensor Networks for Machine Learning:** Explore the application of tensor-network methods to standard machine learning tasks, evaluating their potential to enhance generalization, uncertainty estimation, and adversarial robustness, particularly for one-dimensional datasets.
3. **Quantum Algorithms for Machine Learning:** Develop quantum algorithms tailored to noisy intermediate-scale quantum (NISQ) devices, using tensor-network approximations and the "learning to learn" paradigm to improve the efficiency and scalability of machine learning training processes.
The project aims to bridge quantum mechanics and machine learning, providing breakthroughs in understanding many-body quantum systems, advancing neural network design, and paving the way for practical quantum speedups in machine learning. The results will have significant implications for research and industry, spanning areas like high-temperature superconductivity, autonomous systems, and efficient resource utilization.
Collaborators
Publications
- Žunkovič B, Ribeiro P. Mean-field approach to midspectrum eigenstates of long-range interacting quantum systems. Physical Review B. 2024 Sep 1;110(10):104114.
- Mossi A, Žunkovic B, Flouris K. A Matrix Product State Model for Simultaneous Classification and Generation. arXiv preprint arXiv:2406.17441. 2024 Jun 25.
- Žunkovič B, Torta P, Pecci G, Lami G, Collura M. Variational ground-state quantum adiabatic theorem. arXiv preprint arXiv:2406.12392. 2024 Jun 18.
- Žunkovič B, Zegarra A. Mean-field dynamics of an infinite-range interacting quantum system: Chaos, dynamical phase transition, and localization. Physical Review B. 2024 Feb 1;109(6):064309.
- Žunkovič B, Ilievski E. Grokking phase transitions in learning local rules with gradient descent. Journal of Machine Learning Research. 2024;25(199):1-52.
- Žunkovič B. Positive unlabeled learning with tensor networks. Neurocomputing. 2023 Oct 1;552:126556.
- Žunkovič B. Deep tensor networks with matrix product operators. Quantum Machine Intelligence. 2022 Dec;4(2):21.
- Ulčar M, Robnik-Šikonja M. Sequence-to-sequence pretraining for a less-resourced Slovenian language. Frontiers in Artificial Intelligence. 2023 Mar 28;6:932519.
- Ilievski E. Popcorn Drude weights from quantum symmetry. Journal of Physics A: Mathematical and Theoretical. 2023 Jan 3;55(50):504005.
- Ulčar M, Robnik-Šikonja M. Cross-lingual alignments of ELMo contextual embeddings. Neural Computing and Applications. 2022 Aug;34(15):13043-61.