Jaroslaw Pawłowski (WUST)

Quantum entanglement is a fundamental property commonly used in various quantum information protocols and algorithms. Nonetheless, the problem of identifying entanglement has still not reached a general solution for systems larger than 2x3. Modern deep learning (DL) architectures, that use multilayer neural networks (NNs), have enabled unprecedented achievements in various domains like computer vision or natural language processing. Convolutional neural networks (CNNs) with many hidden layers and complex network structures are extremely powerful in feature learning. In quantum physics, one natural application of DL involves the study of quantum many-body systems [1], where the extreme complexity of many-body states often makes theoretical analysis intractable.
In this talk, I will present our recent research on entanglement detection using modern deep CNN architectures, close to the state-of-the-art approaches in DL, trained in a supervised [1] and unsupervised [2] manner, and compare them in terms of their suitability for building robust entanglement detectors. I will demonstrate that, unlike in technological applications, it is crucial that the architecture of NN respects the symmetries present in physical systems. Additionally, I will show how this can be implemented in complex DL architectures. Before I start discussing advanced examples, I will try to smoothly introduce the concept of machine learning from basics.
[1] Hsin-Yuan Huang et al., Science 377, 6613 (2022).
[2] J. Pawłowski, M. Krawczyk, arXiv:2210.07410 (2023).
[3] M. Krawczyk, J. Pawłowski, M.M. Maśka, and K. Roszak, Phys. Rev. A 109, 022405 (2024).