Identifying quantum phase transitions using artificial neural networks on experimental data

Benno S. Rem, Niklas Käming, Matthias Tarnowski +5 more

Haoxiang Chen, Weiluo Ren, Xiang Li +1 more

Chao Fang, Amin Barzeger, Helmut G. Katzgraber

Abstract Known for their ability to identify hidden patterns in data, artificial neural networks are among the most powerful machine learning tools. Most notably, neural networks have played a central role in identifying states of matter and phase transitions across condensed matter physics. To date, most studies have focused on systems where different phases of matter and their phase transitions are known, and thus the performance of neural networks is well controlled. While neural networks present an exciting new tool to detect new phases of matter, here we demonstrate that when the training sets are poisoned (i.e. poor training data or mislabeled data) it is easy for neural networks to make misleading predictions.

Vladislav Chertenkov, Evgeni Burovski, Щур Лев Николаевич

Valentin Gebhart, Martin Bohmann, Karsten Weiher +4 more

The authors use machine learning methods to identify nonclassicality from experimental data obtained via homodyne detection, opening the possibility for fast sorting and direct monitoring of quantum light in experiments.

Akshay Vijay Gaikwad, Omkar Bihani, Arvind +1 more

In this study we employ a feed-forward artificial neural network (FFNN) architecture to perform the tomography of quantum states and processes obtained from noisy experimental data. To evaluate the performance of the FFNN, we use a heavily reduced data set and show that the density and process matrices of unknown quantum states and processes can be reconstructed with high fidelity. We use the FFNN model to tomograph 100 two-qubit and 128 three-qubit states which were experimentally generated on a nuclear magnetic resonance (NMR) quantum processor. The FFNN model is further used to characterize different quantum processes including two-qubit entangling gates, a shaped pulsed field gradient, intrinsic decoherence processes present in an NMR system, and various two-qubit noise channels (correlated bit flip, correlated phase flip, and a combined bit and phase flip). The results obtained via the FFNN model are compared with standard quantum state and process tomography methods and the computed fidelities demonstrates that, for all cases, the FFNN model outperforms the standard methods for tomography.