Abstract:Measuring complex properties in quantum systems, such as various measures of entanglement and Bell nonlocality, is challenging. Traditional methods require a full reconstruction of the density matrix of a given system. An alternative strategy reduces information by using multiple system copies and joint measurements. However, this approach has been considered so far impractical due to difficulties in preparing and controlling multiple quantum-correlated systems and the amplification of experimental noise. Despite these challenges, we have successfully measured two--qubit quantum correlations of Horodecki's and Werner's states using a multicopy approach. We report on experiments with transmon qubits using IBMQ quantum processors to quantify the violation of Bell's inequality and the negativity of a two--qubit entangled state. We compared these results with those from the standard tomography-based approach and developed a maximum likelihood method to reduce the noise. We trained artificial neural networks (ANNs) to estimate entanglement and nonlocality measures from the top five experimental configurations for Werner's and Horodecki's states. The ANN output, based on a reduced set of projections, aligns with expected values and enhances noise robustness. This could simplify and improve the error-robustness of multicopy measurements without complex nonlinear equation analysis. This approach represents a significant advancement in AI-assisted measurements of quantum systems.

Experimental neural network enhanced quantum tomography

Predictive Modelling of Quantum Process with Neural Networks

Improved Tomographic Estimates by Specialised Neural Networks

Enhancing quantum state tomography via resource-efficient attention-based neural networks

Neural network assisted quantum state and process tomography using limited data sets

Neural networks for quantum state tomography with constrained measurements

Deep Neural Network-assisted improvement of quantum compressed sensing tomography

On how neural networks enhance quantum state tomography with constrained measurements

Data Driven Non-Markovian Quantum Process

Machine-learning-assisted Efficient Reconstruction of the Quantum States Generated from the Sagnac Polarization-Entangled Photon Source

Experimental demonstration of enhanced quantum tomography via quantum reservoir processing

Local-measurement-based Quantum State Tomography Via Neural Networks

Machine-learning-assisted effcient reconstruction of the quantum states generated from the Sagnac polarization-entangled photon source

Quantum State Tomography of Multi-Qubit Systems — a Comparative Study

Quantum Network Tomography

Neural Network Enhanced Single-Photon Fock State Tomography

Neural-Shadow Quantum State Tomography

Neural network based time-resolved state tomography of superconducting qubits

AI-assisted experimental measures of entanglement and Bell's nonlocality without state tomography

Efficient learning of mixed-state tomography for photonic quantum walk

Experimentally exploring compressed sensing quantum tomography