Model-independent estimation of the properties of quantum states is a central challenge in quantum technologies, as experimental imperfections, drifts, and imprecise models of the actual quantum dynamics inevitably hinder accurate reconstructions. Here, we introduce a training strategy for photonic quantum extreme learning machines in which both the learning stage and the optimization of the measurement settings are performed entirely with classical light, while inference is carried out on genuinely quantum states. The protocol is based on the identity between the normalized output intensities following the evolution of coherent states through a linear optical reservoir, and the output statistics obtained with separable input quantum states. Building on this correspondence, we implemented a model-free, gradient-based optimization of the reservoir measurement projection directly on experimental data, without relying on a prior model of the device transformation. We experimentally show that the resulting classical-to-quantum transfer enables accurate reconstruction of single-qubit Pauli observables for previously unseen single-photon states, and extends to the estimation of a two-qubit entanglement witness for arbitrary bipartite states. Beyond demonstrating a qualitatively distinct form of out-of-distribution generalization across the classical-to-quantum boundary, our results identify a practical route to fast, adaptive, and resource-efficient training of photonic quantum learning devices.
Model-independent estimation of the properties of quantum states is a central challenge in quantum technologies, as experimental imperfections, drifts, and imprecise models of the actual quantum dynamics inevitably hinder accurate reconstructions. Here, we introduce a training strategy for photonic quantum extreme learning machines in which both the learning stage and the optimization of the measurement settings are performed entirely with classical light, while inference is carried out on genuinely quantum states. The protocol is based on the identity between the normalized output intensities following the evolution of coherent states through a linear optical reservoir, and the output statistics obtained with separable input quantum states. Building on this correspondence, we implemented a model-free, gradient-based optimization of the reservoir measurement projection directly on experimental data, without relying on a prior model of the device transformation. We experimentally show that the resulting classical-to-quantum transfer enables accurate reconstruction of single-qubit Pauli observables for previously unseen single-photon states, and extends to the estimation of a two-qubit entanglement witness for arbitrary bipartite states. Beyond demonstrating a qualitatively distinct form of out-of-distribution generalization across the classical-to-quantum boundary, our results identify a practical route to fast, adaptive, and resource-efficient training of photonic quantum learning devices.
Quantum extreme learning machines (QELMs) are unconventional computing architectures that bear remarkable promise in both classical and quantum machine-learning tasks, such as the estimation of quantum state properties. However, the probabilistic nature of quantum measurements demands extensive repetitions for training to precisely estimate expectation values, imposing stringent trade-offs among experimental resources, acquisition time, and signal-to-noise ratio, particularly for large datasets. Here, we introduce a paradigm shift by training the QELM exclusively with intense classical fields, namely twin beams generated by stimulated emission, while performing inference directly on previously unseen genuine quantum input states to predict their quantum properties. This strategy dramatically reduces acquisition times while substantially enhancing the signal-to-noise ratio. Using frequency-bin encoded biphoton states, implemented here for the first time in a quantum machine learning architecture, we demonstrate entanglement witnessing of two-qubit states with (93 +- 4)% accuracy, multi-dimensional entanglement detection, and learning of the Hamiltonian governing photon-pair generation with a fidelity of (96 +- 4)%. Our results open a new pathway toward faster and more robust training of photonic quantum extreme learning machines for quantum feature extraction.
Recent developments have led to the possibility of embedding machine learning tools into experimental platforms to address key problems, including the characterization of the properties of quantum states. Leveraging on this, we implement a quantum extreme learning machine in a photonic platform to achieve resource-efficient and accurate characterization of the polarization state of a photon. The underlying reservoir dynamics through which such input state evolves is implemented using the coined quantum walk of high-dimensional photonic orbital angular momentum and performing projective measurements over a fixed basis. We demonstrate how the reconstruction of an unknown polarization state does not need a careful characterization of the measurement apparatus and is robust to experimental imperfections, thus representing a promising route for resource-economic state characterization.
Accurately estimating properties of quantum states, such as entanglement, while essential for the development of quantum technologies, remains a challenging task. Standard approaches to property estimation rely on detailed modeling of the measurement apparatus and a priori assumptions on their working principles. Even small deviations can greatly affect reconstruction accuracy and prediction reliability. Here, we demonstrate that quantum reservoir computing embodies a powerful alternative for witnessing quantum entanglement and, more generally, estimating quantum features from experimental data. We leverage the orbital angular momentum of photon pairs as an ancillary degree of freedom to enable informationally complete single-setting measurements of their polarization. Our approach does not require fine-tuning or refined knowledge of the setup, at the same time outperforming conventional approaches. It automatically adapts to noise and imperfections while avoiding overfitting, ensuring robust reconstruction of entanglement witnesses and paving the way to the assessment of quantum features of experimental multiparty states.
Quantum generative learning is a promising application of quantum computers, but faces several trainability challenges, including the difficulty in experimental gradient estimations. For certain structured quantum generative models, however, expectation values of local observables can be efficiently computed on a classical computer, enabling fully classical training without quantum gradient evaluations. Although training is classically efficient, sampling from these circuits is still believed to be classically hard, so inference must be carried out on a quantum device, potentially yielding a computational advantage. In this work, we introduce Fermionic Born Machines as an example of such classically trainable quantum generative models. The model employs parameterized magic states and fermionic linear optical (FLO) transformations with learnable parameters. The training exploits a decomposition of the magic states into Gaussian operators, which permits efficient estimation of expectation values. Furthermore, the specific structure of the ansatz induces a loss landscape that exhibits favorable characteristics for optimization. The FLO circuits can be implemented, via fermion-to-qubit mappings, on qubit architectures to sample from the learned distribution during inference. Numerical experiments on systems up to 160 qubits demonstrate the effectiveness of our model and training framework.
Quantum reservoir computing is a machine-learning framework that offers ease of training compared to other quantum neural networks, as it does not rely on gradient-based optimization. Learning is performed in a single step on the output features measured from the quantum system. Various implementations of quantum reservoir computing have been explored in simulations, with different measured features. Although simulations have shown that quantum reservoirs present advantages in performance compared to classical reservoirs, experimental implementations have remained scarce. This is due to the challenge of obtaining a large number of output features that are nonlinear transformations of the input data. In this work, we show that even with a circuit quantum electrodynamics system as simple as a single transmon coupled to a readout resonator, we can implement a proof-of-concept realization of quantum reservoir computing. We obtain a large number of nonlinear features from a single physical system by encoding the input data in the amplitude of a coherent drive and measuring the cavity state in the Fock basis. We demonstrate classification of two classical tasks with significantly smaller hardware resources and fewer measured features compared to classical neural networks. Our experimental results are supported by numerical simulations that show additional Kerr nonlinearity is beneficial to reservoir performance. Our work demonstrates a hardware-efficient quantum neural-network implementation that can be further scaled up and generalized to other quantum machine-learning models.
[eng] Quantum photonic technologies have achieved considerable success across a range of disciplines, including quantum computing, quantum communications, and quantum metrology. Conversely, in the current era of information, the ability to discern patterns within voluminous data sets is highly sought after, with Machine Learning (ML) having emerged as a powerful solution to address this challenge. The processing speed of photonic hardware, in conjunction with the potential for computation enhancement resulting from quantum properties, positions quantum photonics as a promising platform for the implementation of ML algorithms. In this thesis, we focus on reservoir computing (RC) as an ML framework that is suitable for time series processing. It provides fast and energy-efficient training in comparison to more conventional methods and is also appropriate for physical implementations. In this thesis, we explore the possibilities brought by quantum photonic setups in the field of quantum RC. Specifically, we propose and study different platform designs that are able to perform different temporal benchmark tasks by providing tunable memory and nonlinearity. Furthermore, we leverage the capabilities of state-of-the-art quantum photonic hardware and identify novel strategies to address key challenges in RC in quantum substrates, as well as expand the potential of quantum machine learning. Firstly, an approach is provided to overcome some of the limitations of quantum projective measurements, thereby enabling information processing in real time. This is achieved by generating a physical ensemble of quantum reservoirs that recirculate through an optical fiber inside a feedback loop. We investigate the detrimental effects of sampling noise due to finite ensembles and provide a strategy to mitigate them. Secondly, the consequences of increasing quantum squeezing in the dynamics of such setups were studied, and it was found that it monotonically improves the noise robustness of the platforms in realistic scenarios. Finally, we propose a novel design in which photonic cluster states, a specific kind of highly entangled quantum network, are used as resources to implement quantum RC. In this configuration, external input information is encoded into the quantum network via quantum teleportation by engineering the measurement basis of the protocol. Furthermore, the platform has the capacity to implement global quantum gates through local operations (single-mode measurements), thus establishing the foundation for distributed quantum ML.