arXiv:2606. 14822v1 Announce Type: cross Abstract: Recent advances in Machine Learning have transformed numerous industrial sectors, yet classical paradigms face fundamental limitations: rapidly growing data volumes, rising computational costs, significant energy consumption, and the physical scaling limits of conventional hardware architectures.
arXiv:2605. 28690v2 Announce Type: replace-cross Abstract: Many applications in quantum simulation, quantum chemistry, and quantum machine learning require not a single quantum state but an ensemble of states characterizing the heterogeneity of a target system.
arXiv:2606. 31536v1 Announce Type: new Abstract: As Quantum Machine Learning (QML) transitions toward practical implementation, the field faces a critical architectural bottleneck that challenges the fundamental assumptions of classical statistical learning theory.
arXiv:2606. 10179v1 Announce Type: cross Abstract: Quantum Circuit Born Machines (QCBMs) offer a natural approach to generative machine learning by leveraging the Born rule.
arXiv:2608. 19497v1 Announce Type: new Abstract: We characterize the learning dynamics of a compact hybrid quantum forecasting model through comparison with a structurally aligned classical baseline.
The paper introduces HELIA, a Hardware‑Efficient and Dynamical Lie‑Algebra Supported Ansatz, and two hybrid training methods that combine classical‑enhanced gradient computation with the quantum Parameter‑Shift Rule. By distributing gradient estimation between classical and quantum hardware, the authors reduce quantum‑resource usage while improving accuracy in Variational Quantum Eigensolver (VQE) ground‑state estimation for 6‑ to 18‑qubit Hamiltonians and in quantum phase classification for up to 12 qubits. The methods achieve up to a 60 % reduction in logical quantum evaluations for VQE and a 2.8‑percentage‑point gain in peak test accuracy for classification, suggesting a promising path toward training larger‑qubit quantum models.
arXiv:2607. 05000v1 Announce Type: cross Abstract: Canonical quantization provides a systematic procedure for constructing quantum models from classical Hamiltonians.
The paper presents a method for learning the mapping from combinatorial optimization problem instances to high‑quality variational angles for the Quantum Approximate Optimization Algorithm (QAOA). By training models that predict these angles, the authors replace the expensive per‑instance classical optimization with a single forward pass, achieving near‑identical solution quality while being several orders of magnitude faster. The learned models generalize to unseen graph topologies, and experiments on quantum hardware confirm that the sampled solutions are indistinguishable from those obtained with traditional solver‑derived angles.
The paper introduces a variational framework that combines a classical neural network with a parameterized quantum circuit to solve Lindblad master equations. It represents the density matrix via purification, guaranteeing physical properties, and evaluates gradients exactly using the parameter‑shift rule. The authors also derive an upper bound on the covering number of this compositional model, compare it to purely quantum neural networks, and provide numerical evidence supporting both accuracy and theoretical predictions about gradient concentration.
arXiv:2606. 09734v1 Announce Type: cross Abstract: Training parameterised quantum circuits (PQCs) on quantum hardware is bottlenecked by the measurement cost of gradient estimation, which under the parameter-shift rule scales linearly in the number of trainable parameters and dominates the total shot budget of training at scale.
The paper investigates whether quantum reinforcement learning algorithms can be matched by efficient classical methods. It focuses on a simplified reinforcement learning setting with a uniform generative model, providing finite‑sample guarantees for classical kernelized Fitted Q‑Iteration that uses kernels aligned with parameterized quantum circuits. The authors identify sufficient conditions on data encoding, kernel choice, and problem structure under which this classical approach dequantizes quantum Q‑learning, and suggest using kernelized Fitted Q‑Iteration as a heuristic when those conditions cannot be verified.
arXiv:2606. 15983v1 Announce Type: cross Abstract: Recent theoretical progress has established conditions under which machine learning models can efficiently predict ground-state properties of gapped local Hamiltonians when trained on quantum-generated data.