Adversaries perturbing shared entanglement in distributed VQAs can manipulate a new Kraus expressibility metric to keep gradients large but steer training to incorrect solutions.
Barren plateaus in quantum neural network training landscapes.Nature communications, 9(1):4812
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
fields
quant-ph 5verdicts
UNVERDICTED 5representative citing papers
SGIR-QAOA uses spectral gap information to construct QAOA parameter schedules, showing better performance than linear-ramp QAOA on Grover's problem at fixed depth, shorter depths for equivalent success probability, extension to MIS, and persistence under extrapolated gaps and mild noise.
Random states from symplectic and orthogonal unitaries show exponentially large strong state complexity and near-orthogonality, with average-case hardness for learning circuits from these groups.
A QAOA-based hybrid method for chance-constrained knapsack problems in insurance achieves performance comparable to classical optimization on IBM quantum hardware with up to 150 qubits.
A quantum framework introduces C-Estimator and E-Estimator for classical covariance matrices using variational circuits, with regularization to ensure positive definiteness and mitigate barren plateaus, validated via simulations.
citing papers explorer
-
Adversarial Effects on Expressibility and Trainability in Distributed Variational Quantum Algorithms
Adversaries perturbing shared entanglement in distributed VQAs can manipulate a new Kraus expressibility metric to keep gradients large but steer training to incorrect solutions.
-
Spectral Gap Informed Ramp QAOA
SGIR-QAOA uses spectral gap information to construct QAOA parameter schedules, showing better performance than linear-ramp QAOA on Grover's problem at fixed depth, shorter depths for equivalent success probability, extension to MIS, and persistence under extrapolated gaps and mild noise.
-
A Quantum Approach to Stochastic Optimization in Insurance Underwriting
A QAOA-based hybrid method for chance-constrained knapsack problems in insurance achieves performance comparable to classical optimization on IBM quantum hardware with up to 150 qubits.
-
Quantum Learning of Classical Correlations with continuous-domain Pauli Correlation Encoding
A quantum framework introduces C-Estimator and E-Estimator for classical covariance matrices using variational circuits, with regularization to ensure positive definiteness and mitigate barren plateaus, validated via simulations.