Q-Capsule: A Localized Capsule-Based Quantum Neural Architecture for Barren Plateau Mitigation

2026-10-08

AI Summary

This study addresses the barren plateau problem caused by circuit scaling in variational quantum algorithms by proposing the Q-Capsule architecture. To maintain gradient stability, it restricts entanglement scope through register partitioning, local readout, and sparse coupling. Furthermore, a quantum Fisher information matrix (QFIM)-guided adaptive depth growth mechanism is introduced to balance representational capacity with training stability, while integrating capsule networks with trainable data re-uploading techniques. Experimental results demonstrate that the proposed method achieves binary and multi-class classification accuracies of 98.1% and 97.7%, respectively, while reducing the number of two-qubit gates by approximately 73%. These findings indicate significant improvements in both parameter efficiency and noise robustness for variational quantum circuits.

Abstract

Variational quantum algorithms are often limited by barren plateaus: gradients vanish as circuit size and depth increase, making quantum neural networks difficult to train. We propose Q-Capsule, a localized capsule-based quantum neural architecture that mitigates this problem through register partitioning, local readout, sparse inter-capsule coupling, trainable data re-uploading, and Quantum Fisher Information Matrix (QFIM)-guided adaptive depth growth. By restricting the dominant support of each observable to a small capsule and controlling inter-capsule entanglement, Q-Capsule preserves useful gradient signals while retaining communication between local quantum representations. As the register width increases, Q-Capsule consistently maintains stable gradient variance, whereas globally entangling baselines exhibit exponential suppression with a log-gradient-variance slope near -ln 2 per qubit. Q-Capsule also produces more structured optimization landscapes, higher parameter efficiency, improved robustness to depolarizing noise, and lower measurement requirements. Its adaptive policy achieves 98.1% accuracy on binary classification and 97.7% on four-class classification, while using approximately 73% fewer two-qubit gates than the fixed-deep model on the multiclass task.

Problem

• Barren plateaus
• Variational quantum algorithms
• Quantum neural networks
• Vanishing gradients

Innovation

• Quantum Neural Architecture
• Barren Plateau Mitigation
• Capsule Network
• Quantum Fisher Information Matrix
• Adaptive Depth Growth