TBNQuantum
Federated Unlearning of Quantum Data via Classical Shadows
Publication: quant-ph updates on arXiv.org
Author: Zihao Ding, Jun Huang, Cong-cong Xing
Source published: 15/10/2026, 04:00:00 UTC
Briefing created: 15/10/2026, 04:28:38 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 15/10/2026, 04:13:24 UTC.
Briefings
Researchers propose SHADE, a method for federated unlearning of quantum neural networks using classical shadows. By recording classical measurements of observables at training time, the method estimates gradients and curvature for forgotten data, allowing nodes to remove their influence without replaying quantum states. Simulation results show SHADE achieves forget accuracy within 0.5 points of retraining on 50-node edge boards in under 0.2 seconds.
Key points
- SHADE uses classical shadows of quantum states to estimate gradients and curvature for unlearning purposes, avoiding the need to replay quantum data.
- The shadow size required for a target unlearning error grows exponentially with the backward light cone of the observable, not with the number of qubits.
Context and limitations
Simulation results are limited to 50 nodes and edge boards; real-world performance on larger networks remains unverified.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 15/10/2026, 04:28:38 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeifeohd2lh4afmetqwn3qtxgmcpz27twk6nbrdlrius2hn23basswe