Quantum
Advantage of Sample Complexity in Quantum PAC Learning Requires Inverse Access to State-Preparation Unitaries
Publication: quant-ph updates on arXiv.org
Author: Natsuto Isogai, Satoshi Yoshida, Mio Murao
Source published: 01/10/2026, 04:00:00 UTC
Briefing created: 01/10/2026, 04:22:21 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 01/10/2026, 04:13:25 UTC.
Briefings
A 2026 study by researchers at arXiv demonstrates that in quantum PAC learning, forward-only access to a quantum state's preparation unitary does not improve sample complexity compared to classical or quantum data copies. The research establishes that optimal query complexities for realizable and agnostic learning are both Theta((d+log(1/delta))/epsilon), matching classical bounds. This finding implies that inverse access to the state-preparation unitary is essential for achieving known improvements in realizable settings, while forward-only access remains limited to classical-like performance.
Key points
- Quantum PAC learning with only state copies achieves optimal sample complexity matching classical learning.
- Forward-only access to state-preparation unitaries yields query complexities Theta((d+log(1/delta))/epsilon) for realizable learning, identical to classical bounds.
Context and limitations
Results apply to the worst-case scenario over compatible state-preparation unitaries and finite ambient dimensions. The study does not address forward-only access in realizable learning settings where inverse access is available, nor does it cover scenarios beyond the specified VC dimension and accuracy parameters.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 01/10/2026, 04:22:21 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeicdbqx7jxc2hfnn7q53h3rgpiculy4mkgcsjafc6zq4aszbnv7ifm