Quantum

Experimental visual quantum state discrimination

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 10/11/2026, 08:43:24 UTC.

Briefings

A new AI-driven approach using diffractive optical neural networks (DONN) optimizes quantum measurements for computer vision tasks. The system learns to extract maximum information from incoming photons, achieving exponentially lower discrimination errors compared to direct imaging or single-photon-optimized detectors. Validation was conducted in both noiseless settings and a photon-counting experiment using an electron-multiplying CCD with noise equivalent to 10,000 photons per signal photon.

Key points

• The system uses a diffractive optical neural network (DONN) trained with a quantum information-theoretic objective to maximize the rate of discrimination error decay as photons accumulate.
• The resulting measurement achieves exponentially lower error than both direct imaging and a DONN optimized for single-photon discrimination in both noiseless and noisy photon-counting experiments.

Context

This research addresses the suboptimal nature of traditional vision tasks that rely on capturing images with optical instruments, which leads to irreversible information loss in photon-limited and noisy conditions. The study introduces AI-driven optimization of quantum measurements to overcome these limitations.

Limitations

Validation was performed in a noiseless setting and a photon-counting experiment with an electron-multiplying CCD; the specific noise characteristics of the real-world application environment are not fully characterized.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 10/11/2026, 10:19:00 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeiabhfkcituskf37bud5o35grmvuzlkvxae6linhf7txiwsimmqh7y