Original title: Virtual quantum neural networks
Benchi Zhao, Xuanqiang Zhao, Yinan Li, Yingzhou Li +1 · September 2026
Quantum computers try to learn patterns using quantum neural networks. Usually, these networks only search through standard quantum operations, which limits what they can learn. To solve this, researchers created virtual quantum neural networks. Instead of running just normal physical setups, this method uses random sampling on quantum circuits alongside classical data processing. This process blends quantum operations together mathematically into linear combinations, creating a much larger search space. By opening up this bigger space, the virtual networks become more expressive, meaning they can represent more complex patterns. They also handle unwanted noise much better. The researchers tested their approach on three tasks: mitigating quantum errors, sorting data into two groups, and estimating ground-state energies of physical systems.
Imagine you're mixing paints but you can only pour from a fixed set of pure colors. You can't create a subtle shade you need. Now suppose you instead let a tiny bit of each paint fall, random amounts, and then blend them all together with a computer. Suddenly you can match any color, even if one paint is slightly muddy. In this paper the researchers use random samples of quantum circuits, combined classically, to achieve computations no single circuit could.
For researchers working with today's noisy quantum computers, this approach makes quantum neural networks more powerful without requiring better hardware. It achieves better error correction and classification by cleverly combining quantum circuits with classical processing. This could make quantum machine learning practical years sooner than expected.
The abstract does not state limitations explicitly. The method relies on random sampling and classical post-processing, which could introduce computational overhead. Benefits are demonstrated on only three specific tasks, so general scalability remains unaddressed.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Virtual quantum neural networks — Benchi Zhao, Xuanqiang Zhao, Yinan Li et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.