Rigorous noise reduction with quantum autoencoders

Abstract

Reducing noise in quantum systems is a significant challenge in advancing quantum technologies. We propose and demonstrate a noise reduction scheme utilizing a quantum autoencoder, which offers rigorous performance guarantees. The quantum autoencoder is trained to compress noisy quantum states into a latent subspace and eliminate noise through projective measurements. We identify various noise models in which the noiseless state can be perfectly reconstructed, even at high noise levels. We apply the autoencoder to cool thermal states to the ground state and reduce the cost of magic state distillation by several orders of magnitude. Our autoencoder can be implemented using only unitary transformations without the need for ancillas, making it immediately compatible with state-of-the-art quantum technologies. We experimentally validate our noise reduction methods in a photonic integrated circuit. Our results have direct applications in enhancing the robustness of quantum technologies against noise.

Copyright and License

Acknowledgement

This work was supported by a Samsung GRC project and the UKRI EPSRC Grant Nos. EP/W032643/1 and EP/Y004752/1. The authors thank Jielun Chen, Hsin-Yuan Huang, and John Preskill for insightful discussions.

Contributions

Wai-Keong Mok, Hui Zhang, and Tobias Haug contributed equally to this work.

Wai-Keong Mok: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Hui Zhang: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Tobias Haug: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Xianshu Luo: Methodology (equal); Resources (equal); Writing – original draft (equal); Writing – review & editing (equal). Guo-Qiang Lo: Methodology (equal); Resources (equal); Writing – original draft (equal); Writing – review & editing (equal). Zhenyu Li: Methodology (equal); Resources (equal); Writing – original draft (equal); Writing – review & editing (equal). Hong Cai: Methodology (equal); Resources (equal); Writing – original draft (equal); Writing – review & editing (equal). M. S. Kim: Funding acquisition (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Ai Qun Liu: Funding acquisition (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Leong-Chuan Kwek: Funding acquisition (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal).

Data Availability

The data that support the findings of this study are available from the author on reasonable request.

Conflict of Interest

The authors have no conflicts to disclose.

Files

Embargoed

The files will be made publicly available on May 31, 2025.

	All versions	This version
Views	0	0
Downloads	0	0
Data volume	0 Bytes	0 Bytes