Neural network code

Neural network code[1–3]

An approximate code obtained from a numerical optimization involving a reinforcement learning agent.

Depends on the parameter being optimized.

Neural network codes can be obtained by optimizing the coherent information [2].

Kitaev surface code — Reinforcement learners can be used to optimize the geometry of the surface code to be more suited to a noise channel [3].
Quantum Lego code — Quantum Lego and more general tensor-network code optimization can be done using reinforcement learning [4,5].
Numerically optimized bosonic code — Numerically optimized bosonic codes can be obtained via reinforcement learning [6,7].

[1]: T. Fösel et al., “Reinforcement Learning with Neural Networks for Quantum Feedback”, Physical Review X 8, (2018) arXiv:1802.05267 DOI
[2]: J. Bausch and F. Leditzky, “Quantum codes from neural networks”, New Journal of Physics 22, 023005 (2020) arXiv:1806.08781 DOI
[3]: H. P. Nautrup et al., “Optimizing Quantum Error Correction Codes with Reinforcement Learning”, Quantum 3, 215 (2019) arXiv:1812.08451 DOI
[4]: V. P. Su et al., “Discovery of Optimal Quantum Error Correcting Codes via Reinforcement Learning”, (2023) arXiv:2305.06378
[5]: C. Mauron, T. Farrelly, and T. M. Stace, “Optimization of Tensor Network Codes with Reinforcement Learning”, (2023) arXiv:2305.11470
[6]: Z. Wang et al., “Automated discovery of autonomous quantum error correction schemes”, (2021) arXiv:2108.02766
[7]: Y. Zeng et al., “Approximate Autonomous Quantum Error Correction with Reinforcement Learning”, Physical Review Letters 131, (2023) arXiv:2212.11651 DOI

Page edit log

“Neural network code”, The Error Correction Zoo (V. V. Albert & P. Faist, eds.), 2024. https://errorcorrectionzoo.org/c/reinforcement_learning