Denoising quantum states with Quantum Autoencoders -- Theory and Applications
This work addresses the problem of noise in quantum states and protocols for quantum computing researchers, demonstrating a method to mitigate noise and enable robust quantum applications.
This paper implements a Quantum Autoencoder (QAE) as a quantum circuit to correct noisy Greenberger-Horne-Zeilinger (GHZ) states, achieving almost perfect reconstruction. Surprisingly, the QAE can also act as a generative model for noise-free GHZ states and enables a Quantum Secret Sharing (QSS) protocol to succeed despite noise.
We implement a Quantum Autoencoder (QAE) as a quantum circuit capable of correcting Greenberger-Horne-Zeilinger (GHZ) states subject to various noisy quantum channels : the bit-flip channel and the more general quantum depolarizing channel. The QAE shows particularly interesting results, as it enables to perform an almost perfect reconstruction of noisy states, but can also, more surprisingly, act as a generative model to create noise-free GHZ states. Finally, we detail a useful application of QAEs : Quantum Secret Sharing (QSS). We analyze how noise corrupts QSS, causing it to fail, and show how the QAE allows the QSS protocol to succeed even in the presence of noise.