GAN decoder on a quantum toric code for noise-robust quantum teleportation

Abstract: Generative adversarial network (GAN) is a strong deep learning model that has shown its value in applications such as image processing and data enhancement. Here, we propose a GAN-based quantum topological toric code decoder and we apply it to devise a quantum teleportation protocol which is robust to depolarizing noisy environments. We construct the generator and discriminator networks of GAN, train the network using the eigenvalue dataset of the toric code, and obtain an optimized decoder with a high decoding threshold compared to some existing decoders. The simulation experiments at code distances $d=3$ and $d=5$ show that the fidelity threshold of this GAN decoder is about $P=0.2108$, which is significantly larger than the threshold $P=0.1099$ of the classical decoding model. Also, the quantum teleportation protocol, optimized for noise resistance under $d=3$ and $d=5$ topological code, shows a fidelity improvement within the depolarizing noise threshold range of $P<0.06503$ and $P<0.07512$, respectively. With appropriate dataset training, the decoder can be adapted to other error models. More broadly, the proposed GAN model provides a novel approach for topological code decoders, offering a versatile framework for different types of noise processing.