Adaptive neural-network decoding for quantum error correction

Abstract

Quantum error correction is one of the key requirements for reliable large-scale quantum computation, while efficient decoding of syndrome measurements remains an important practical challenge. Neural-network-based decoders provide a promising alternative to conventional decoding algorithms, as they can learn complex spatial and temporal correlations directly from error-correction data. In this work, a neural-network approach to decoding the surface code is considered. The decoder is based on a recurrent neural network architecture that processes sequences of stabilizer measurements and predicts logical Pauli errors affecting the encoded qubit. The current implementation is studied for a distance-3 rotated surface code and is designed to distinguish between the logical I, X, Y, and Z outcomes. Synthetic error-correction data generated under circuit-level noise models are used for training and evaluation. Particular attention is paid to the ability of the neural decoder to capture correlated error patterns and to its robustness with respect to variations in the underlying physical noise model.

Speaker

Daniil S. Bagaev
Russian Quantum Center
Russia

Discussion

Ask question