Neural network-based quantum error correction decoding method and apparatus, chip

摘要

The present application discloses a neural network-based QEC decoding method, apparatus, and chip, and relates to the field of artificial intelligence and quantum technology. The method includes: obtaining error syndrome information of a quantum circuit; performing block feature extraction on error syndrome information using a neural network decoder to obtain feature information; and performing fusion decoding processing on the feature information using a neural network decoder to obtain error result information, wherein the error result information is used to determine a data qubit in the quantum circuit in which an error occurs and a corresponding error type. used - includes . In the present application, a block feature extraction method is used, the quantity of channels of feature information obtained by each feature extraction is reduced, and the input data of the next feature extraction is reduced, which is used to reduce the quantity of feature extraction layers in the neural network. This is helpful, and thus the depth of the neural network decoder can be shortened. Therefore, the decoding time used by the neural network decoder can be reduced accordingly, to meet the requirement of real-time error correction.