Toward Uncertainty-Aware and Generalizable Neural Decoding for Quantum LDPC Codes

The paper introduces QuBA, a Quantum Bayesian graph Attention decoder that provides expressive error-pattern recognition and calibrated uncertainty estimates for quantum error correction. It also presents SAGU, a multi-phase training framework that enhances cross-domain robustness, allowing decoding beyond the training set. Experiments on bivariate bicycle codes show that both QuBA and SAGU outperform classical belief propagation, achieving up to two orders of magnitude lower logical error rates and comparable or better performance than domain-specific training approaches.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.