MMPD combines Tanner-graph message passing with bidirectional Mamba blocks to deliver 0.45 dB better performance and 1.5x lower memory than attention-based decoders on a 1056-bit LDPC code.
Hybrid mamba- transformer decoder for error-correcting codes
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.IT 2years
2026 2representative citing papers
Code automorphisms used for data augmentation during training and inference allow syndrome-based neural decoders to closely approach maximum likelihood performance on short high-rate codes with small datasets.
citing papers explorer
-
Scalable Mamba-Based Message-Passing Neural Decoder for Error-Correcting Codes
MMPD combines Tanner-graph message passing with bidirectional Mamba blocks to deliver 0.45 dB better performance and 1.5x lower memory than attention-based decoders on a 1056-bit LDPC code.
-
Leveraging Code Automorphisms for Improved Syndrome-Based Neural Decoding
Code automorphisms used for data augmentation during training and inference allow syndrome-based neural decoders to closely approach maximum likelihood performance on short high-rate codes with small datasets.