MASC internalizes multi-order n-gram patterns via shared PCM and MMER routing to refine source probabilities, shorten codelengths, and reduce sensitivity to channel errors in SSCC for low-SNR regimes.
Language models are unsupervised multitask learners,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IT 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Contextual Memory-Enhanced Source Coding for Low-SNR Communications
MASC internalizes multi-order n-gram patterns via shared PCM and MMER routing to refine source probabilities, shorten codelengths, and reduce sensitivity to channel errors in SSCC for low-SNR regimes.