Bregman conditional random fields use mean regularization and iterative Bregman projections to make sequence labeling inference GPU-parallel, matching CRF accuracy at higher speed.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Bregman Conditional Random Fields: Sequence Labeling with Parallelizable Inference Algorithms
Bregman conditional random fields use mean regularization and iterative Bregman projections to make sequence labeling inference GPU-parallel, matching CRF accuracy at higher speed.