EviTrack is a test-time inference framework that performs selection over latent trajectory hypotheses rather than marginal sampling to handle delayed disambiguation, outperforming baselines on a synthetic benchmark.
IEEE Transactions on Aerospace and Electronic Systems , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
EviTrack: Selection over Sampling for Delayed Disambiguation
EviTrack is a test-time inference framework that performs selection over latent trajectory hypotheses rather than marginal sampling to handle delayed disambiguation, outperforming baselines on a synthetic benchmark.