Contrastive pair presentations yield exact identifiability characterizations via a geometric refinement of Angluin's condition, a new contrastive closure dimension for generation, mutual incomparability with text identification, and a single algorithm that tolerates any finite corruption budget.
Safe language generation in the limit
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Defines mistake-bounded generation and gives an algorithm for finite classes achieving optimal last-mistake time Cdim(L) with floor(log2 |L|) mistakes, plus a trade-off for infinite classes and noisy extensions.
citing papers explorer
-
Contrastive Identification and Generation in the Limit
Contrastive pair presentations yield exact identifiability characterizations via a geometric refinement of Angluin's condition, a new contrastive closure dimension for generation, mutual incomparability with text identification, and a single algorithm that tolerates any finite corruption budget.
-
Mistake-Bounded Language Generation
Defines mistake-bounded generation and gives an algorithm for finite classes achieving optimal last-mistake time Cdim(L) with floor(log2 |L|) mistakes, plus a trade-off for infinite classes and noisy extensions.