Presents an efficient PAC algorithm for multiclass linear classifiers under nasty noise with sample complexity O(k² (d log d + log k)) when the marginal is a mixture of bounded-variance distributions satisfying a margin condition.
Lipton , title =
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
-
Efficient and Noise-Tolerant PAC Learning of Multiclass Linear Classifiers
Presents an efficient PAC algorithm for multiclass linear classifiers under nasty noise with sample complexity O(k² (d log d + log k)) when the marginal is a mixture of bounded-variance distributions satisfying a margin condition.