The paper proves equivalence of Fisher objectives under multilabel scatter constraints, a rank characterization allowing discriminant dimensionality beyond C-1, and near-minimax optimal finite-sample bounds for multilabel LDA subspace estimation under sub-Gaussian noise.
Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , year =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On the Spectral Structure and Objective Equivalence of Orthogonal Multilabel Fisher Discriminants
The paper proves equivalence of Fisher objectives under multilabel scatter constraints, a rank characterization allowing discriminant dimensionality beyond C-1, and near-minimax optimal finite-sample bounds for multilabel LDA subspace estimation under sub-Gaussian noise.