POUR derives a provably optimal forgetting operator by showing that orthogonal projections of simplex equiangular tight frames remain ETFs in lower dimensions, enabling representation-level unlearning with closed-form and distillation variants.
Chundawat, Ayush K
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
POUR: A Provably Optimal Method for Unlearning Representations via Neural Collapse
POUR derives a provably optimal forgetting operator by showing that orthogonal projections of simplex equiangular tight frames remain ETFs in lower dimensions, enabling representation-level unlearning with closed-form and distillation variants.