Tag
collapse
- Jun 9, 2025 Collapse and Redundancy Reduction in Self-Supervised Training Why self-supervised objectives tend toward constant or low-rank outputs, how whitening and decorrelation terms restore full rank, and what each approach costs in compute and stability. deep learning
- Jan 3, 2022 Alternation and the Structure of the Polynomial Hierarchy How stacking existential and universal quantifiers over polynomial-length certificates produces a hierarchy of classes, why it collapses if any two levels coincide, and what complete problems look like at each level. computation theory