Blog Notes on learning theory, optimization, neural networks, and related theory. 2026 “Row Normalization and Orthogonalization Are Asymptotically Equivalent on Neural Networks.” Related Note: Matrix factorization Deep linear network 2-layer ReLU May 8, 2026.
“Row Normalization and Orthogonalization Are Asymptotically Equivalent on Neural Networks.” Related Note: Matrix factorization Deep linear network 2-layer ReLU May 8, 2026.