Oct 2026
Adaptive parameter manifold learning for low-dimensional neural network optimization.
This work proposes Adaptive Parameter Manifold Learning (APML), a novel optimization framework that leverages Local Tangent Space Alignment to discover the underlying manifold structure from training trajectories and optimize neural networks on the parameter manifolds.
Zhe-Kai Du, Wei Ao, Jie-Yan Liu et al.
· Neural Networks · 0 citations