Entropy-Optimal Manifold Regression for Nonstationary Nonlinear Problems
July 31, 2026
The proposed Entropy-Optimal Manifold Regression (EOMR) enables joint identification of feature subsets and subspaces in nonlinear regression. The method maintains linear scaling for both iteration and memory complexity, demonstrating robustness on Lorenz-96 chaotic systems and Hasegawa-Wakatani tokamak plasma models.
HOW THIS AFFECTS YOU
●
researcherYou can apply this method to high-dimensional chaotic systems where memory and computational scaling are critical constraints.