EOMR improves regression accuracy in chaotic systems by identifying relevant feature subsets and subspaces.
problem Learning relevant feature subsets and subspaces in nonstationary and nonlinear regression problems.
method Jointly identifies relevant feature subsets and subspaces using Entropy-Optimal Manifold Regression (EOMR).
result EOMR achieves orders of magnitude better prediction accuracy than state-of-the-art AI and ML tools.