CNN performance for various convergence regions’ selections. Left panel shows σ_S8, which is the constraining power on S₈; right panel shows H, which is the information entropy. Low, mid, and high κ denote the low-, mid-, and high-convergence regions. The second row presents the performance of a network taking the high-convergence regions in one channel and the Fourier transform amplitude in another to mimic a summary statistics analysis method capable of extracting non-Gaussian information: the combination of the PS and peak counts or Minkowski functionals.