Fig. 2: Pre-training performance on the source dataset. | Nature Communications

Fig. 2: Pre-training performance on the source dataset.

From: Visualizing nexus of porous architecture and reactive transport in heterogeneous catalysis by deep learning computer vision and transfer learning

Fig. 2

a Framework and b Learning curves for Generator 1. c Framework and d Learning curves for Generator 2. e Statistical information and f Visual information for original and reconstructed 3D models. g Predicted results for normalized effective reaction rate (Rnorm) at different Damköhler number (Da) and Péclet number (Pe). h Predicted results and error for the normalized concentration field (cid). Normalized concentration field (cid) was calculated by cid=ci/c0, where ci and c0 (mol cm−3) represented the local and initial concentration, respectively. Then, the local reaction rate, effective reaction rate, and Rnorm could be calculated from Eqs. 12–14. i Heat map for Generator 2. Source data are provided as a Source Data file.

Back to article page