Figure 2 | Scientific Reports

Figure 2

From: Convolutional neural networks for approximating electrical and thermal conductivities of Cu-CNT composites

Figure 2

Schematic for the data preparation. The image data size of Cu-CNT networks is reduced by converting RGB scale into grayscale. The Cu-CNT interfacial resistance is encoded by the selection of Cu domain color intensity. The resistance levels are defined as follows. Level 1: Rc = 20 kΩ or R″t,c = 10−7 m2K/W, Level 2: Rc = 2 kΩ or R″t,c = 10−8 m2K/W, Level 3: Rc = 200 Ω or R″t,c = 10−9 m2K/W, and Level 4: Rc = 20 Ω or R″t,c = 10−10 m2K/W. The images are flipped horizontally, vertically, and diagonally to amplify the training dataset by four folds.

Back to article page