Skip to main content
. 2023 Oct 9;24(19):15009. doi: 10.3390/ijms241915009

Table 3.

Performance of Keras/TensorFlow NN in the chemokine receptor subtype selectivity tasks.

Training Set Number of Datapoints Validation Set Number of Datapoints Loss Accuracy (Change) Average Tanimoto Coefficient
Training vs. Validation Set
Mode Tanimoto Coefficient
Training vs. Validation Set
CCR2 1995 CCR2 399 5.406 0.190 0.139 0.16
CCR3 121 9.621 0.231 (+0.041) 0.141 0.16
CXCR3 199 17.878 0.126 (−0.064) 0.125 0.13
CCR3 603 CCR3 121 5.332 0.223 0.243 0.15
CCR2 399 14.013 0.115 (−0.108) 0.142 0.10
CXCR3 199 14.853 0.191 (−0.032) 0.143 0.13
CXCR3 994 CXCR3 199 4.727 0.322 0.198 0.15
CCR2 399 13.681 0.125 (−0.197) 0.124 0.13
CCR3 121 8.029 0.182 (−0.140) 0.142 0.13