Fig. 8: Machine-learning prediction.

a Learning curve of the neural network. The errors are averaged over 25 randomized runs and the error bars indicate the 95% confidence interval; b parity plots between predicted and DFT-calculated ΔGH values.
a Learning curve of the neural network. The errors are averaged over 25 randomized runs and the error bars indicate the 95% confidence interval; b parity plots between predicted and DFT-calculated ΔGH values.