Closed BlindApe closed 7 years ago
As stated yonghuajiang, according to (2) formula in the reference paper, 'Practical Bayesian Optimization of Machine Learning Algorithms'
line 28 at Utility.R should be: Utility <- (GP_Mean - y_max - eps) * pnorm(z) + sqrt(GP_MSE) * dnorm(z), (instead of pnorm(z))
EI utility function expression fixed. I didn't merge that pull request because it didn't pass checks. Thank you very much.
As stated yonghuajiang, according to (2) formula in the reference paper, 'Practical Bayesian Optimization of Machine Learning Algorithms'
line 28 at Utility.R should be: Utility <- (GP_Mean - y_max - eps) * pnorm(z) + sqrt(GP_MSE) * dnorm(z), (instead of pnorm(z))