Affine transformation and train S samples from low fidelity and 1 sample among the low fidelity observations (g = af + b where g is the high-fidelity func and f is the low fidelity func. Basically, they are discussing how you train it)
To pick among low-fidelity observations, we use UCB.
Multi-level Training and Bayesian Optimization for Economical Hyperparameter Optimization
Affine transformation and train S samples from low fidelity and 1 sample among the low fidelity observations (g = af + b where g is the high-fidelity func and f is the low fidelity func. Basically, they are discussing how you train it)
To pick among low-fidelity observations, we use UCB.
Main points
Confusing points