Issues found by Shuhei

[dengdifan]

• Chapter 6: Bayesian Optimiziation

• • Cheap Acquisition functions
• • • page 9: Only PI uses different time notation from others (t - 1 --> t ?)
• • • Page 19: Yes, I agree that c(λ) and \hat{c}(λ) are mixed up sometimes (when we define EI)

• Chapter 8:

• • In evaluation of solutions I, i think the argument of S-metric should be F (pareto Front set), but not pareto Set P and thus λ in the slide must be c(λ) or f(λ)
• • Other errors in bayesian optimization of multi criteria optimization (i might be checking old slides, so please ignore in this case)
• • [Scalarization: ParEGO] |W| = s+m choose m - 1, but not s+m-1 choose k - 1
• • [Why the Tchebycheff norm?] Cost function with maximum weight into account --> confusing, it's not determined by only weight, right?"

• Chapter 11

• • 1.Overview
• • • page 5 --9: typo. might can

• Chapter 11

• • 4. Population-based training
• • • page 14 --16, 25: typo. pertubation --> perturbation

• Chapter 12: Interpretability

• • 3. Local importance
• • • page 9: Enhancement. The referencing to the original ablation study paper
• • • page 10 -- 13: Typo. which choices where --> which choices were
• • • page 24: Typo. (s1) --> (s2)
• • • page 24: Enhancement. The blue numbers are hard to see for me and it helps if the slide explicitly says those blue numbers are sorted in the importance order