GE of model is based on Loss and does not consider e.g. AUC

[giuseppec]

In slides-evaluation-generr.tex, the first slide on "GE for a fixed model" is based on a Loss (formally, this definition does not allow for AUC). Later, you introduced $\rho$ and use this notation to define the "GE for inducer" so that it allows for perf. measures such as AUC but why not already introducing this in the first slide when you explain the GE for a fixed model? You could directly introduce $\rho$ in the very beginning.

[scholbeck]

Also, the data set to be evaluated is missing as a parameter / function input in the GE estimate.