Quality function: compare sum of squared difference between two weight vectors

[mainwaringb]

Another important quality function will be checking the sum of squared differences between weights. If two sets of weighting targets produces very similar weights, then the decision between these weighting targets is trivial.

This is conceptually similar to seeing whether two models produce very similar predictions for y; in the model-motivated approach, we would generally prefer a simpler model unless the complex model produces sufficiently different predictions.