It always depends on your problems, your targets and your metrics. The most critical aspect for many industrial applications is false positives, so that machine

It always depends on your problems, your targets and your metrics. The most critical aspect for many industrial applications is false positives, so that machine learning extreme / redundant optimisation becomes cheaper for the business in the long term.