multi-variate

Okay, just in regression and multi-variate statistics, just from my bookshelf, there's: William W.

Through that inventory, we can spin any text and images for multi-variate testing.

We apply engineering and data science to tough problems such as dynamic pricing, shipping estimations, personalized emails, and multi-variate testing.

That is, we want a detector that is multi-variate.

The point of my post here was to urge the authors of the OP and also others in anomaly detection to address, for their work, carefully, the issues of the error rates and being both distribution-free and multi-variate.

But, in practice, the usual situation, e.g., signal processing, multi-variate statistics, there's no good reason not to use L2 and many biggie reasons to use it.

So, we want to be able to adjust and know false alarm rate in advance, get for that false alarm rate a rate of missed detections, be distribution-free and multi-variate, and have essentially no data on the cases of sickness we are trying to detect.

I've indicated likely most of the main points, but my answer here is rough and ready (I typed too fast), and a quite polished answer is also possible -- I just don't have time today to dig out my grad school course notes, scan through Rudin, Dunford and Schwartz, Kolmogorov and Fomin, much of digital filtering, much of multi-variate statistics, etc.