Matrix Methods & Applications (DRAFT)
Stephen Boyd (Stanford) and Lieven Vandenberghe advise:
The textbook is still under very active development by Lieven Vandenberghe and Stephen Boyd, so be sure to download the newest version often. For now, we’ve posted a rough draft that does not include the exercises (which we’ll be adding). The first few chapters are in reasonable shape, but later ones are quite incomplete.
The 10 August 2014 draft has one hundred and twenty-two (122) pages so you can assume more material is coming.
I particularly like the “practical” suggested use cases.
The use cases create opportunities to illustrate the impact of data on supposedly “neutral” algorithms. Deeper knowledge of these algorithms will alert you to potential gaming of data that lies behind “neutral” processing of data.
Inspection of data is the equivalent of Mannie’s grandfather’s second rule: “Always cut cards.” (The Moon Is A Harsh Mistress)
Anyone who objects to inspection of data is hiding something. It may be their own incompetence with the data but you won’t know unless you inspect the data.
Results + algorithms + code + data = Maybe we will agree after inspection.
I first saw this in a tweet by fastml extra.