Bluh, had the misfortune of gmail opening that one for me. So much for succesfully avoiding that image for the past 20 years.

Interesting to see that at the same time the EU privacy watch dog is proposing to mandate encryption and outlaw these kind of decryption methodologies[0]. While still an opinion, it is good to see that in this area the EU is among the most progressive governments around. [0] https://secure.edps.europa.eu/EDPSWEB/webdav/site/mySite/sha...

Interesting indeed. I thought it peculiar that the UK government released the Tony Blair report just before the BRexit vote. That report certainly got many people riled up about the 'establishment' and possibly could have changed the BRexit voting result. My point is that if the UK had stayed in the EU then the UK surveillance bill might have been dead in the water because it might have comflicted with EU law.

The EU referendum was on the 23rd June 2016 and the Chilcot report (as it is known formerly) was released 6th of July 2016. So those facts aside your point, do you have any conflicting EU laws in mind? I ask as currently the UK has not left and then looking at 2 years notice so any new UK law coming into play for years (due to the time to formulate them), will be complying within current EU laws and regulations.

TL;DR No and no. But I respect the premise.

Thank you for the correction! I have made this cement to some friends, so I will let them know I was wrong.

Whenever I see some of these 'countermeasures' I usually run them through some high end commercial face recognition algorithms I have access to. In many of the cases the countermeasures are completely ineffective. They may fool some things like opencv DIY algorithms, but not the expensive stuff :). For those the most effective strategy is to just tilt your head to the left or right a lot, though that will possibly give you a sore neck and you'll definitely look silly.

just shave off your eyebrows, or makeup to make them appear gone. Most (nearly all) facial feature finders expect eyebrows of some kind. (I work in facial recognition.)

> just tilt your head to the left or right a lot

That seems like an trivially simple thing to defeat from a code perspective.

Or just wear an Islamic veil with some sunglasses?

I think one of the caveats was that the function should be relatively smooth and continuous, a mandelbrot set isn't very smooth is it?

Smoothness isn't necessary to prove the result, just continuity.

You do need smoothness to prove a bound on the rate of convergence of the basis representation, and given that the boundary of the Mandelbrot doesn't have a closed form representation (as far as I'm aware), I think that the convergence of the neural network representation would be extremely slow.