Facebook's '10 Year Challenge' Meme Could Train Facial Recognition Algorithms On Age Progression, Age Recognition (wired.com) 38
If you've spent any time on social media lately, you've probably noticed a trend where users are posting their then-and-now profile pictures, mostly from 10 years ago and this year. While this "10 Year Challenge" appears harmless, founder of KO Insights and the author of Tech Humanist, Kate O'Neill, says all this data "could be mined to train facial recognition algorithms on age progression and age recognition." She adds: "It's worth considering the depth and breadth of the personal data we share without reservations." From the report: Imagine that you wanted to train a facial recognition algorithm on age-related characteristics, and, more specifically, on age progression (e.g. how people are likely to look as they get older). Ideally, you'd want a broad and rigorous data set with lots of people's pictures. It would help if you knew they were taken a fixed number of years apart -- say, 10 years. Sure, you could mine Facebook for profile pictures and look at posting dates or EXIF data. But that whole set of profile pictures could end up generating a lot of useless noise. People don't reliably upload pictures in chronological order, and it's not uncommon for users to post pictures of something other than themselves as a profile picture. A quick glance through my Facebook friends' profile pictures shows a friend's dog who just died, several cartoons, word images, abstract patterns, and more. In other words, it would help if you had a clean, simple, helpfully-labeled set of then-and-now photos.
What's more, for the profile pictures on Facebook, the photo posting date wouldn't necessarily match the date that the picture was taken. [...] Through the Facebook meme, most people have been helpfully adding that context back in (e.g. "me in 2008, and me in 2018"), as well as further info, in many cases, about where and how the pic was taken (e.g. "2008 at University of Whatever, taken by Joe; 2018 visiting New City for this year's such-and-such event"). In other words, thanks to this meme, there's now a very large data set of carefully curated photos of people from roughly 10 years ago and now. In closing, Kate says it's not necessarily bad that someone could use your Facebook photos to train a facial recognition algorithm -- it's inevitable. "Still, the broader takeaway here is that we need to approach our interactions with technology mindful of the data we generate and how it can be used at scale."
What's more, for the profile pictures on Facebook, the photo posting date wouldn't necessarily match the date that the picture was taken. [...] Through the Facebook meme, most people have been helpfully adding that context back in (e.g. "me in 2008, and me in 2018"), as well as further info, in many cases, about where and how the pic was taken (e.g. "2008 at University of Whatever, taken by Joe; 2018 visiting New City for this year's such-and-such event"). In other words, thanks to this meme, there's now a very large data set of carefully curated photos of people from roughly 10 years ago and now. In closing, Kate says it's not necessarily bad that someone could use your Facebook photos to train a facial recognition algorithm -- it's inevitable. "Still, the broader takeaway here is that we need to approach our interactions with technology mindful of the data we generate and how it can be used at scale."
They already have that data (Score:5, Insightful)
Facebook is over 14 years old. They already have a series of photos showing people ageing more than 10 years.
They are evil, untrustworthy and despicable, but this current program is in NO way an additional threat. They do not need to use this data for age regression algorithms, they already have the necessary data.
Stop complaining about people feeding the baby Dragon - it is already fully grown and eating whole herds.
Re: (Score:2)
besides if some dude just asked you for that for scientific research, I guess you would just give it, most people would.
then the research synopsis would be published on some for pay science paper, while the actual research had been sponsored by someone like facebook anyways.
Like, come on, you can't just frame _everything_ as a _threat_ like the blurb. of course then there's the faction who thinks any photos of them are a threat to their personal being.
Re: (Score:2)
Usually ML papers are on Arxiv. Almost all of them, including the FB ones.
Re: (Score:2)
Facebook is over 14 years old. They already have a series of photos showing people ageing more than 10 years. They are evil, untrustworthy and despicable, but this current program is in NO way an additional threat. They do not need to use this data for age regression algorithms, they already have the necessary data.
That's what I'm thinking too, I mean if I was doing this research the first thing I'd do is make a filter to find sequences of images that my general facial recognition software with >99% probability says is the same person. I'd probably not care unless I had like 5+ image sequences, there's plenty people who constantly take pictures of themselves and the "fire and forget" pictures will outnumber the "look at me 5 years ago cool" pictures by a massive factor. Then I'd iterate on that and see how well I'd
Re:They already have that data (Score:4, Interesting)
Facebook is over 14 years old. They already have a series of photos showing people ageing more than 10 years.
Yes, but it's really handy if people will go online and confirm the results of their AI.
Re: (Score:2)
Relevant/highlighted parts reveal that the author has never worked on large data projects. What is called out is regularly dealt with as a matter of course in said projects.
Re: (Score:2)
Seems like Facebook could just do that anyway (Score:2)
At this point Facebook has a ton of images of people over quite a long timeframe they can just mine on a whim for any changes over time, without people specifically posting comparison shots.
As for other groups training age changing AI, well I'm not sure why I should really be concerned about that anyway???
I guess the larger message is "any image you post publicly can be used by anyone for anything" but isn't that already obvious.
Publicly Share Lies Pirvately Stick To The Truth (Score:3)
Of course by publicly sharing lies, I mean grossly excessive exaggeration, with humour and relational database poisoning. So for me preferred gender Bungil, preferred pronoun cheeky bastard, preferred web mail server Yandex because and only because it's RUSSIAN (and so I end up in UK and US Russian statistics are mwa hah hah Russian sourced). My likes and dislikes, well being a moody bastard, entirely dependent upon my moods, it could be any thing at any time, like or dislike, depending upon the story I wish to tell at that time.
You 'MUST', absolutely 'MUST', treat the public internet as a fantasy game you play for fun and never, 'NEVER' ever to be taken seriously. The public internet, the place of grossly excessive exaggeration, slander and insults, purposeful offence, silly lies, empty truths, endless distortions of reality, not just about yourself but everything around you, mocking beliefs, creating beliefs, lies about beliefs, if it amuses be mysandrist, misogynist, prejudiced and racist all at the same time.
Work hard at making fun of internet social media, have as much fantasy fun with it as you can and poison those relational databases in every way you can. Invent genders, pronouns, religions, all new 'isms', be an SJF a social justice freak and freak the fuck out of all the brass rod up khyber pass (cockney rhyming slang) types, have them all huffing and puffing, chest pumping and frothing at the mouth and laugh at them, mock them and laugh at them some more.
Take the internet too seriously and you will kill us all, treat it like a joke and a fun fantasy place to play and we will live much more happily, don't be a dick, treat the internet like it is a big ole massive bag of dicks, don't become one of them, mock the fuck out of them (PS equal opportunity, there are just as many dick brained wamen as there are dick brained alpha males (apparently dog boys by inclination), the are less of the 'other gender' dick brains but only because there are lot less 'other genders' but probably a much higher proportion in that particular group, lots of real bitches, which I am sure they would agree with, well the normies amongst them).
Lie all over the public real name based internet, it is what it is for and the only thing it is good for, fuck the public ego strokers, the internet poseurs of no real value, greedy and shallow in equal measure.
Re: (Score:2)
I completely agree and have been saying this for years. The internet was, and should be treated like the wild west, anything goes. If you have any presence online at all, expect to be flamed/trolled/hacked, and don't feel bad about doing the same to others. Meatspace is the place to be polite and civilised (because of physical proximity you are more easily punished for not being civilised). Don't bring your immature, bedwetting need for niceness to our internet, its not wanted here.
Then the internet beca
Good for tracking illegal migrants (Score:2)
Years of swapping documents and creating new documents could have obscured the tracking of illegal migrants.
Now governments, police and mil have more digital tools to work with.
CCTV and needed photo ID years later will allow for a more easy way to look back over past attempts by illegal migrants at creating fake ID sets.
Re: (Score:2)
You are assuming that politicians actually want to solve illegal immigration, rather than keeping it going as a wedge issue.
While the politicians have ranted and dithered, illegal immigration has mostly faded away to near zero net migration. It is mostly a non-issue. We should move on to something that matters.
Re: (Score:2)
Who just create a new social media account under the name of their fake documents.
Did they have social media back in their own nation?
Years later social media can help governments link the old account that was not updated with the new account.
One from the no longer used account in anther nation.
The new social media used under a different name in another nation.
The illegal migrant with a new name, new social media and no legal reas
Re: (Score:2)
Why not package a decade of user tracking as a real data set that a gov will want to buy.
Re: Good for tracking illegal migrants (Score:3)
Illegally immigration is a red herring. It's fully lawful "guest worker" (unwelcome guest) programs that are driving down wages for American workers while enriching a handful of scumbags.
The "give up more of your privacy" 'challenge' (Score:3)
Memes, crap data and filters (Score:4, Interesting)
Sounds good, except....
Half of these 10 years later pics are now memes. Good luck filtering them out.
I've only seen a few that were actually 10 years. Most are more like 7-8 years. You don't know the actual time span.
People are selectively choosing what pictures to compare. There's a lot of bias in what is being chosen.
Filters, filters, filters. 10 years ago they weren't as common, or they were simple instragram-like color filters. Now they are actively distorting the geometry of people's faces, smoothing blemishes, making their eyes subtly larger, etc.
In other words is a bunch of crap data that is totally uncontrolled. Not exactly what you want to be training algorithms on.
Just white (Score:2)
That's really creepy (Score:2)
No, seriously, it really is.