Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Thursday January 31 2019, @03:19PM   Printer-friendly
from the you-should-be-in-pictures dept.

IBM hopes 1 million faces will help fight bias in facial recognition

IBM thinks the data being used to train facial recognition systems isn't diverse enough.

The tech giant released a trove of data containing 1 million images of faces taken from a Flickr dataset with 100 million photos and videos.

The images are annotated with tags related to features including craniofacial measurements, facial symmetry, age and gender.

Researchers at the company hope that these specific details will help developers train their artificial intelligence-powered facial recognition systems to identify faces more fairly and accurately.

And then the police adopted the new facial recognition algorithms and everyone lived happily ever after.

IBM blog post. Also at TechCrunch and VentureBeat.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 3, Informative) by physicsmajor on Thursday January 31 2019, @04:49PM (7 children)

    by physicsmajor (1471) on Thursday January 31 2019, @04:49PM (#794579)

    Two quick points to completely deflate your arguments:

    First, "White" needs to not lump in those of Hispanic or Arab/Middle Eastern descent. Yes, I know the liberal rags like to conflate these, and I'm calling it out as the complete bullshit that it is. Run the numbers again and you won't be able to make those claims.

    Second: By far the most mass murders are blacks on blacks in inner cities, but these don't fit the narrative and get no media time despite being an insane proportion of gun violence. Despite multiple casualties they are also rarely classified or termed "mass murders," a poorly defined moniker.

    Starting Score:    1  point
    Moderation   +1  
       Flamebait=1, Troll=1, Insightful=1, Informative=2, Total=5
    Extra 'Informative' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   3  
  • (Score: 0, Interesting) by Anonymous Coward on Thursday January 31 2019, @05:27PM (1 child)

    by Anonymous Coward on Thursday January 31 2019, @05:27PM (#794601)

    I have never seen anyone combine hispanic and middle esstern as "white", and gang violence is classified differently from mass shootings, sort of like how military deas ths are not considered murder. You had a good point about how the simple statistics will result in a racist algorithm, but now you seem to be desparately trying to hammer "blacks are most violent!!!" while dismissing valid criticism. Not a good look.

    • (Score: 2, Touché) by Ethanol-fueled on Thursday January 31 2019, @10:08PM

      by Ethanol-fueled (2792) on Thursday January 31 2019, @10:08PM (#794730) Homepage

      Didn't the FBI statistics consider Hispanics/Latinos "White" for quite some time? Kinda like this? [nationalvanguard.org]

  • (Score: 4, Informative) by Anonymous Coward on Thursday January 31 2019, @05:57PM (4 children)

    by Anonymous Coward on Thursday January 31 2019, @05:57PM (#794618)

    I used to think exactly like you! I even made it the subject of my masters thesis thinking it was a genetic thing. After all, everyone knows that blacks are more violent to blacks than white folk are and blacks are just statistically more violent in general right?

    That was my personal experience and the experience of everyone I knew, whether black, brown, white or other.

    My thesis assumption was that this tendency towards violence, must be a genetic legacy from being tribal and uncivilized for so long. It makes perfect sense that gangs would be the modern replacement for tribes. But in the end, my research ended up pointing me somewhere else and I was forced to draw a different conclusion.

    Your argument doesn't hold when you control for poverty levels especially when you look globally and examine homogeneous cultures.

    The reason it appears to hold in the USA is because in the USA, the overwhelming majority of blacks live in poverty here and per household, blacks have the lowest average income, although hispanics tend to have the lowest personal incomes, they make up for it by craming a whole lot of family into a house. Ergo total household income, not personal income is what I mean when I say poverty.

    If you control for poverty levels and population density, it paints a different picture.

    Take Utah for example, there are many cities where there are literally no black people. In fact pretty much anywhere outside of Salt Lake City proper and blacks are astonishingly rare.

    Mexicans tend to be over-represented in the jails but also over-represented in the victim statistics. Then you get to towns like Hurricane & Eureka where the culture is completely homogeneous and completely white and so are the jails.

    Per capita crime rates are about the same everywhere when you look at per household income in the area vs population density.

    Thus, your white trailer trash kid is just as likely to commit a violent crime as a your black ghetto rat assuming the household income is about the same.
    Turns out household poverty rates are more likely to be the causative factor, not race, but race and poverty are strongly correlated.
    This makes sense when you consider the desperation and stress that becomes a daily way of life for people who are truly impoverished.
    It's not to say that all poor people commit crimes. Most people don't commit crimes and most poor people don't commit crimes. But the line to cross from upstanding citizen to crook is a lot easier to cross when you have nothing to lose, especially if you have mouths to feed.

    • (Score: 3, Insightful) by Azuma Hazuki on Thursday January 31 2019, @06:39PM

      by Azuma Hazuki (5086) on Thursday January 31 2019, @06:39PM (#794636) Journal

      The kind of person who posts the shit "physicsmajor"--oops, pardon me, just some AC who is in no way at all physicsmajor--posts, is not going to listen to reality. He, and this has *got* to be a he, Has A Narrative. And if the data don't fit The Narrative (TM), then the data are wrong and/or don't exist, and he will simply ignore them.

      --
      I am "that girl" your mother warned you about...
    • (Score: 0, Redundant) by Anonymous Coward on Thursday January 31 2019, @07:09PM (2 children)

      by Anonymous Coward on Thursday January 31 2019, @07:09PM (#794652)

      Someone's projecting a lot, and it isn't physicsmajor - all the grandparent stated was that these things are not predominantly white (which is what the great-grandparent outright stated). They're right about that.

      You are the one making it about race/projecting it into a white supremacy post, when all it says is to reexamine your data after an incorrect statement was made (that whites were mass murderers and terrorists).