Slashdot Mirror
Researcher Trying To Teach Computer What Women He's Attracted To
jfruh writes: Harm de Vries, a post-doctoral researcher at the Université de Montréal, is trying to build an algorithm that will sort through pictures on Tinder and OKCupid and pick out women he'll find attractive. "Tinder kept giving me pictures of girls I wasn't attracted to," he said in a phone interview. "So I wondered if I could use deep learning." His program, built using deep learning techniques, has about a 68 percent success rate, which isn't that bad. (A human friend to whom de Vries described his preferences managed 76 percent.)
I call bullshit. Women don't date men, especially techies.
Let me guess: female and alive.
"-1 Troll" is the apparently the same as "-1 I disagree with you."
...wait, opps, wrong de vries....
I usually just type in "blonde Thai ladyboy" into the search box and the results are almost 100% what I'm looking for.
Is that so hard?
SJW's don't eliminate discrimination. They just expropriate it for themselves.
News flash: pictures can lie. Here's an idea - try talking to a girl in order to see if you might like her.
Wasn't that line of thinking the basis for a short story from Isaac Asimov?
A better algorithm would be one that can spot which images of men or women are either fake or taken in such a way as to mislead as to the appearance of a person. It doesn't really matter if the computer can pick with 100% accuracy the people you're attracted to if none of them actually look like their pictures.
I'd suggest building a dating site where there are no pictures and people only describe their interests or personality, but people would just lie about that too.
The algorithm to find girls attracted to him can be represented by a simple: "return false;"
And all 68% of the matches think he's a 'nice guy'.
If I'm reading this correctly, our post-doc protagonist has created a deep learning algorithm to automate the process of being shallow? I have no words. . .
The guy is creating a profile selection algorithm to sift through mostly fake algorithmically generated profiles because the fake profile site's own selection algorithm is inadequate.....
Sexbots are going to be really, really popular.
Maw! Fire up the karma burner!
Tinder shows him pictures of women that have some chance of finding him attractive. It's a supply and demand thing - there are more men than women on Tinder, so he may have to tone down a bit his expectations.
His own profile manages to garner zero responses.
Cheers to you, Mr "4everalone6269"
But I wonder how long it'll be before someone comes along and labels this as misogyny?
More interesting would be an application that only gives women (or men) who find YOU appealing.
If you think you're attractive, create an image-reading algorithm that picks girls that looks like yourself (facial features, shape, etc.)
Done.
Look at successful, power couples in the entertainment business sometime (especially athletes) and see if you can spot the narcissistic pattern...
If you don't think you're attractive, I guess you can put a NOT in front of the algorithm...
The triggered and whining masses are sure to arrive and tell us how misogynistic this is.
Or a snatch.
Or a cunt.
Or a pussy.
Or a gash.
Etc.
If he can't bother to look at the pictures himself, then he's looking for a woman that doesn't need attention from a man. That should be easier to search using text rather than pictures.
Or maybe he's looking for a woman who isn't afraid to find a man named Harm and meet him in person.
Researcher leaves academia and gets a job at Ashely Madison.
I can't believe he's having something do that for him when there's something willing to tirelessly look out for his happiness right there. Maybe one day he'll realize he had love right in front of him, and he and Eliza/Alice v.2 or whatever he's calling his algorithm can live a happy, long life together.
Let's see if he can use deep learning to filter only the women that would have anything to do with someone who would do this.
Gimme the following
5 feet to 5 foot 3
A to B cup
Short, powerful thighs, wide hips
Raven-haired
Didn't we see this in Weird Science?
Table-ized A.I.
Internet shall give the answer. Just run a facial recognition algorithm over your Internet discoveries and you'll see an accumulation of favorite characteristics. Mach those with the pictures from the dating services.
A foreign student burning through my tax dollars on his dumb-ass research. Go back to your home country and waste their money.
The dude needs help. Serious help.
Researcher Trying To Teach Computer What Women He's Attracted To
Well, at least he's trying something new. The usual, and completely wrongheaded, approach a computer nerd uses when trying to get laid is to try to teach women he's attracted to about computers.
Isaac Asimov wrote a short story about this years ago ("True Love"), abeit in a much more entertaining fashion.
"I will say to her, 'I am Joe, and you are my true love.'"
I keep hearing love is blind,guess i herd wrong. lol
Jack of all trades,master of none
if your first criteria is what the lady looks like, then maybe you have some issues.
I always wanted to know if she was rich. What's wrong with this guy?
there are 3 kinds of people:
* those who can count
* those who can't
... to feed their pr0n collection into some experimental software?
CUR ALLOC 20195.....5804M
Imagine, using a computer for dating?
Doubtlessly the computer matches would be so perfect as to eliminate the thrill of romantic conquest. Ha-ho-ha-hey-hoo!
Alas, only the five richest kings in Europe will ever know for sure...
Neither the summary nor the linked article provide the necessary statistics to tell us how well this algorithm actually works. We're told it has a 68% success rate, which presumably means that 68% of the time it gives the same answer as de Vries (the human subject/programmer).
The problem is, we're not told anything about the sensitivity or specificity of the technique. What is the rate of false positives? False negatives?
Let's say that de Vries typically finds 1 out of 3 (33%) of the profile pictures "attractive". His computer could score 67% accuracy just by rejecting every single picture. (Such an algorithm would have zero sensitivity, but perfect specificity, and a terrible false negative rate. The "reject-everything" algorithm also scores better the more picky de Vries gets.)
This sort of story is only interesting if it includes specific information about where and how his algorithm fails (and succeeds).
~Idarubicin
Toiling into the night in his mother's basement...
Although you can't get there on a dating site (yet) there's evidence that women are attracted to some men by smell. Maybe that's also why so many men like to spend their weekends in a boat fishing.
https://en.wikipedia.org/wiki/Hot_or_Not
Maybe OKcupid thinks he should judge women on more than his eagerness to fuck them. In general, the male isn't allowed to be fussy. Besides, attraction is a two-way street: If you're not thinking of someone else's feelings you're not boyfriend material.
Yeah, I used to prefer women with big tits or a beautiful face: That didn't contribute anything to a relationship. Now I look for women who admit they like fucking. Alas, women tend to not say this.
The Computer should select the pretty girls, not the beautiful girls.
It's hard enough finding people who aren't obese! And women are very good at using "MySpace Angles" to try to mask their fatness. (Taken from above angled down.)
If he can tell obesity from any angle, that would be good. (Or a verified dating service where height/weight get verified, and biometric login so a a fatty can't get a normal friend to get weighed for them.)
Pussy, two legs, at least one arm. Done.
already been done