Slashdot Mirror
Scientists Propose App That Detects Emotions Based On Walking Style
An anonymous reader writes: Chinese researchers claim to be able to deduce a person's emotional state using accelerometer data from mobile devices attached to the wrist and ankle. The study recorded baseline data and then comparitive data after showing either disturbing or amusing videos to test subjects. The paper envisages the ultimate development of smartphone and wearable apps capable of providing systematic long-term and short-term data on someone's state of being, based mostly on the movement of the ankle whilst walking. They posit the usefulness of the information in medical applications, but do not address possible unsuitable uses, such as for the purposes of employment assessment or insurance premiums.
Nice! In the future I can buy employee walking data from Apple or Google and fire all those unproductive depressed employees.
We need not unhappy personnel.
>> movement of the ankle whilst walking
Great - I can't wait until we're all issued our "optional but wear it if you want a job" ankle bracelets.
effenheimers. the F-word is so fucking commonplace that the way it's used in conversation can be a clue as to how the users feel.
if this is supposed to be a new economy, how come they still want my old fashioned money?
I wonder how the ankle device would cope with a silly walk. At the very least, the workplace would get more interesting as people don't walk from point A to point B in a normal manner.
My sci-fi novel, Ghost Thief, is now available from Amazon.com.
With lithium and Benzedrine! I will be happy all the time! (the correct amount of happy)
"detecting" emotion! what a laugh and a lie, they are doing nothing of the sort. They are just guessing.
When the day comes that such a thing is invented by sociologists there will surely be a scope-creep coda to the tune of "more research needed" within the vast sphere of human malfeasance.
Just what we need is a technological literature brimming with amateur hand-wringing and armchair ethics. I'd just love to read what Shockley might have written about his invention in the last paragraph of the last page if given a greenish-yellow editorial light to paint the future.
While we're at it, how about some moral footnotes from Fritz Haber?
A sad end, but a fine act of ethical commentary by the first woman to be awarded a doctorate in chemistry in Germany. To think what we might have learned if only she'd been wearing a mood bracelet.
Just got a raise, but exhausted from pulling a 30 hour shift? Depressed and slouched.
How accurate is this really going to be considering the multitude of reasons you could have for walking in one way or another? What about people with a bad knee who start walking strangely because the knee starts hurting?
-=This sig has nothing to do with my comment. Move along now=-
You can go with this
Or you can go with that
...considering they cannot walk in a straight line, act absurdly happy, yet alcohol is a depressant. Better yet would a woman on her period cause it to just fluctuate wildly until the device spontaneously combusts?
Your gait may not be able to tell you anything about your emotional health, but it certainly can tell you something about how long you'll probably live.
http://www.livescience.com/104...
You are welcome on my lawn.
Science is for Luddites. Only modern appers with app apps can app apps using apps!
Apps!
by the way I use my walk, I'm a woman's man, no time to talk?
Will it also detect silly enough walks that suffices the ministry of silly walks rigorous standards?
all the time...
Terrific! AI is slowly catching up to psychopaths!
Book, A., Costello, K., & Camilleri, J. A. (2013). Psychopathy and victim selection: The use of gait as a cue to vulnerability. Journal Of Interpersonal Violence, 28(11), 2368-2383. doi:10.1177/0886260512475315
We have this already, it's called a polygraph. A device that detects fear and stress, used by a so-called technician to calculate how normal and 'honest' the subject is. A person can control their limbs far better than they can control their breathing and sweating. Leaving aside personality modification therapy, the main purpose of this will be teaching the subject to hide their emotions.
The authors report results from 10-fold cross-validation experiments. Yet this evaluation approach is unsuitable for the segmented time-series data used in this work, as they are not statistically independent (adjacent frames overlap to a large extent). There was recently work presented on ubicomp (a conference on this sort of stuff), that illustrates how this leads to a bloated, overly optimistic performance estimation. Until they show the user-independent performance of the system, e.g. in a leave-one-user-out evaluation, you are safe to ignore these results.
It would be easier to look at someone's face.
Smile mean they are happy.
Frown means they are sad.
How would this be useful in any way? The person will know what they feel like, so this has to be a way for someone else to know what you are feeling without them telling them.
http://craphound.com/littlebro... comes as close as recommending how to hack them.
> Scientists Propose App That Detects Emotions Based On Walking Style
The United States of America and Germany are going to be entirely immune, thanks to Otto, Diesel, Daimler, Benz, Packard and Ford.
1. discriminate against people with one or two legs amputated. Because I can't imagine they will give those people and exception.
2.Look up books titles "Human Walking" (one by Inman, one by Rose. I think Rose took Inman's first edition and updated it). There is a lot going on in walking, and having only two feet--three would have been a lot more stable--makes it a lot of the time a game of "continuously interrupting falling down." Developing a program to be sensitive and selective, in other words, accurately interpreting the raw data is an issue.
If the results are used for the purpose of helping your doctor consider further diagnostic tests, or changing meds, or referring you to a therapists, maybe 80% accuracy is okay.
But if the issue involves criminal courts, employment screening, and the results include 1%, 10%, or 30% false positives, maybe there is a bigger problem.