Slashdot Mirror
How Facial Analysis Software Could Help Struggling Students
moon_unit2 writes "Tech Review has a story on research showing that facial recognition software can accurately spot signs that programming students are struggling. NC State researchers tracked students learning java and used an open source facial-expression recognition engine to identify emotions such as frustration or confusion. The technique could be especially useful for Massive Open Online Courses — where many thousands of students are working remotely — but it could also help teachers identify students who need help in an ordinary classroom, experts say. That is, as long as those students don't object to being watched constantly by a camera."
beers given as treatment? or is crack more in vogue?
world was created 5 seconds before this post as it is.
I know xkcd is usually oblig, but for the constant camera in your face, lets go with some oblig. Penny Arcade
Just like anything, it depends on how this technology is used. A big part of being successful is learning to deal with the times when you're flat out lost and have no clue what you're doing, but are still expected to produce results. If someone else deals with these situations for you, then there are things you will never learn.
Seriously, if people would do a ROI analysis and figure out how long it will take them to pay off the loans they are taking out, fewer would spend $100,000.00 get PhDs in Medieval English and then whine about not being able to find a job that will pay for their loans.
...to get everybody in the class to show up wearing one of these.
is it me or are these questions more blatantly marketing test-questions?
Seriously? What ever happend to the idea of a student being responsible for asking for help when they need it?
A teacher could help struggling students.
I remember when technology was fun. It's getting really sour and ominous, and I am starting to fear for future generations.
Unless it identified almost all of them it wasn't working ;-0
I would be more interested in the ones who don't show that emotion, since they are the ones so lost and confused that they have abandoned all hope and given up. If you are trying to learn Java, and you aren't frustrated and confused, you're doing it wrong.
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
If an instructor can't detect who is struggling, they are not a very good teacher and should find another job where ignoring people they are there to help is ok.
If a student can't open their mouth and say "I am struggling" and have to depend on an external agent to recognize this and help them, they deserve to continue struggling.
The answer to the problems posed by MOOCs is to teach human beings (instructors and students) to be self-responsible and willing to communicate, not to dream up some half-assed technological solution that can't perform the task better than a human.
I did a lot of that in my time. Although I can see why one would automate some parts (like getting a sample to analyze the facial from), but I really don't see any fun in letting the software do everything.
It's only metadata, isn't it?
Sorry, I forgot there are ads on the Web; I use Lynx.
...the teacher could just be encouraging of students asking questions and then helping them out. And when they get home, parents could take an interest in their lives.
"Facial recognition software to detect struggling students.." gg
You have Teacher and Professor ego to deal with. Will the students care, probably less so than the teachers, whose goal is to make them look like a better person than the rest of society. If you were to show that kids are struggling in their class and it isn't due to their own laziness, will make them look like a less of the perfect person. Off to complain to the union about this. Screw if kids are missing out in a good education it is always about the teachers, If they are teachers they must be trying work for the students best interests.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
Nice try NSA...
It's dead easy to spot who is struggling. A good first hint is, "fails exam, still doesn't show up to class." Another is, "shows up to class, spends it all watching YouTube videos." And, my favourite, "Asks bloody stupid questions that make it clear this student passed the pre-reqs by pity alone."
Identifying the students is easy. Successful interventions is much harder, and to some extent, futile. Sure, you can recommend additional reading, pair up high-performers with low-performers for peer teaching, or re-teach the same concept in multiple ways (boring & disengaging the high-performing students), but some fraction of students will do poorly anyway. And when a otherwise high-performing students struggles, they either directly seeking out assistance from the instructor (one-on-one teaching, recommendations additional resources, etc), or independently self-teach around the problem concept. You don't need to identify them, because they take responsibility for their own learning and either ask for help or fix the problem.
With a small group (100 for sure, certainly MOOCs), students will perform on a beautiful gaussian curve, with some stellar, "Why are you even taking this class?!" students and some, "So, you know you're going to fail and you're not doing anything to change that? Great." students. The eventual grade distribution is as predictable as the rash of grandmother-funerals, sick notes, and other absences that pop up during exam-season.
So, as a student, I would find this tech presumptuous and invasive, and as an instructor I'd find it utterly useless.
That is, as long as those students don't object to being watched constantly by a camera.
I don't meant to sound like a card-carrying member of the Fringe Lunatic Association, but after the multiple recent revelations that the LEO's ride around photographing cars and license plates, USPS photographs all mail, the NSA collects metadata on all phone calls, the FBI and NSA together mine data from social networks—in short, that the US government in fact does all those things that the fringe lunatics warned about for years—it's hard to trust a university, whether state-run or private, with a camera to watch me at my computer in much the same way that it's impossible to trust Microsoft to watch me with an always-on X-Box One camera/mic setup.* I feel that recent events have given students very good reason to question whether the benefits of automatic frustration-recognition software are worth the risk that some sort of data might make its way from the camera to an FBI/NSA/Fusion database, despite the sturdiest ringfences and firewalls of promises, hope, and trust. Really, if the MOOC designers are really concerned about frustration, why not just include an "I'm frustrated! Give me a hint!" button on the user interface? Why monitor faces through a camera, and why propose the idea at the same time that MS's creepy XBox camera idea went down in flames?
One of my friends did his masters thesis project in code quality metrics. As part of it he wrote some code that will find the average LOC per function, code/comment ratio etc etc and spit out a letter grade. His thesis guru was a fiend. He fed the source code of this analysis code into itself. Poor guy graded himself a C-minus or something.
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
I always thought that if you were a student and confused (aside from being the normal state of affairs for a student) it would be the student's responsibility to speak up and ask for help. Perhaps it would be more helpful for Profs and TAs if they got a report when more than x% of the class is confused. And I'm sure administrators would like to see those numbers too come tenure time!
Assignment 12-A is designed to require an average student approximately 20 minutes to solve.
Student #001A solves the problem in 3.5 minutes - Too easy for his skill level
Student #312Q solves the problem in 42.3 minutes - he is struggling and needs further assistance
Problem solved, and you didn't have to spend a dime placing spy-cams at every workstation. You're welcome.
An enigma, wrapped in a riddle, shrouded in bacon and cheese
They would be better off teaching the students "What the Face Reveals" and teaching them how to be better human beings than teaching a machine to find better ways to remove humanity from human interaction
Java is a horribly convoluted mess and completely unsuitable for learning programming of any kind. Hence every student struggles and that makes detecting it quite trivial.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Rather than dealing with the privacy issues and with the inevitable plethora of false positives, wouldn't it make more sense to have a button called "Raise Hand" or "Express Confusion" or "I'm lost"? In fact, very much like the buttons that already exist on distance learning interfaces?
Is this supposed to handle the case where users are confused but can't make themselves ask for help? Or is this setting up a framework to later require that all distance learning students have a camera trained on them? I was going to preface that with "tinfoil hat time" but constant surveillance seems to be less and less the stuff of conspiracy theories and more and more the stuff of daily life.
Oliver's law of assumed responsibility: If you're seen fixing it, you will be blamed for breaking it.
1 spot the horny female students
2 ???
3 profit
http://i.imgur.com/uhSNk4V.png
.
The Slashdot post says "open source facial-expression recognition engine", but the article links to a commercial web site that makes no indication that their product is open-source.
The problem with this type of innovation is that it's a shot in the dark.
We haven't the first clue for the most effective way to teach people. We study things in HS because the subjects are "classics", not because they are useful (Geometry versus Probability, for instance). The "walk around lecturing in front of passive students" model doesn't fit with the need to be rambunctious. The fixed, level-based scale of achievement: "all children should be at this level of achievement at this age, else they are disabled" doesn't take into account variations in maturity or birth date. (Be born a day earlier, get put into a class where you're competing with class mates a year older.)
For reference, check out redirect. The author carefully details a large number of education techniques and social services which have no scientific basis whatsoever. Predictably, when actually studied, many of these ideas do more damage than good; for instance, regarding teen pregnancy, government teaching initiatives tend to increase the teen pregnancy rate.
There's simply no evidence that a) this system works to the degree of accuracy needed, b) doesn't have a high false-positive rate due to unforseen factors such as drapes waving in the background, c) can be used to any good effect (double-blind studies anyone?) as a teaching aid.
Our track record for using technology to help education is not good.
It makes for a good story, though. "We don't know the best way to teach, but here's something that should work!"
Here's another thought problem for you. Recall the 2009 Star Trek movie which shows a young Spock standing in a pit while a computer presents audio and video lessons. (I don't think the pit model works, but a student in front of a screen seems natural enough.)
Assume that you have control over this content, and can do double-blind studies of minor changes. Each video is a computer program, so any small piece can be redone without retaping the entire lesson. The program allows student interaction.
What features would your ideal teaching machine have, what sorts of things would you teach, what sorts of experiments could you do to home in on the optimum teaching method?
Using phone apps that do much more make more sense than trying to scan a room full of faces trying to detect the difference between constipation and consternation.. Students can run an iPhone/Android app that let students answer questions from their instructors... the quiz results can be used to get immediate feedback on the learning process..
...to a non-problem. If a student is confused, he has the responsibility of seeking help.
To quote Sugata Mitra, any teacher that can be replaced by a computer should be.
In a real class, spotting the struggling student is obvious. Teachers already do what the proposed system acheive
It is interesting for a MOOC. But if MOOC teachers have to handle struggling students, I fear it will destroy MOOC viability
CERT looks really cool, but it doesn't appear to be open source; at least there doesn't seem to be a download link anywhere. It looks as if there's a company called "Emotient" that's commercializing the tech. It might have been open source at some point but it looks proprietary now.
Thanks.Read more............. http://computersbds.blogspot.com/