Human Behavior Understanding

Post Reply New Topic

I've been reading a new book: Human Behavior Understanding. (ISBN-13: 978-3642147142 for anyone who can't view the Amazon preview.)
There's nothing in it absolutely new, since the areas it covers (feature detection, social signal analysis, etc.) are established fields individually. But this has never been run as a holistic independent project before, and because of that, it should begin receiving funds to allocate to separate research areas. I intend to request assistance for creating a functional treatment device for people with AS.

From the abstract:

I've attempting to put together an in plain English write-up on this for further online discussion and I'd appreciate any help. If anyone has any questions about this, from a non-technical perspective, please let me know.

Thanks!

are you intending to go whole-hog on the AI aspect of this project? that is a major job.

Yes, I intend to go whole-hog on the AI. Yes, it is a major job. Yes, it will take some time for it to be completed. But it'll be so cool when it happens.

Imagine the ability to mathematically predict exactly how people will react to you, the optimal conversation topics to have, the most emotionally satisfying banter possible.
In other words: artificial social cognition is worth the effort.

Here's an analogy that I'm working on:
I'm a math genius. I'm not less of a math genius because non-math-geniuses use math calculators.
I'm an emotional idiot. NTs won't be less of emotional geniuses because I will use an emotional calculator.

Are you suggesting something along these lines:

Nanny Robot Senses Emotions in Autistic Children.

Yes, that's a decent reference. However, this robot's abilities are rather primitive compared to what the latest research is showing.

Go to Human Behavior Understanding on Amazon, click on the book icon showing "Click to look inside", then use the scroll-bar to navigate down to page 161. "Table 7" shows the confounding matrix of emotional states. These are much more complicated than the robot in the article you linked.

I intend to use the above research to implement a device that is at least as accurate in predicting emotions as that table.

The link I provided above was merely an example of using technology to interface with autistic children. My mistake was in missing the direction you were focusing..(I skimmed, sorry) So you are interested in an AI or at least an ES for reading NT emotional states? In a sense like an emotional prosthetic for those with dificulty reading others feelings?

"tl;dr"


Yup...


B I N G O

And from there, it'll be possible, though not easy, to start adding in conversational intelligence for optimal response. Given enough data, you can compute anything.

My apologies, but my time is short at the moment.. I was able to locate this paper and others at this site which my be of help in your research:

Expert System For Automatic Analysis of Facial Expressions

Thanks for the ref! Unfortunately anything written earlier than 2008 is a bit too old for the latest research since this is such a super-dynamic field. And expert systems are all well and good, but we really need something portable by a user, rather than on a super-computer hive, like most expert systems. (Technically, this paper is on implementing the older Hidden Markov Models/HMMs while the latest research is on the more efficient Markov Random Fields/MRFs.)

In fact, the newest algorithms are so efficient, they realistically can be run on microcontrollers. i.e. Can run on something the size of a cell phone!

(Actually, am I being too harsh here? I don't mean to be negative--I honestly like that someone has finally ref'd a decent paper. So, please... send me anything you've got!)

Keep in mind that in 2000, the 1 gig hard drive on most computers (state of the art at the time) was the size of a ham sandwich and 80 mhz processor was pretty good, for comparison the cell phone I am posting this with has a 32 gig SD card the size of my pinkie nail..

Point is while the info may be a bit dated, the hardware has come a long way in 10 years.. what took a large desktop computer can easily be done on a cell phone sized device now...

Btw, I took no offense, and think you have an idea which technology might have finally caught up enough to make practical.

Just look at the evolution of the hearing aid....