The Core

Why We Are Here => Hardware & Technology => Topic started by: rcjordan on June 04, 2017, 02:42:58 AM

Title: Algorithm Accurately Reconstructs Faces From A Monkey’s Brain Waves
Post by: rcjordan on June 04, 2017, 02:42:58 AM
http://blogs.discovermagazine.com/d-brief/2017/06/02/brain-waves-faces/

Interesting ...in an ominous  sort of way.
Title: Re: Algorithm Accurately Reconstructs Faces From A Monkey’s Brain Waves
Post by: ergophobe on June 04, 2017, 04:19:48 PM
And in 5 years, your smartphone will read your mind - currently 90% accurate at numbers...
https://twitter.com/ValaAfshar/status/871068025109303296?cn=bWVzc2FnZQ%3D%3D
Title: Re: Algorithm Accurately Reconstructs Faces From A Monkey’s Brain Waves
Post by: BoL on June 05, 2017, 10:13:18 AM
Great for criminal line-ups I would suppose, and evidence... unless the patterns are hackable with training (I'd suspect probably not as they'd be fairly primal)
Title: Re: Algorithm Accurately Reconstructs Faces From A Monkey’s Brain Waves
Post by: Brad on June 05, 2017, 11:33:49 AM
>hackable

Wave a banana around while monkey brain is memorizing or identifying.
Title: Re: Algorithm Accurately Reconstructs Faces From A Monkey’s Brain Waves
Post by: ergophobe on June 05, 2017, 10:57:54 PM
Quote from: BoL on June 05, 2017, 10:13:18 AM
unless the patterns are hackable with training

From everything I've read, these patterns only emerge as language after training. That is, if you want to translate brainwaves into language, the person needs to learn how to think. Or perhaps more accurately, how to talk using subvocalizations.

https://en.wikipedia.org/wiki/Subvocalization

The facial reconstruction thing from chimps is new territory... very interesting
Title: Re: Algorithm Accurately Reconstructs Faces From A Monkey’s Brain Waves
Post by: simplytheresa on June 06, 2017, 12:10:01 AM
Whoa! Super interesting! I'm also excited because the full Cell article is available online. I don't have time to look through it in depth right now, but a few notes:

- Although they used fMRI to initially identify the face cell clusters, it looks like the actual data was collected via "electrophysiological recordings" --> tungsten needles. Very specific and invasive.
- Also, these are not memories of faces, but faces that the monkeys were looking at while the recordings were being taken. Memories are going to be much more complicated.