ProgrammingGodJordan
Banned
Well, there are two classes of brain based learning:
(1) Deep learning(a) This began by using hints from neuroscience.
(b) However, deep learning benefits from paradigms outside of neuroscience.
(c) This is because deep learning experts don't know enough about the brain, to get much performance.
(2) Computational neuroscience(a) Stricter versions take a path to as closely as possible, attempt to imitate brain neurobiology, unlike Deep Learning. (example is Markram's project)
(b) Mixture of computational neuroscience and deep learning both attempts to imitate the brain, while adapting deep learning tasks completion. (Example IBM)
FOOTNOTE(0):
Both of the above are crude approximations.
- Deep learning computes on layers of low scale simulated neurons (see 1(c)), and basic synaptic weights that record representations of inputs, for appropriate prediction/classification tasks.
- Computational neuroscience scales from models that literally strive to achieve brain neurology, to models that leverage deep learning.
Both paradigms yield models that either exceed or equal human performance on some cognitive tasks, and more and more cognitive tasks enter the machine doable domain as time passes.
FOOTNOTE(1):
As I mentioned in original post, and as is observable otherwise, a lower bound for when machines achieves human level cycles per moment is doable.
And I repeat that I don't equate human level intelligence with this probable arrivial in matching cycles per moment par human/machine models.
Artificial consciousness (or whatever human intelligence is) is probably a bit further off, roughly 2025 - 2050.
FOOTNOTE(2):
Anyway, we shall note that these cognitive machines don't need to be conscious to displace jobs.
They are already better disease diagnosis machines, etc, and even machines like these have already began to displace jobs.
Last edited:
!
