Naked Science Forum
On the Lighter Side => New Theories => Topic started by: smart on 28/10/2017 14:22:23
-
What is deep programming?
Is it possible to develop neural pathways in the brain from machine-compiled instructions?
What do you think?
-
Is it possible to develop neural pathways in the brain from machine-compiled instructions?
If we could replicate the entire nervous system then yes, in theory, it could work, if we look at a circuit board it is a prebuilt pathway that information travels along. Therefore if we developed a circuit that a neuron could travel along we are already halfway there, again, in theory, you'd need to also develop a machine capable of compiling the instructions into a neuron, mainly because our bodies rely on these to give it instructions. Theoretically, it is a possibility, though I'm sure I read somewhere about a guy who could control his arm using electrical impulses after he damaged nerves. Seems like we are halfway there.
-
What is deep programming?
Do you mean "Deep Learning"?
Many problems that are solved regularly by bees and dogs are beyond our most powerful computers when programmed by humans. We simply don't know how to represent and solve these problems using conventional computer algorithms on conventional computer hardware.
So AI researchers have tried to imitate biological structures like neurons in a computer, and rather than "program" them, they "train" them by giving them examples that match the desired criteria (or not).
Normally learning requires considerable high-precision, high-powered processing and analysis. The learned results can often be executed on different, low-precision, low-power hardware (which is itself incapable of further learning).
This still falls short of biological systems where continual learning and execution takes place on the same low-precision, low-power "wetware".
See: https://en.wikipedia.org/wiki/Deep_learning
-
What is deep programming?
Do you mean "Deep Learning"?
Many problems that are solved regularly by bees and dogs are beyond our most powerful computers when programmed by humans. We simply don't know how to represent and solve these problems using conventional computer algorithms on conventional computer hardware.
So AI researchers have tried to imitate biological structures like neurons in a computer, and rather than "program" them, they "train" them by giving them examples that match the desired criteria (or not).
Normally learning requires considerable high-precision, high-powered processing and analysis. The learned results can often be executed on different, low-precision, low-power hardware (which is itself incapable of further learning).
This still falls short of biological systems where continual learning and execution takes place on the same low-precision, low-power "wetware".
See: https://en.wikipedia.org/wiki/Deep_learning
Learning itself is not anything useful, remembering is ''learning''. Learning just means remembering, a person will be good at their job if they remember their ''programming'' that they were taught to remember.
Almost any person can remember and follow due process according to memory. A monkey is not smart because the monkey can place a shape in the correct shaped hole, the monkey uses trial and error until the monkey remembers which hole goes with what shape.