1) Do you think that the weights acquired by the inputs could be construed as memory elements? And that, in theory, if the weights (neuron input sensitivities) could be read out, we could divine actual memories? 2) Then if these weights were transferred to another network, it would perform in the same way as the original. 3) Would this mean that if this was also done in humans, would we be able to download our memories into an artificial neural network?
The question is -- can it monitor its performance, analyse it and then re-program itself to achieve better results? Can it then logically re-define the terms "better" and "results"?
1) Do you think that the weights acquired by the inputs could be construed as memory elements? And that, in theory, if the weights (neuron input sensitivities) could be read out, we could divine actual memories?
2) Then if these weights were transferred to another network, it would perform in the same way as the original.
3) Would this mean that if this was also done in humans, would we be able to download our memories into an artificial neural network?
The question is -- can it monitor its performance, analyse it and then re-program itself to achieve better results? Can it then logically re-define the terms "better" and "results"?
How much does it cost? Sales package? How to buy?
Tell you what.. mitl ayri into w ibm wyalli kherikon 3ars.
Spikes goes in, Spikes goes out, you can't explain that!