Thanks Diogenese, you are so kind to take the time to respond, I'm really sorry but I really don't understand any of that, it's well beyond me. In regards to the paper, is there any way akida could help with their research? Again thanks for your time.
Hi Rach,
As I indicated, statistics is not my strong suit.
Short answer is "no", Akida cannot be used with this concept.
This process uses analog transistors, and an inherent characteristic of analog transistors - manufacturing variability. Analog transistors are much more sensitive to manufacturing variations and to changes in tempera
Akida uses digital circuits which have a binary output (1 or 0). This gives a binary circuit a high margin for error - it's ON or it's OFF.
ture.
Analog signals can have any value within the limits of the system (the supply voltage), but, because of manufacturing variability, the same input to one analog circuit may produce a slightly different output from a second apparently identical analog circuit. Although an analog circuit may be intended to produce a current of 1 milliamp, it may produce a current of, eg, 1.1 milliamps. In an analog neuron, these currents are collected in a capacitor and the voltage across the capacitor is the neuron's output. When there are hundreds of analog transistors in a circuit, the variations in current can affect the accuracy of the process.