Neural networks

Question 1

Give the simplest explanation of what is a neuron

Answer 1

Something that keeps a number.

Question 2

For an single geyscale image what is a neuron?

Answer 2

A pixel with the corresponding value

Question 3

How it works in the very general perception

Answer 3

We have a layer of neurons which activate another pattern of neuron and so on. On the final layer, the neuron with the higher value, is the networks choice

Question 4

Why it is reasonable for a layer like this to behave correctly for the test case of handwritten digits?

Answer 4

Each layer can represent a line edge a pattern and at the end we have the end pattern.