|
We have now seen how a single neuron works, but what can it be used for.
Let's take an example: Say we want to create a binary calculator by using "neural networks".
We will only need a single neuron, the reason why we are making a binary calculator is that it is less
difficult when you define the weights by yourself. The neuron has got 8 synapses and a linear activation-function.
This is a quite unusual neuron because the weights goes beyond -1 to 1, but it reflects what we are using it for very well. The weights are listed on the following table.
You can check out if it works yourself.
| wk1=1 |
wk2=2 |
wk3=4 |
wk4=8 |
| |
| wk5=16 |
wk6=32 |
wk7=64 |
wk8=128 |
|
