Innovation by DX

Lesson 13  Artificial Neuron

Artificial Neuron is a model of creatures' neuron. The architecture of Artificial Neuron is as follows.

It accepts multiple inputs, and exports outputs after a defined process, which multiplies each input with prepared coefficient, and sum up all values.

Suppose the number of inputs is N, and see how the calculation works.

Multiply an input $x_i$ with a pre-defined coefficient $w_i$ The $w_i$ is called Weight. Multiply all inputs with corresponding coefficients, and sum up all values. And then, add a value $b$ . The output is $u$ . Where, the value $b$ is called Bias. Finally, apply a function $f$ to the output value. The final output is $z$ . Where, the function $f$ is called Activation Function.

The formulas of the calculation process above are as follows.