Why a 1-3-1 BP neural network with 10 parameters (weights and thresholds, lay particular stress on)? Pray god give directions

CodePudding user response:

1 is input, set to x, the middle layer is set to y, the output is set to z,
So 1-3 is y1, y2, y3=(w1, w2, w3) * x + (b1, b2, b3), namely w1, w2, w3, b1, b2, b3 six parameters,
3-1 is z=y1 * w4 w5 + y3 + y2 * * w6 + b4, namely four parameters,
The above total 10 parameters,