Artificial neural Network (Artificial neural netwroks) Note-discrete single output perceptron algorithm

Recently in the study of Artificial neural network (Artificial neural netwroks), make notes, organize ideas

Discrete single output perceptron algorithm, the legendary MP

Two-valued Network: The value of the independent variable and its function, the value of the vector component only takes 0 and 1 functions, vectors

Weight vector: w= (W1,W2,W3.....WN)

Input vector: x= (X1,X2,X3.....XN)

Training Sample Set

{(x,y) | Y is the output of input vector x}

The training process is relatively simple

As follows:

1, initialization weight vector w

2, repeat the following procedure until the training is complete:

2.1 For each sample (X,y), repeat the following procedure:

　　　　　　　　　　　 2.1.1输入X
　　　　　　　　　　　 2.1.2计算 O=F(XW)
　　　　　　　　　　　 2.1.3如果输出不正确，则改变权重
　　　　　　　　　　　　　　　　　　　　 当O=1时，W=W-X
　　　　　　　　　　　　　　　　　　　　 当O=0时，W=W+X

It's interesting to adjust the weight vector.

It's hard to figure it out at first, but if you're just asking for an approximate solution, then it's easy. The value is just a constant approximation of the process of change.

A little bit of initialization weights are random, so the process of adjusting weights is more random.

I've been thinking about support vector machines (SVM), so it's a bit sleepy to adjust for such a weight.

SVM is the global optimal, and Ann is the local optimal, so the process of weight adjustment of Ann is simpler.