LEARNING OF ANEURON (1/2)
Perceptron an early version of
neuron
(about 60 years ago)
Given input X1, X2 and output Y1, obtain
the value of W11 W12, and b1
Linear kernel function: y1 = W11x1 +
W12x2 + b1
noise
9.
LEARNING OF ANEURON (2/2)
X1 X2 Y1
0 0 0
0 1 0
1 0 0
1 1 1
W11*0 + W12*0 + b1 < 0
W11*0 + W12*1 + b1 < 0
W11*1 + W12*0 + b1 < 0
W11*1 + W12*1 + b1 > 0
b1 = -4
W11 = 2
W12 = 3
W11 < -b1
W12 < -b1
W11 + W12 > -b1
y1 = W11x1 + W12x2 + b1
b1 < 0
Target:
Learning “AND function”
Obtain a line to separate
blue and red points into two dimensions
10.
XOR PROBLEM
X1 X2Y1
0 0 0
0 1 1
1 0 1
1 1 0
Not linear separable in one layer
other non linear functions
Need 2nd line to separate 2 domain
2 layers
#6 AI -> Machine learning -> Supervised -> NN
Deep learning = Deep NN in Machine learning
有不deep的learning 嗎? Shallow learning
SVM -> 低維度不可分, 那就拉到高維度分看看