Plot decision lines for first layer

Course Q&A Machine Learning Specialization Advanced Learning Algorithms
1

Hi guys,

In the multi-class lab (week 2), there is stage where the decision boundary lines for the first layer output are added to the plot:

image
image987×466 101 KB

I tried to do the same by myself, and I don’t know if I did it right :frowning:
My code is:

line1 = W1[0, 0]/W1[0, 1] * X_train[:, 0] + b1[0]
line2 = W1[1, 0]/W1[1, 1] * X_train[:, 0] + b1[1]

plt.scatter(X_train[:, 0], X_train[:, 1])
plt.plot(line1, X_train[:, 0])
plt.plot(line2, X_train[:, 0])
plt.xlim(-7.7, 7.7)
plt.ylim(-4.5, 4)
plt.show()

and the result I get is:
image

What do you think? Is this correct or did I miss something?

2

Hello @Evgy,

It becomes clearer if we write down the first equation and derive from it step-by-step:

image
image618×669 54.5 KB

Cheers,
Raymond

3

Thank you Raymond,
Now it looks good :slight_smile:

image

line1 = -W1[0, 0]/W1[0, 1] * X_train[:, 0] - b1[0]/W1[0, 1]
line2 = -W1[1, 0]/W1[1, 1] * X_train[:, 0] - b1[1]/W1[1, 1]

plt.scatter(X_train[:, 0], X_train[:, 1])
plt.plot(X_train[:, 0], line1)
plt.plot(X_train[:, 0], line2)
plt.xlim(-7.7, 7.7)
plt.ylim(-4.5, 4)
plt.show()

4

That’s good, @Evgy!

Raymond