A doubt in C2_W2 Lab

Course Q&A Machine Learning Specialization Advanced Learning Algorithms
1

Dear Administrator,

Could you please help me on this issue?

image
image982×583 123 KB

For example,
We are creating a learning model by using ReLU for predicting whether a tumor is

c0 malignant level 1
c1 malignant level 2
c2 malignant level 3
c3 malignant level 4

x0 is the size of tumor
x1 is the size of patient’s age

According to the description in the figure,
“The output of the unit is represented by the color of the background. This is indicated by the color bar on the right of each graph.”

“since these units are using a ReLu, the outputs do not necessarily fall between 0 and 1 and in this case are greater than 20 at their peaks.”

May i know what does the output of unit (ranged from 0 to 20) mean and how does it apply to my example?

Thank you

2

The output here should be the malignant level i.e. a range of lets say 0-4 corresponds to malignant level 0 and so on in the same logic (note I have just taken this as an example it might not correspond to the factual number in this particular case).

3

We use relu Activation function for HIDDEN LAYERS ONLY .
For the output layer , we use SOFTMAX activation function which
returns a ‘N’ number vector where ‘N’ is number of classifications that is, here 4.