Week 1, Coffe Roasting Numpy, Lab 2. I dont understand how should we expect the amount of parameters

Course Q&A Machine Learning Specialization Advanced Learning Algorithms
1

How did we expect layer 2 to have 4 parameters? (3 W parameter, and 1 B). I would expect it to only have 2 W paramerer since the input is in the shape of (2,)

Screen Shot 2024-07-04 at 15.11.11
Screen Shot 2024-07-04 at 15.11.111006×656 118 KB

Screen Shot 2024-07-04 at 15.11.17
Screen Shot 2024-07-04 at 15.11.171073×575 102 KB

The classroom referenced

2
  1. input is 2 right.
  2. layer 1 parameters will be 2 x 3+3, (input size x size of layer i.e 3 + size of layer x 1 for parameter b ), this layer will output the size 3.
  3. layer 2 parameters will be 3 x 1+1. (hope you get this one)
1 Like
3

Thankyou for answering, however I am still confused why is the layer 2 parameter will be 3x1 + 1. I thought it would the same as layer 1, 2x1 + 1 (because layer 1 will outputs matrix of (2,1) ?

4

see layer 1 has 3 units hence it will output the matric of size (3,1), and layer 2 has 1 unit, therefore to match the dimension of input it will have 3 weight parameters and 1 bais term. Got it?

1 Like
5

Owww, I understand, I forgot that layer 1 will output matrix of size (3,1). Does it mean that, if for example layer 2 has 2 unit, it would be 3x2 + 2?

6

yes,yes!!

If your doubt is solved, feel free to my mark my answer as SOLUTION to close the thread. Let me know if there is anything I can help you with.

2 Likes
7

My sincere gratitude and thankfulness for helping me learn ML :grin:

2 Likes
8

@tarunsaxena1000, note that marking a reply as a Solution does not close the thread.

1 Like