Loss function - P.15 vs P24 of the course slides

Course Q&A Deep Learning Specialization Sequence Models
1

Dear all,

I believe the correct loss function is that of P.15 and not P.24.

Is that correct?

Thank you.

2

Hey @AntonioMaher,
Can you please let us know which week’s slides are you referring to?

Cheers,
Elemento

3

Sorry, I am referring to W1.

Thanks for the quick reply.

4

Hey @AntonioMaher,
I don’t see how the loss formulation mentioned on P 24 is incorrect.

Here, I assume you know that the loss formulation mentioned on P.15 is for binary classification (whether a word is a person’s name or not), while the one mentioned on P.24 is for softmax classification. Also, the binary classification loss function can easily be converted into softmax classification loss function, by just changing how the labels are defined. If for binary classification, instead of defining the labels as 0 and 1, you define them in one-hot encoding form, i.e., [1, 0] and [0, 1], you will find that both the loss formulations are the same. Let me know if this helps.

Cheers,
Elemento

5

Hi @Elemento,

Thanks for the clarification! Crystal clear now.

Have a great week!