Question 1 “Which of the following do you typically see in ConvNet? (Check all that apply.)” doesn’t seem to have a correct answer since none of the options are entirely correct, e.g.:
- FC layer is used to flatten the volume, not after flattening the volume.
- Pool layers are used after CONV not vice-versa
- ConvNet is made up of CONV layers, Pool layers etc
- FC layers are not followed by a CONV layer.
I’d have to read the exact wording of the questions, but I’d say that option 2 and 3 sound true to me just based on what you listed there.
Hey @paulinpaloalto thanks for getting back to me. I would love to share the exact options provided so the mods could maybe raise an issue (if it’s valid) but there is no way to do so without violating the forum rules.
I have made this post “unlisted”, which means other learners cannot see this anymore. Only people with the link to this or the ones who have the permissions can see this now.
Feel free to freely talk about this and share whatever information you want.
We are happy to help
Hi @Mubsi , these are the answer options given. None of the options sounds entirely correct.
The first option should be correct. My belief is rhat the others are not.
I’m confused here, I thought a non-linear activation layer such as softmax is used instead for generating output classes?
Firstly, the “check all that apply” has been left behind by mistake (I have removed it), this question version only has one right answer.
As for the answer, as Paul mentioned, the first option is the correct answer. You are right that in order to get the prediction when there are more than 2 classes, softmax is used. But remember, the question is specifically about ConvNet. Maybe going through the lectures again regarding ConvNet will clear your doubts.
Option 2 and 4 are incorrect as they are not common practice.