Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Comments on M.Sc project #8

Closed
Anri-Lombard opened this issue Dec 27, 2021 · 3 comments
Closed

Comments on M.Sc project #8

Anri-Lombard opened this issue Dec 27, 2021 · 3 comments
Labels
comment A general comment on the repo

Comments

@Anri-Lombard
Copy link

Well done with your deep learning algorithm!

I thought of a simpler U-Net architecture for image segmentation similar to an x-ray lung cancer detector introduced by the "AI in medicine" Coursera course; although I like your ConvNet setup the softmax classifier (I hope; I can't open your paper because of its size).

Good luck with work or a Ph.D.!

@Adamouization
Copy link
Owner

Hi Anri, thanks for the comment, I appreciate it! My project colleague explored using U-Net for image segmentation, you might want to check it out.

You can try downloading the paper through my website at the very bottom if you're interested http://www.adam.jaamour.com/

@Adamouization Adamouization added the comment A general comment on the repo label Dec 28, 2021
@Anri-Lombard
Copy link
Author

I have just read most of the 14,345 words, and it is overall fantastic. In fact, I thought a master's dissertation will be overwhelming (to a freshman at least), but it is explained and described so well that I could follow along. I predict this is the result of great writing!

What I learned:
(1) This paper bore down to fundamentals, even techniques like regularization (and dropout), data augmentation, and transfer learning. If you describe gradient descent it could be a full overview of machine learning :-)
(2) This paper took ethical considerations seriously; I do not know if this is common across researchers good and bad, but I respect it highly.
(3) You used imageNet for transfer learning and built your own classifier on top of it with a softmax for 3 classifications (although this is common it is the first time I see it implemented in an M.Sc. dissertation).
(4) You chose TensorFlow over Pytorch because it is older and has a larger community to help you where you get stuck. This is surprisingly clever as my naive reaction would have been "Pytorch, newer and shiny. Let's use it".

Selfish outtakes: I now understand what a master's dissertation involves and how little I know.

Lastly, I am a big fan of your website.

@Adamouization Adamouization pinned this issue Jan 19, 2022
@Adamouization
Copy link
Owner

Thank you very much for all the comments, I really appreciate them! Ethical considerations are always very important, especially when dealing with data like this! I do hope it is standard across all researches 😄
PS: I also like your website, it has cool features!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
comment A general comment on the repo
Projects
None yet
Development

No branches or pull requests

2 participants