Hi,
A couple of months ago, I had saved a few links from the week 3 discussion forum that had a great explanation of all the week 3 material. I had gone through a couple of links, and it had cleared many of my doubts.
I was planning to go through the remaining links now, but those links are not working anymore since everything is moved to discourse.
So, can someone help me to find this content on discourse?
Here are the links that I had saved (unfortunately I hadn’t saved the titles of those threads, so I can’t search them on discourse using title)
https://www.coursera.org/learn/convolutional-neural-networks/discussions/weeks/3/threads/tcCkmLwvT9SApJi8L5_UNQ
(I think this was a pinned thread)
https://www.coursera.org/learn/convolutional-neural-networks/discussions/weeks/3/threads/xpZORO2EQUCWTkTthOFAEQ
https://www.coursera.org/learn/convolutional-neural-networks/discussions/weeks/3/threads/nFHwetqgTKSR8HraoMyk2Q
https://www.coursera.org/learn/convolutional-neural-networks/discussions/all/threads/hHcad5WPS9W3GneVj8vVtQ
https://www.coursera.org/learn/convolutional-neural-networks/discussions/weeks/3/threads/ceyCHYbVS_Osgh2G1RvzwQ
https://www.coursera.org/learn/convolutional-neural-networks/discussions/weeks/3/threads/KcJ1QOefSUuCdUDnnwlLwA
I have tried searching some of the key terms that I remember from the above threads, but I haven’t been able to find any of the above content.
Yes, when the course team decided to migrate to using Discourse as the discussion platform for the new version of these courses, no mechanism was provided for automatically bringing forward valuable material from the Coursera Forums from the previous 4 years of activity. I totally agree that the particular posts you reference were incredibly valuable and well done. They explore all the aspects of YOLO in a much more detailed way than is available from the lectures.
So there are two problems here:
-
It would take a significant manual effort to port the material over. Those threads are quite detailed and there are also some lacunae (shall we say) in what Discourse supports relative to the forums. Discourse is better in almost every way, but it does not (yet) support LaTeX e.g.
-
As @ai_curious mentions, the original author of those posts has deleted them. I don’t think it would be ethical to “undelete” them and then simply appropriate that material. Well, there are really two layers of issues here: a) is it ethical to just copy them without permission and b) how will we support the material if there are followup questions without the assistance of the original author. I do not understand YOLO at anywhere vaguely approximating the level of depth in the posts in question. Please realize that the mentors are just fellow student volunteers and we do not get paid to do any of this support. I would not feel comfortable porting all that material without permission from the original author. But even with such blessing, I’m not sure I could muster the time and energy to recreate all of that material.
If you want to dig deeper on understanding YOLO and how to implement it efficiently, the best I can advise is to follow the references to the YOLO papers which Prof Ng lists and then do some web searching and see if you can find any material that covers these issues.
1 Like
Thank you @ai_curious and @paulinpaloalto for your responses.
I wasn’t aware that the author himself has deleted the posts.
@paulinpaloalto
Yes, I think there should have been some mechanism to get the valuable content from discussion forums to this platform. Manually copying content will require a lot of time and efforts.
Anyway, that leaves me with no other option but to go through the referred research papers, which is exciting but time-consuming as well.
Sorry, but to be fair it is not an easy problem. There is a huge volume of content in the old forums and most of it is “temporal” and has no long term value. It’s fundamentally a manual task to evaluate which things should be brought forward. Even in the case of “pinned” threads, someone needs to evaluate the material and make sure it is still relevant and perhaps modify it as appropriate. That is a huge amount of work. Which is all to say that there is no easy solution here and it’s pretty clear why things have worked out this way.
1 Like