I like this part the most among DLS. The approach is very reasonable and can be applied to other fields. It was very rewarding. I would like to recommend it to my friends, but I feel sorry that my friends would not understand it because it is explained through the mask of deep learning.
In the first video of W1, NG mentioned that ML strategies keep changing. If you have a better ML strategy or approach based on this ML strategy lecture, please share it. thank you!!(the news that ng has turned favorable to early stopping is also good)
1 Like
Dear @sunblockisneeded,
Thanks for sharing your thoughts, it’s great to hear you found this part of the course so rewarding. I agree, the approach is very practical and can be applied well beyond deep learning.
I understand your hesitation in recommending it; the deep learning framing can be a bit much for those new to the field. Still, the core ideas are powerful.
And yes, ML strategies are always evolving, Andrew sir’s shift toward early stopping is a good example.
If you or anyone else has tried different approaches based on these ideas, I’d love to hear about them!
–
Girijesh
Keep Learning AI with DeepLearning.AI