First, I would like to say that the course is fantastic! However, I not able to really practice what I’m learning. The labs and the quizzes are great, but I was wondering if there are more resources that could help me practice training networks using what we learned in python or TensorFlow so I can become more proficient with these tools? I feel I just run the Jupyter labs and watch what happens instead of practicing and learning the syntax. This also goes for some of the formulas and getting those drilled into my head with practice.
2 Likes
Hello Erich,
what you could do run the same Jupiter assignment in your local Jupyter with codes removed like an exercise for you to know if you have learnt and understood what has been taught.
Also other way would be explore Kaggle, there you will have numerous project to exercise with but then you need to have a better understanding of python and programming for doing those projects.
Regards
DP
3 Likes
@Erich_Woessner I would advice you to get a linux computer, install a docker, if you have a gpu great, you can link it to the docker, if not, well, it is ok. Lauch a docker with jypiter, and start doing projects from Kaggle. That is a great way to learn.
There are also some free environments I think but they usually last some months, it might be cheaper.
There are definitely a lot of resources online if you’re looking to continue learning!
I recommend the Deep Learning Specialization course offered on Coursera as well. The first 2 courses reinforces and builds upon a lot of the deep learning concepts covered in the Machine Learning Specialization, and the last 3 courses covers even more complex systems (Convolutional Networks, RNN, Transformers, etc).
As others have mentioned, Kaggle is also a great website and helps you practice more practical aspects of machine learning.
1 Like
You do not need a linux computer for this.
2 Likes
Thanks I am actually doing the coursera specialization now. I found the labs to be great stepping through the material, but they almost helped too much because I felt I was just running the code they provided. Kaggle seems like it might be a good option. Thanks for your suggestion.
I do have jupyter up and running on my windows machines.
Thanks everyone for the suggestions.
True, you don’t need it.