Hacker News new | ask | show | jobs
by shamino 2819 days ago
Just wanted to chime in - you mentioned Andrew Ng's Machine Learning Coursera lectures are outdated. I want to say that while this could be true, the reason he does everything in Octave is because he finds that students learn faster when programming in Octave. He's aware that Python is the leading practical machine-learning tool. His focus is on learning the fundamentals and gaining an intuitive feel for the algorithms. I greatly enjoyed using Octave, which I felt was actually faster than if I had done everything in Python.

With that said, I'm so excited to learn from your new course, and can't wait to start.
2 comments
jph00 2819 days ago
Andrew's new deep learning courses (http://www.deeplearning.ai) all use Python. When I last spoke to him he was no longer enthusiastic about using Octave/Matlab for teaching.
link
shamino 2819 days ago
I also saw that his new deep learning courses are in Python. He may have changed his mind about teaching in Octave for his traditional Machine Learning courses, but I still learned a great deal and I feel like I have a strong intuition of how ML algorithms (and neural nets) work.

There is some truth to Octave having a faster turnaround than Python, if you're new to programming. I feel like with Deep Learning you really have to bite the bullet, but his courses are just fine with Octave/Matlab - with respect to getting an intuitive feel for the algorithms.

link
jph00 2819 days ago
Yes I do think his ML course is still wonderful, as I said in the post :) However I don't think that Octave is better than Python for teaching or understanding machine learning concepts.
link
shamino 2819 days ago
I just don't think his courses deserve a knock, or that your course will be better because it uses Python. Both courses can be great :)
link
shamino 2819 days ago
I'm going to go out on a limb and say that it's probably for similar reasons that you chose PyTorch over Tensorflow :)
link
rahimnathwani 2819 days ago
Thankfully, someone has translated the assignment skeletons from Andrew's course into Python, so there's no need to use Octave if you go through that excellent course today:

https://github.com/dibgerge/ml-coursera-python-assignments/b...

link