Do the machine learning lectures uploaded to Coursera cover the contents of cs231n?

Hi there,

cs231n is a deep learning program for computer vision. You can find the official here â†’ https://cs231n.stanford.edu/

Here are the lecture series taught by Andrej Kaparthy (one of the deep learning heroes) â†’ https://www.youtube.com/playlist?list=PLkt2uSq6rBVctENoVBg1TpCC7OQi31AlC

Just taking a quick look at the cs231n information that Sithu has provided, the closest match would be the DLS specialization here: C1 covers fully connected nets for image classification and then DLS C4 (Convolutional Nets) covers a lot of the other more advanced topics mentioned in the cs231n syllabus. Not all of it is covered and Iâ€™m guessing that a graduate CS course at Stanford will present the material in a more mathematical way than the Deep Learning courses, but Prof Ng is also a Stanford professor and taught cs229 for many years. His lectures are excellent. It would be worth taking DLS C1, C2 and C4 first and see how far that gets you.

Is knowledge of machine learning required as a base, Right?

You can take DLS with no previous ML experience. The prerequisites are good competence as a python programmer and a solid knowledge of math up to the level of basic Linear Algebra. You need to be very comfortable with vectors, matrices and operations like matrix multiplication, but you donâ€™t need anything as sophisticated as knowing what an eigenvector is.

If you want to get a more general â€śsurveyâ€ť of Machine Learning first, you could start with the MLS Specialization. That does an introduction to a wider variety of topics, but DLS covers only deep neural networks, which are the core technology used in most computer vision systems.

Yes, you should first take the DLS, it is more fundamental than cs231n. It will give you the solid foundation.

I have a questionâ€¦

If Iâ€™m bad at math, should I give up on machine learningâ€¦?

If you are new to machine learning, donâ€™t worry too much about it now. You can proceed with MLS and DLS.

But if you want to deep dive into deep learning, like how your model works, for better debugging, then it is worth to know math.

Could you please set the order of the lectures?

Machine Learning â†’ MLS | DLS?

You should first start with MLS if you donâ€™t have any previous experienced in ML. Then DLS.

@sithu Does MLS stand for â€śmachine learning specializationâ€ť, a Coursera course?

Is DLS a â€śdeep learning specializationâ€ť?

yes

Oh, Thanks !

I am currently taking course 1 of the machine learning specialization.

That is awesome! Enjoy your courses

@sithu Broâ€¦ I donâ€™t know quadratic functions and Iâ€™m really ignorant of math. In this case, how can I take the MLS classâ€¦?

Is it better to skip the math-related parts and gain knowledge about the theoretical parts?

DeepLearning.AI has math specialization if you want.

Khan Academy has math courses if you just want to learn the required math while doing MLS.

Iâ€™d suggest you should get the foundation math knowledge, but this is just my opinion. Everyone is different.

Not having experience with quadratic functions is a little worrying, but those donâ€™t really play that much of a role here. As I mentioned earlier on this thread, the math that you cannot avoid in machine learning is Linear Algebra. You will need to deal with that everywhere. You need to be familiar with vectors and matrices and know how operations like matrix multiplication work. If you donâ€™t have at least that basic knowledge, it is going to be challenging. There are lots of ways to learn enough Linear Algebra. The DeepLearning Math for Machine Learning (M4ML) Course 1 will give you that, although it may be a bit of overkill because it covers other Linear Algebra topics like Gaussian Elimination that arenâ€™t really used in MLS or DLS. The Khan Academy has a nice LInear Algebra program.

The other prerequisite knowledge you need for MLS and DLS is solid competence as a python programmer. Both MLS and DLS involve lots of programming assignments and they are all in python. They teach you about numpy (the linear algebra package for python), but they do not teach you python: you need to already know it.

No, because youâ€™ll need to understand some math in order to implement the programming labs and gain practical skills.

I will write a comment about the results in 3 months.

See you in 3 months!