Hey guys. Today there’s been a new announcement about a new MOOC on coursera on a deep learning specialization. More info on

Basically, 43€/month, you get 5 courses by Ng & Co and earn your certificates.

What do you think about it, given the fact that we followed and loved


I think it’s good to have some “competition” or other options, and good to have both paid and free material, coming at the problem of educating AI from different angles.

It’d be great to hear a comparison from somebody who’s reviewed both of these courses.

I suspect each course has its merits. Kudos to anyone who wants to further education in this area =)


I think it’s not as much about competition as it is about pushing the widespread adopting of deep learning (the democratization of ai)

I just wanted to add that you can access all lectures/assignments for free in “audit” mode

1 Like

In “audit” mode we can not neither see the assigment nor submit it

I can access all the assignments while being in preview / audit mode.

Here’s the first programming assignment :

I just realized that only I can open this link.

In audit mode, I can see the quizzes but not submit, and I have access to the notebooks.

I’m very curious about this too. I tried Ng’s old course on Machine Learning, but it was a bit too math-y for me. And Octave… :expressionless:

I really like the “for coders” approach of Rachel & Jeremy, and love Francois Chollet’s book “Deep Learning with Python”. However… During the course I also watched some of the Standford CS231n courses and they were quite useful, the math made much more sense too me.

So I’m tempted to give a try. But if Ng starts talking about Octave again, then I’m out. :smile:


From what I understood, Python is the de facto standard for deep learning at the moment! So no more Octave :smiley:

There is a nice bonus to the first course: interviews with “deep learning heroes” – including Hinton and Goodfellow. I would sign up for those alone!

1 Like

I did the original Stanford (before coursera) Machine Learning class in 2011 at (nowadays redirects to Coursera).

Based on that experience I have to say that Andrew Ng is one of the best teachers I’ve ever encountered during many MOOCs.

Of course Jeremy is one of the best teachers also, but Andrew and Jeremy are using different teaching methods, which - for us as students - gives the best of the both worlds! Best of the teachers in the world teaching the same thing using different approaches.

Personally I’m at least going to to audit (ie. not paying) the Coursera Deep Learning course.


Well… I’ve subscribed (paid, with a 7 day free trial)

I just wanted to add that you can access all lectures/assignments for free in “audit” mode

Is that true? I don’t see the “audit” link for this course, only the “7 day trial - Enrol now!” buttons.

The option to audit the course is only available when you click the enroll button on the individual courses.
The option will not be shown if you click on the enroll button that is displayed on the landing page of the deep learning specialization.
Therefore you have to search for the courses individually and sign up for each one separately in order to audit the courses for free.


For those of you that haven’t used Coursera before

These are the 3 out of 5 courses that are currently available:

If you open the links and click on enroll, then you have the option of clicking on “Audit Only” which allows you to see all the videos as well as the assignments. Since the assignments are Jupyter notebooks, you should be able to download them through File > Download as > Notebook.
But in order to be able to submit the assignments and get marks and feedback on your assignments, you have to sign up on Coursera. Pretty much if you don’t care about the certificate and can figure things out using google/stackoverflow you should be able to get the most out of these courses without subscribing!

I have glossed over the Coursera courses, and it roughly corresponds to part one of fast ai.
Part two is what I think is the best way to approach a field like deep learning,which is rapidly changing.
It showed me how to read state of the art papers , and now without a lot of fuss ,I find it easy to understand/implement the absolute bleeding edge in deep learning.

1 Like

Part two is what I think is the best way to approach a field

I 100% agree, and I believe that Rachel&Jeremy’s approach is much more effective for a “coder” audience: first get the grips looking at the code and building toy problems, then look deeper into the theory.

The only thing I would change in is making it less computer vision-centric by spending more time in other domains (e.g., NPL? sequence generation? RNN in general?). Also, a wish for part3 is to spend more time on unsupervised techniques.

ps: thanks @simon and @danialk


Hello Friends,

I wrote an article about the deep learning courses by Andrew. I drew comparisons to our courses. Hope it helps!

“Thoughts after taking the courses”


ok, Thank you. I have not seen it before

Thanks, nice one!

By the way, I completed the course a couple of days ago. I can say that the first three courses are really easy, beginner level, if you come from having completed the two courses (even online, like me).
It’s extremely useful as a review of the basics, though, given the fact that courses are hands-on, while Ng’s are a bit more on the theory-side (nothing impossible to understand, of course).

Will gladly pay a full month when CNN and Sequence Models will come out.


Has there been any indication, to when the remaining two courses of the specialization will be released?