Although the course is great from a theoretical point of view, but it has two major flaws. First, it doesn't provide the fundamentals of Machine Learning but instead directly moves to Deep Learning, although building those concepts from ground up. Also, from a practical point of view, this course is really lacking as there is not a single explanation video on any of the coding aspect of Deep Learning and the videos that even exist just ask us to read through the Documentation to learn the practical aspect.

Much weaker than Stanford offerings. Strange buildup of topics for a breezy, but not particular accurate understanding. For example: multiple layers of a neural network is introduced before multiple category classification. Transfer learning is introduced incorrectly. The matrix representation of multiple features of an example with multiple examples is introduced very late in the course. The instructor is conscientious and seemingly knows the material despite using non-standard terminology. One wonders if he is primarily a teacher/researcher and rarely a practitioner. One wonders if Duke is a leader in machine learning research.

The course is okay, the teaching is helpful explaining the concepts of machine learning well. The problem is the connections between theory and practice. The assignments in pytorch are completely decoupled from the course materials and not explained very well. They expect you to code a model directly from mathematical theory with poor explanation of the pytorch framework and syntax.

This is very wrong thing that happen to us.Coursera tolds that this course if free due to our 9th birthday celebration.But when the time of claiming certificate .Its shows us to buy a certificate.This is cheating .I hate this app ..

It's really an amazing field to learn new things and from institute is like Amazing to me I've learnt more ...it's not at all boring and we'll will be excited for future experience with you 💯

Thanks Coursera and Duke University for this course. It is very insightful to get understood the basics of ML and applied ML in numerous fields. It really made me to move ahead with ML domain.

I felt that I took the best descition in taking this course, because the professors took this course with atmost clarity and made even the difficult concepts understand easily.

Thank you Professors

Very good introductory course, I highly recommend it to anyone looking to get a flavour of the methods behind the recent advances in AI without going into super-technical details.

Excellent course. Concepts such as gradient descent and convolutions as they pertain to neural networks are explained without going into the mathematical details but, in my opinion, are explained more intuitively and better, as compared to most other courses. The course does include some ungraded Jupyter notebooks exemplifying key elements of deep learning networks. Highly recommended to 'cement' understanding of neural networks.

The course covers all the topic's regarding the machine learning and has an excellent explanation of concepts and the slides are very easy to understand thank you for such a wonderful course !

A very concise and yet beautifully constructed course for introduction to machine learning for absolute beginner having basic knowledge of probability and mathematics.

Very good introductory course ,very well designed and professors explaination is very easy to understand .Go for it guys !

Happy learning !!!!

Sonic Somanna PK

This was a really great course for understanding the basics of machine learning through a lot of simple but relevant, real world examples.

The course is a good balance between learning key concepts and doing coding, the coding being optional. The phrasing of quiz questions and answers were sometimes confusing.

Good introductory course, quizzes are alright, but the coding assignments are a fail.

I liked the videos, they give a founded overview over the mathematical and conceptual foundations. In fact, you can watch at 1.25x speed with no problems.

The course skips topics like decision trees, knn, and r regressions other than logistic regression, as that is the basis for neural networks. But it's all good, there's another course here tha covers these. NNs and applications are covered well.

Quizzes are good, sometimes a bit ambiguous, but if you pay attention to the videos, well doable.

My biggest problem with this course is the programming assignments. I think they display a complete lack of didactical effort or understanding of their students' perspective.

I'm not talking about suggesting a python version from 2019 or so, that is incompatible with latest libraries which are also necessary for the exercises. Also the (recommended) Anaconda is probably the worst package manager ever known to mankind, I'm not going into detail. I ended up doing pip and it worked fine, just sorry for the lost time.

The biggest problem I had was the coding assignments. The labs just present some code examples and go like "you figure out the rest of it". Not even hints or some master solution to look it up, after trying by oneself for several days.

And even if you figure it out in the end, you're still left guessing why your 4 layer network does 96% accuracy, while the two layer network from the example does 99%. No comment, not even a ballpark figure how it should perform.

There is some help in the forums, but to me the descussions there show that most people are struggling with that.

Even having a reasonable background in linear algebra from uni and some Python experience I ended up looking up stackoverflow and other resources for hours. Especially the NLP assignment cost me several days, bc the lab creators think it's good to swap a few dimensions in the input, as opposed to the rest of the internet apparently. And skip over the batching in RNNs, because, yeah you go figure it out.

In summary, I'm not really convinced about this one.

On the one hand, good intro to understand the concepts and applications and you can finish in like two weeks, depending on how well you do on the labs, And of course if you can spend 1-2hrs per day on videos and quizzes.

On the other hand, the conding assignments are frustrating as hell and I don't feel like I got any routine in pytorch. I guess more, but simpler coding tasks separating individual aspects, and some kind of help to get to a solutions, I would have given it 4 or 5 stars.

It was great to touch new professional area and to understand its fundamentals. The course gives a broad view on machine learning, so I think now I really understand, what the machine learning is and how to use it in my work and even my political investigations.

The concept is explained in a great way but i didn't understood even a single part of programming part as no one explained that

Great course ! Pr Carin is clear enough to make you understand complex concepts like LSTM. The Math, calculus, algenra and prob are not too difficult. I enjoyed to follow this course ! To conclude a good introduction to ML to make you go deeper into the subject

A beginner like me who wanted to learn and expirement with machine learning but didn't know where to start, well this is the best course for you. The learning process is highly engaging and the concepts are explained in a well-refined manner.

I thought this was a great course to build up an intuitive understanding of a few different machine learning techniques. It is certainly skewed more towards breadth than depth, but this is unavoidable given the short length of the course.

The practicals are not at all aligned with their introduction. Specifically, in 2B you're asked to perform something that hasn't been introduced, and 3B could really use some hints. Also, you have no way to ensure that you actually complete the practicals as intended.

The course has a nice and concise presentation of applicative deep learning theory. It is not "machine learning" that is stated in the course name.

A serious drawback is that all the practice exercises are based on TF v1, which is currently obsolete.

Nothing is well defined, it looks like the classes are taken from other programs and put together in this "introduction". For instance, you have a professor in week two whom was never seen before, saying that we addressed a never seen before problem "previously".

Too tough

The evaluation/assignment labs are really difficult (the bar is high at the beginning specifically). But the course itself is excellent in general. The teacher is a definitively a fantastic one, with obviously great knowledge, but also nice fine teaching skills in addition. I'm quite impressed and utterly satisfied with that course. This course include LLM applied to text. Would there be any future module (7?) related to understanding GenAI basics for image or sound, it would be even more complete and amazingly awesome.