I have done the initial Deep learning courses of Andrew, and they were very thorough and well explained. I was expecting the same quality, however, it was not so. Explanations were generally good, but the examples and the details around the architecture of the models were barely discussed or considered, besides pointing me to the next course (which I have done). I was a bit disappointed TBH, for an "applied" course I do not think this provides enough material to begin applying this knowledge into real life problems.

Really poor quality, sadly. The notebooks are full of errors, the quizzes are mostly coding questions instead of being about deeper understanding of the notions studied, and I don't think the videos are clear enough about what decisions are most important when building this type of model and how to make those decisions. Love the topic, but very disappointed, and don't think this is worth what I'm paying..

The course is merely a walk-through some Jupiter notebooks of Laurence. There are no proper slides with explanation of what's going on. I also don't see much activity from the course creators on the discussion forums. It is incredibly easy to complete the course without forming any deep understanding.

The weekly programming exercises are not even automatically checked for accuracy.

If you already have good knowledge of Neural Networks like CNN, RNN, LSTM, etc. then only opt for this one. Because they keep referring to previous courses in the specialisation for these. Also, they are only writing the code but never cleared about, what they are writing and why.

quiz's are pathetic! throughout the whole course the instructor talks about the advantages of RNN and LSTM and CNNs for time series prediction while not being able to prove this not even for one in the entire course, what a disappointment !

Me parece que no dan un buen enfoque a muchos puntos, los códigos no se explican bien, y abordan temas que son densos en minutos lo cual hace que quedes sin mucha información. No me parece que sea un buen curso por eso.

The course was not as detailed or in a flow like I expected from a deeplearning.ai course and the editing was also very bad, one thing was shown and something else was spoken.

A bit too high-level with lacking explanation on intuition. E.g. Conv1D was added to LSTM layers which helped reduce loss value, but did not go into the explanation of why.

Course is very quick and does not cover the topics in sufficient depth - explanations and discussion are all very brief.

Junk course. Andrew Ng is a great specialist but I'll never try courses from deeplearning.ai.

assignments are repeated. No need to have 4 different courses, each course can be one week.

This course will be good only for very beginners. It's not deep and challenging enough.

To make it better you have to develop more challenging and GRADED! exercises

Poor audio quality.. Cant even hear in full volume

Very bad quizzes, no challenge whatsoever.

Too much repetition in the content.

It is a surface-level introduction

Could have been better

Too easy.

Unfortunately this course does not do a great job in teaching in my opinion. The course states that it is around 25 hours, however the "teaching" portion of it totals to only about 2 hours of short videos, that don't do a great job of really explaining anything. The larger portion of the time is taken up by the labs, which are basically just sheets of unexplained code that you click play on to see what they do. At the end of each section, you have to take a quiz and a test. The quiz generally contains questions that are designed to just trick you into the wrong answer, with multiple choice options that all look similar. The tests are ok, however if you are unable to work out the answers there is no real help provided, apart from the ability to post on a forum for the course. In the forum you are unable to post any actual code, which means that the threads are just a mess of people trying to explain what they have tried without actually posting their code, and others trying to help them in the same way, which makes the forums practically useless to get help from. In the final test, you have to create a modal that needs to get MSE and MAE scores that are under a certain number to pass. These numbers are seemingly completely arbitrary, and since there is no available recourses to help yourself, and the forums provide little insight, the final test to attain the qualification just becomes based on random trial and error.

This specialization is false advertising. It does NOT prepare you for the Tensorflow certification exam. It’s especially disappointing after taking the fantastic specialization by Andrew Ng, and makes this specialization feel like a cheap cash grab by Coursera and DeepLearning.ai. This series of courses fails to prepare you for three reasons:

1 – The certification exam is done on whatever is the current version of Tensorflow (v2.6 as of writing). You can’t expect a specialization like this to update every minor release, but much of the coding is still on the v1.X version!

2 – The certification exam requires you to work in the PyCharm IDE. The IDE doesn’t even get a mention in this specialization and it is all done through Google Colab.

3 – The material is covered at a very superficial level. I was hoping to walk out of it feeling confident in using Tensorflow on novel problems, but I’ve barely learned anything about Tensorflow that I didn’t already get from Andrew Ng’s specialization. There’re a few minutes of lectures (some weeks less than 10 minutes). The programming assignments are either pathetically easy, or lack any guidance on what to do (seriously sometimes there’s no instructions at all, you have to guess what to do by the variable names), or both.

Save your time and money and go elsewhere to learn Tensorflow.

Even worse than the NLP course. Week 1~3 contains nearly no new material for tensorflow. It's just some replicated knowledge from previous courses. Studying synthetic data is good, but is off-topic for a tensorflow course. The course should focus on models and model structures for different types of time series data. My biggest complaint is that this course does not cover even the basic knowledge required by the tensorflow certificate exam (as advertised). Where is the multivariate time series forecasting? This is the most important part of the exam but the course totally neglects that.

Extremely shallow and sloppy made course. It is sad to see that the optimization done in the neural network is at the very least non-robust (if not totally random). The techniques used are simple illustrations that one can find better in youtube videos for free. The fact that people have to pay for this course is basically a scam, you should be ashamed of yourselves.

Maybe I had wrong expectations from this course. But to me it felt like the material in this course was extremely superficial. I was hoping to learn something, but it turned out to be a very basic overview of the material. Everything boiled down to "compile + fit" without the explanation of nuances associated with time-series settings.

The module on time series did not help at all in the certification exam. It's full of simplistic examples and broken links and optional assignments. All the while, the new iteration of the exam is more complicated and touches on topics not covered in this workshop on time series. Very disappointing.