Hi Andrew,

It's just Awesome Andrew !.. it was a pleasure to achieve this course 8 years after I achieved your first course in 2011 (before coursera borns). Thanks to you and to all your team at stanford.

I can't wait for the next four courses :)

I was teaching the machine learning course to my students in the past 3 years and I plan to teach this current course to my students this year. They have the barrier of English language and I'm trying to do my best to explain the main ideas I understood from your course.

Abdelhak,

Professor at Mohammed V University

Rabat, Morocco

This course is a perfect introduction to neural networks. It builds on simple concepts to then put together larger processes that link and compound these concepts together. In all, the user begins to see how something like fitting a Logistic Regression model is not that different from some other learning models that are regarded as more complex, such as neural networks. Surely, as one continues, these things will diverge, but this course does a good job os using the knowledge users likely already have, in order to better introduce them to some more complex concepts.

I liked the course. Without any previous knowledge of neural networks or deep learning, now I can claim that I know the basics and the reasoning behind fundamental steps of deep learning. The slides that were downloadable also were a great point of this course, since it made it easier to take and add your own notes. The instructor also explained topics quite clearly. I also like the system of grading for programming assignments where were divided in different sections unlike some similar courses on Coursera. Totally recommended for people interested in this topic.

The best thing about this course was that the course gets easier and easier! Prof. Andrew Ng, the community, and the arrangement of the assignments always got my back! I never felt like "I must skip this line as this is out of my scope". With a little bit of hard work, anyone can build any layer NN for image classification problem with 80%+ accuracy. If there is any scope for improvement, I'd say that the notebooks get disconnected frequently, which should be improved. Otherwise, this is the best course I've ever had! Thanks all who are involved with this course!

Amazingly well done course. The best thing I liked about is the attention to detail that Prof. Ng has paid. For example I always had tons of problems with the rank 1 matrix. The frustration levels used to be so high. However, being the consummate practitioner and teacher, he identified what kind of problems one encounters when one learns python and deep learning for the very first time. It was more like symphony. I tried other courses in other websites. But this easily is the best of it all. I strongly recommend it to everyone who wants to get into deep learning.

This is an excellent starting point for learning about Neural Networks and Deep Learning.

Many technical derivations and details are left out but this is only a plus. These details are much better learned with a working knowledge of the basics/implementation of neural networks. Besides, it is clearly stated whenever such details are omitted. This course will not make you an immediate expert in coding nor neural networks but it is the best starting point out there for a broad audience. Regarding becoming an expert, always remember that Rome wasn't build in one day.

Great quick overview and introduction to neural networks and basic deep neural nets. Great intro for those without a lot of the required math background. I would have liked to see some more quizzes (even if optional) on the derivations of the gradients. That was a bit of black box and we were just given the equations. I also thought it was a bit odd to have examples-by-column rather than rows. Assuming this was done to simplify notation (less transposes) - but it's counter to almost every other presentation in machine learning and stats that use example by rows.

Well motivated. Clearly structured. Generalizing from Logistic Regression over shallow Neural Network to Deep Neural Networks was easy to follow and reinforced the structure of the approach. I overall liked the presentation of the maths and assume that it is well suited for an audience of differing affinity to maths. For myself, I will have to do the calculations again on my own to get a real grip on them. [Writing db (=something that should grow with steeper b) for dL/db (=which shrinks with steeper b, given the same change in L) still feels wrong.] Thanks!

Nice basic course, gives a clear look at what is happening inside neural networks, all details are explained in quite clear and understandable form with practical tasks of implementing everything, so that you really know what is going on.

After that course you will have a knowledge of how to implement a simple neural network and it's learning algorithm from zero. Also you will get some knowledge about matrices operations, derivatives and python programming.

I do highly recommend this course for novices and for more skilled people. It was a positive experience.

Excellent presentation of the material. The homework assignments made this approachable by holding my hand as I learned "how to walk" with matrices and multilayer neural networks. I feel like there could have been one more "do everything yourself" assignment, where we had to build another L-layer neural network completely from scratch, but maybe that isn't the point of this course, since I expect I'll be using keras or something in "the real world". An optional quiz involving some of the derivations for some equations might have been a nice stretch, as well.

A solid introduction into discussing the basics of machine learning. Although I had to research some details on specifics topics which I could not completely understand in the course, that was my own problem, really. The basic steps for creating a neural network and understanding the functions behind initializing parameters, forward propagation, cost and backward propagation are explained well, and since the assignments are in python, I've learned a few packages and helpful coding hacks to better implement efficiency in programs. Overall, I would recommend!

It is a great course to get started on Neural Networks and their practical implementation. The whole course is constructed keeping the end result of building an OPTIMIZED program in python for building a neural network and everything connects together in the final programming assignment. Not only do you learn what are neural networks and how they work but you also learn very importantly how to code in a very optimized manner so that you decrease the training time as much as possible. Definitely a great course, looking forward to complete the specialization.

This course is the best course to start learning deep learning. You will enjoy the step-by-step creation of shallow and deep neural networks.

Frankly speaking, I am amazed at the creativity and brilliance of the Andrew Ng's team for preparing the programming assignments and quizzes. Therefore, I am speechless.

Last but not least, I am also really grateful and thankful for Coursera who has given me Financial Aid for this course. In return for this act of kindness, I am able to finish it before any deadlines.

Thank you deeplearning.ai!

Thank you Coursera!

Thank

Great intro to deep learning. Although it's a bit repetitive at times, especially coding bits - one is not really forced to understand the components at times but can instead just follow instructions and copy paste bits and pieces. Would for example have appreciated that more time was spent on explaining the details of derivation of backwards propagation especially within "deep domain". The intuition is clear, but either forcing the user to do (or giving a link to) a step by step derivation would have been useful and saved time. Thanks for a great course.

I completed the original ML course earlier this year which gave the fundamentals of the practice. What I got out of this course was a reinforcement of the practices and ways of collecting my thoughts. There was enough difference in the approach and especially in the back prop areas to help clarify the understanding from what was a bit of magic, to a clear and more structured set of calculations. The platform of using the notebook is very solid, and of course there is the usual outstanding support from the community with respect to answering questions

Andrew NG's approach is one of its kind. Previously, I had taken several courses with other reputable online providers, and also did a lot of reading in tandem. What amazed me about Andrew's approach was the fact that crucial concepts were explained in much detail, one-by-one; this helped me complete the overall puzzle and/or fill in any missing links. I'm not sure if I could've followed the course without any previous experience, but if you're familiar with Python, NumPy and basic ML concepts, then this course will help you understand DL a lot better.

I think this course is very accessible, and gives you enough know how to hit the ground running. My only caution is that there is a bit of hand-holding involved (because of the limited background they assume), and that if you want a more rigorous foundation, you'll have to supplement this course with other materials. This doesn't detract from the quality of the course though, rather, it's amazing how much you can do with Deep Learning without fully understanding all of the finer details, and this is a good place to springboard into more advanced study.

After a full course of Machine Learning, of course, this one is rather weak. The feeling that all 4 weeks we are talking about the same thing. This is probably done for those who are not at all in the subject. I see this course as an introduction to the specialization. I hope the continuation will be stronger. It's great that practical work is done in Jupiter on Python. Program exercises are easy, but it takes a lot of time to figure them out if we don't know Python very well. This is not a plus or a minus, just a statement of fact. Thank you Andrew!

This was an amazing course for me. I've always wanted to get to the bottom of deep learning fundamentals and this course did not disappoint. It walks me through the basics to the more deeper concepts in incremental steps without overwhelming me with too much derivatives (but just enough to carry the point across). Just the right mix of theory and practice. Highly recommended as a starting point for deep learning, or if you're like me, developing more intuition towards the practice that I am already doing. Fills in the gaps in my understanding nicely.

A very nice introduction to deep learning. Covers the basics and builds up slowly. There is some prerequisite knowledge of Python programming and calculus to have success with the course. Professor Ng's explanation of the topic is focused on practical applications, and builds on years of experience gained in academia and industry. The exercises are focused on mastering core concepts. The notation takes a little bit of time to get accustomed to, but you begin to understand why the notation is the way it is. Very good course; I definitely recommend it.

A really good intuition and introduction to neural network and deep learning. What I enjoyed the most was the fact that we needed to implement the learning algorithm step by step through the guided programming assignments as opposed to calling an in built function in libraries ( such as tensorflow etc). I felt the programming exercises were quite very successful in an attempt to draw and maintain the learner focus on the algorithm itself as opposed to other programming aspects, which can be learnt elsewhere/improved elsewhere. Great course. Thank you

This course makes you implement your own neural network without using Tensorflow or Torch. As a result, the student gets to learn what neural networks are implemented internally instead of only learning how to use a particular software package. The course is full of small, practical, and highly useful information such as why we use a cross-entropy loss instead of sum of squared errors loss and why do we need to initialize parameters using not-too-large random weights. This information is very useful in implementing NNs at work or for job interviews.

It's a very good course. It illustrates the idea of neural network and deep learning in an intuitive way. I think this time I fully understand the idea and details behind them. Also, the python programming is very friendly. I have used R for years but not so familiar with python. However, folloing the instructions I can do the coding very efficiently. I think i just spent less than 1 week on this course but get 100% score on it. So it's not so challening compared to Machine Learning and PGM. I think PGM is the most difficult one among these courses.

Dear Prof. Andrew,

It is my pleasure to express gratitude and thankfulness to you and your team.

I am grateful to have you as a mentor in learning AI for everyone, neural networks and deep learning. It was a great journey with you in this learning process. Lectures and assignments made me realize the importance of the ANN and other advance tools in real world applications. The mathematical content behind neural network theory and programming assignments encouraged me to pursue this area in future.

Thank you once again.

Dr.Hari Krishna Gaddam, India

Overall, an excellent course! The material is taught very well. The programming assignments were enjoyable and fairly straightforward. The Jupyter programming notebooks were really cool and fun to work with.

The only criticism I have is that week 1 material was extremely easy, easily doable within a day. Week 2, on the other hand, was quite difficult. I think it was the most difficult week overall because it introduced a huge amount of new concepts and math. After I had a good understanding of week 2 material, the rest of the course was not so bad.