Gained a lot of insight on how to structure machine learning projects, but I believe it would help for this course, and for the deep learning specialization to put lecture notes after each video in order to get a short and concise summary of all the relevant info we need to know, like the one in Andrew's into to ML course; however, Andrew is an insightful teacher, so I had to give this course a 5.

There is a tendency to dive straight into applying ML to a problem and with the tools available today this is all too easy. It therefore becomes necessary to make sure that we are aware of how we can structure the process of machine learning. How we organise our skills/intuition/measures is what this course is about. Essential and in some ways as fundamental as the scientific process itself.

very good guidance on how to start a machine learning project, including many interesting discussions including how to choose the size of training/test/dev set, how to analyze the errors, how to deal with mismatched distributions of test/traning/dev set by adding a training_dev set and how to do end-to-end and multitask training. The contents are well exercised by two well defined case studies.

This is the best series of ML that I have taken so far on Coursera. Andrew Ng is a master at instructing others. I cannot say enough about this series, you would need to take the series to comprehend what I am trying to say. Somedays I watch and I am just amazed how Andrew takes a concept and turns it comprehensible at such a fundamental level. Great course it deserves more than 5 stars!!!!

Excellent Course on how to structure the Machine Learning projects so that the developers do not waste time following a random trial and error approach and rather take on an approach which is proved to work well in improving the accuracy of the model in spite of the changing requirements and data. I would like to thank Professor Andrew N.G. and his team for developing such a wonderful course.

Another great course from deeplearning.ai. Again, Many thanks! to Andrew Ng and Coursera. Great lectures as usual. I have learned lots of new concepts and methods such as Dropout Regularization, RMSprop, Adam Optimization, Learning Decay, Batch Normalization, etc. I think the assignments were also helpful but not enough to absorb what I have learned. I'm looking forward to practicing more.

This course was very interesting. The examples are good chosen and the exams have great questions (they are summarising everything from the lessons). Great suggestions and also personal tip. I'm studying and I'm learning a little bit of these neuronal systems and machine translation which are based on language while your examples were more visual like the car case for example. Thank you :)

This Part of study is a aimed to improve the skills during the Modelling and Calculation. In the realistic problems, people need time to get familiar with the process of how to build a sophisticated network. And the time to learn these experiences could be long. This course give us a lot of useful information and tricks. It saves our time and reduced the hardness for the work! It's great!

I wish there was more such cases that I can learn from. I found this course very valuable. Thank you :)

I would be interested in participating in research. Do you think that Coursera could help with creating PhD degree/ applied research. I would like to improve the world by applying the knowledge I gained from this specialisation. Do you think Coursera could help with something like this?

Really thoughtful course on which less emphasis is put less on the actual work on the learning models and more on the meta-data and the provided criteria as well as general strategy. It is crucial one know these topics as very few are the places such information is taught, for the material is slightly overshadowed by the actual making of models and long-term perspective can be neglected.

This course should be mandatory for any machine learning practitioner, researcher, or student. Ng shares excellent insights and provides a clear structure for thinking about how to manage our most valuable resources in machine learning -- labeled data! The course discusses the concepts in a deep learning context but I would recommend even for those not working on deep learning problems.

I took this course soon after completing the Machine Learning course, before starting the Neural Network and Deep Learning. And found it extremely helpful, the simulator approach takenup in the course is absolutely spot-on and unique to this course (as compare to any knowledge source on internet).

Andrew NG has poured in his tacit knowledge and made it explicit in the best possible way !

This course gives insight to all the errors and their analysis, different approaches to deal with problems in machine learning and also working of different models such as Face recognition, Speech recognition and Automated driving models. Andrew sir explains all this concepts in a very learnable manner. I do recommend this course to those who are going to build their first ML model.

One of best courses I have taken on Coursera. There are not much available online resources to learn about how to structure and manage a Machine Learning projects. I would like to express my appreciation for all of the hard work and dedications professor Andrew Ng and his team spent on designing such a great course with understandable lectures as well as well-designed assignments.

This course is the most hands-on deep learning class I have seen so far... and have taken a lot. Most courses focus on the technical details of feedforward, backpropagation, activation functions, etc. but this is the only one I have seen where guidance is provided on how to tackle real-life situations. So far, the BEST course I have takes on deep learning projects tips and tricks.

This course is light on math and programming but loaded with great advice that I have already been able to put into practice at work. Some things are lessons I have learned by being in the field for a few years and others are lessons that might have taken a while to learn on my own. This course has extremely valuable real-world advice that will impact the work I do right away.

I understood such concepts as: evaluation metric, percentage of distributions, estimating train and dev set errors,

training a basic model first,

choice

softmax activation,

carrying out error analysis

on images that the algorithm got wrong,

algorithm will be able to use mislabeled example,

dev and test set should have the closest possible distribution to “real”-data, and so on.

This course offers insights into organizing and structuring machine learning projects. It is different than the other courses of this specialization by not going to much into technical details. I found it still very rewarding since Andrew offers some very niche tricks that can help researchers in practical application of machine learning and deep learning algorithms.

Great job!

The topics discussed in this class are very closely associated with the title `Struturing Machine Learning Projects`. These topics are more than just concepts, I think they would be very useful in real projects (Though I haven't done one :) ). There are a lot of use cases discussed in the course. Hoping in the near future, I have an opportunity to use them in practice.

This course was smaller and a bit more theoretical than the previous two courses. Although the lectures give you a good insight on error analysis, things to check in order to optimize your model and finally how you can use a pre-trained model to solve a different task - of the same input data type.

Thanks both to the instructor and the crew for this great series of lectures.

Really changed my thinking about how to run an ML project. I just wish my projects were the kind that could exploit these methods to the fullest. They're more like the autonomous driving example. There are parts that DL is useful for (particularly sequence learning with RNNs) but big parts that aren't (e.g. use of probabilistic graphical models). Anyway, awesome course!

Lots of principles and skills about how to organize machine learning projects and diagnose problems. Especially for the error analysis part, you will definitely save much more time in solving these errors than you expected by following the suggestions taught by Andrew. Thanks Andrew, I really learned a lot from your awesome deep learning courses and felt closer to industry.

This course is just magical. It covers so many concepts that would require years of experience to gain. Thanks to Professor Andrew for sharing his great knowledge with us. The bias/variance and train and dev/test distribution concepts are very well explained with examples. Also, the quiz helps to practice these concepts which require a better understanding of all of these.

Great course as always! Andrew Ng is a great teacher, and he actually can inspire all of us on being better professionals (and researchers) on the field. The idea of the case studies was great! It was very fun to experience how it is to be part of deep learning projects and the decisions associated with this. Congratulations for all of you guys from coursera! Thank you!

This short course focuses primarily on non-technical aspects of deep learning projects. The value of this subject matter is the focus on aspects that can make or break the success of a machine learning project. Given the fact that as much as 80% of deep learning efforts never make it "into production" (Gartner et al) spending time on these issues is highly recommended.