Unlike most of the Deep Learning knowledge which can be found in literature and other MOOCs, this course provides you with insights that can only be acquired trough (often painful) trial and error. Here you learn how to approach Deep Learning projects, how to avoid most common mistakes, and how to quickly identify errors in your model.

Do yourselves a favor, and finish this course before taking on your very first DL project.

This turned out to be quite an interesting and useful course, despite the fact that at first it seems boring without any programming assignments, but really teaches you how to steer an ML project towards good performance. I think it is worth extending it to touch more on how to tune a model (like architecture, hyperparameters) based on interpretation of bias/variance/erroneous results, perhaps with some real-world examples.

This course is my favorite so far. It has really given me a way to systematically and strategically set up a machine learning experiment and iterate in a way that make sense. For me the toughest part of ML projects has always been figuring out where and how to start. Now that I have some solid guidelines to follow, I don't feel as anxious about jumping into a new problem and it turning into a wild goose chase. Thanks a lot!

Such an amazing course. This course should be done by every Deep Learning researcher or enthusiast. If the previous course taught how to debug Neural Networks then this course teaches how to perform advanced debugging on the dataset. The contents of this course are invaluable and not taught in most institutions and they are not known to the wider community that actually follow the guidelines and techniques presented here.

Wow! This course was more of real time application scenarios and the kind of tweeks to build, transform learning plus multi-task learning was excellent. The end-to-end learning with a split approach of solving was really something new I found in this learning. Not to forget the application level quizzes were so tricky it was challenging to understand and interpret the possible solutions but, was great learning experience.

More of a general comment for the specialization but I love the Andrew and the teaching team have set up the content delivery. Simple, clear and well-paced delivery with consistent use of well-considered examples. On top of that, the summaries are great representations of key concepts. I am greatly appreciating the entire specialization and seeing the bigger picture in terms of why it is structured as such. Thank you!

I find this particular course in the whole specialization especially useful so far. Andrew teaches great strategy that helps think and act on deep learning projects in a more systematical way, and does so with crystal clean examples. Quizzes in this course, similar to flight simulator, are great at ensuring you can apply the principles you have just learnt and see where your understanding is a little bit vague.

Fantastic lectures combined with case-studies for real world applications. In this course you don't program, but don't underestimate the ability to abstract out and systematically assess your thinking. This could speed up your project development and save you tone of time. Any potential employers would also be happy that you know some practical aspects of implementing a deep neural network for a particular use.

Honestly speaking, this is the best course in the whole deep learning specialization. This course is the one which tells us what to do as a Deep learning engineer in real world scenario.

People can do the coding and everything, but without proper directions the product might fail.

Andrew Sir has given his expertise in a very neat and compact way, good enough for starting our own research or whatever we want to.

Amazing 3rd course, I learned so much related to error analysis, bias, variance, data mismatch, data synthesis as well as transfer learning, multi-task learning, end to end deep learning and more. I really loved both case studies in the end of each week. The interviews especially with Andrej Karpathy was my favorite :) Excellent best practices and strategies that you don't learn from any other course or book.

This course in the specialization is less about how to build a model but gives you a structured way of how to approach a deeplearning project. It shows how much some manual (and maybe boring) counting can speed up your project and that starting with a simple model and iterating on that often outperformes very detailed thinking about your project at forehand.

The practical tips Andrew shares are very valuable!

Liked this way more than I thought I would. Machine learning project management is vital in a professional setting (I would assume), and I often leave it as an afterthought. It's not just building the fanciest model, it's about how to iterate from an okay model to your best model in an efficient manner. This course teaches you what to look for with your results and pinpoints what areas to focus on to improve.

Very nice introduction to the aspects of a machine learning project that is not covered other places, but is very important. Most of it is very intuitive and comes as no surprise, but it is still very usefull to collect it into a single course. It is a good resource to have in case you are in doubt about how to structure your project, where to focus your energies and how to make progress in a systematic way.

Great course by Andrew Ng, coming from his Machine Learning Course and seeking to work on Kaggle Competitions, this course provides all the knowledge necessary to approach any machine learning problem (with or without a team), and efficiently work towards a better algorithm. It's almost as if he gives you the tools necessary to optimize yourself which in turn allows you to efficiently optimize any algorithm.

This great course provided quite a few good ideas for addressing scenarios that may arise when constructing a deep learning model. In terms of improvements, would suggest updating the name of the course to 'Structuring Deep Learning Projects' rather than 'Structuring Machine Learning Projects,' as the model-specific techniques (and examples for other techniques) all mention types of Deep Learning models.

Hello Andrew, I am a big fan of you. Learning from your every course. Very unfortunate that I can do that remotely only.

One thing that I want to mention - Can we have lecture notes on coursera, just like the way used to in CS229 that we can read before coming to next lecture. I found that that was very useful in understanding when things get harder.

Thanks hope we can improve coursera in that matter.

Based on personal experience with other courses on deep learning, much of the insights focus on what I'd call the "technical details" (e.g., maths and computations, using libraries, training strategies). I recommend this course for focusing on the "non-technical" aspects.

I find the lessons in this course helpful in streamlining the planning/strategizing process in the projects that I have been part of.

Good engineering tips, tricks, bolts and nuts, very useful! Andrew Ng is more dedicated to engineering and best practices that are very important in the machine learning field which is not only theory (a lot less than some believe or pretend) but very empirical, with a lot of practices, try and and error, recipes. Don't be snob, maths are awesome but good engineering and best practices are crucial too.

A really helpful course on how to take decisions when approaching a problem, how to make error analysis and what conclusion to draw from that. What to give importance to have a more robust model. This courses learning materials are what will be making a difference between someone who just trying out things randomly and someone who knows how to approach a problem to get better solution in a quick time.

The practical aspects of what to consider and what not to worry about while getting started with DL projects were very helpful. It puts us in the right direction. The whole field seems to be a little overwhelming, but watching the experts breaking things down to more easy and understandable concepts with relevant examples is very much encouraging for a student like me to dig deeper and make progress.

My Sincere "thank you" to Andrew Ng for teaching me ML and Structuring ML Projects. I find the content and presentation are on the highest level; which will definitely make the learner to think and workout in the correct direction when working in ML engineering and/or managing ML projects. I believe, in the coming times I'll learn more relevant topics from him to excel in my career.

Thank you again.

Excellent course - a unique short course where you'd get tons of insights from one of the top AI/ML experts Andrew Ng about how to curate data and structure your ML projects. Lot of practical and actionable tips.

Most useful course of the entire specialization to help you understand soul of the AI/ML development, you'll appreciate it even more if you have some experience of real-life AI/ML projects.

Thank you for providing this course. This course is something different, it takes you away from the technicality of the algorithms and makes you focus on a different but very important aspect of ML problems, i.e. error analysis. The professor is once again great in compiling all his practical real life experiences in teaching a subject which is not commonly found in other online training curriculum.

Great topics and discussions.

I get the dynamic aspect of writing the lecture notes in the videos, however the lecture notes should be "cleaned up" in the downloadable files (i.e., typos corrected and typed up). Additionally, the notes written in the video could be written and organized more clearly (e.g., uniform directional flow across the page/screen rather than randomly fit wherever on the page.

This course was helpful in basic undestanding of how to evaluate the data from deep learning models.

It took very diffrent aproaches like the precision and recall metric and even get faster evaluation with a f1 score. It was also helpful to get insight on diffrent types of errors which could show some direction how to optimize dev and test sets and why it is possible to pass beyond human performance.