Its takes to long to train the models (6 hours each case). I was lucky because some students give me de advice to use google Free GPU to complete the each train in 2 minutes. Without the students tips It is impossible.

I suggest to add to this course information about to use either cuda with own local jupyter lab or either recommend some serice like googel colab to get better performance.

I guess the course was OK, but the complete limitations of the "provided" compiling environment is inexcusable. I tried five different emulators, eventually using Google's Colab tool in order to get any type of results within six hours. I don't know what you can do, but something needs to be done.

Also, I would recommend NOT having two different tracks (Keras & Pytorch). Because of the aforementioned coding issues, most of the real instruction occurs in the discussion forum. It is INCREDIBLY confusing when there are essentially two different assignments posting questions in the same space. Also, can you do something about all my "classmates" asking for people to review their work in the forum? In my opinion, that is NOT the purpose of the discussion forum. As the admins, can you please just delete those requests to make it easier to find the REAL discussions. You are smart computer scientists, can't you create an AI to filter all those posts into the bit-bucket?

Course opened late. One instructor did not prepare his materials and the exercises were not even accessible. I will not be purchasing from Coursera in the future because of this specialization course. Truly a huge waste of money.

Is this really a professional certification ? Right now I don't think so. These tasks are for very very beginners. Not for professionals..

Most of this course is lab-work. However, the lab environment was inadequate. It kept crashing, disconnecting, or went to slow. While I understand that the Lab is a 3rd party tool, my payment was made to Coursera, hence they need to help ... at least by extending access for the lost time.

But need to study extra as these topics are not taught like Transfer Learning

Not enough instructions as I wasted many hours without going to Google Colab. Please change it to AI review of project rather than peer review as most times there is insufficient submissions.

This course was incredibly frustrating. There were errors in the quiz questions out of one's control, which won't let one proceed without the correct answer to mis-written questions. It takes many days for instructor/and course assistants to respond to course forums. The IBM cognitive labs are too slow/not usable/crashes consistently and IBM Watson has severe credit limitations. Ultimately, searching the course forum threads from other students posts and running Google Colab on GPU were the only way I made it through. I definitely expected more from Coursera - more instructor interaction, working software and more TA visibility/guidance.

Course lecture are well done, but labs don't work properly. IBM should be able to do better. I am very familiar with Python and Jupyter notebooks. I have taken many courses with similar labs and have had almost no issues. I am disappointed, and will move on to AWS and Azure. When you move to the cloud, your stuff needs to work or people move on to the next service with a couple of clicks of a mouse. :-(

Heavily flawed A.I Grading system. Failing and invalidating those reaching this point.

A very nice project based course to get hands on experience with deep learning

and transfer learning.

The Course is good, The labs were crashing which were causing lot of issues in completing the course

This is an excellent course if someone wants to learn transfer learning. However, having said that, there should be another task for which students should build their own model and compare its accuracy with the predefined one. With this, students would get insights as to how to build a deep learning model from scratch.

Great course content! One thing that can be improved is the Skills Network portal. It's incredibly difficult to train any of the pre-trained models there.

This is a good enough project if it is your first Pytorch implementation. However, the program is unevenly difficult, with very few information for week3 assignment, and almost copy/paste assignment for week4.

Capstone project had delayed for a month, and after that the course resource also not ready.

Code is littered with non-fatal errors, inconsistent spacing in every line. Many core methods used are deprecated by a decade, have no use anymore, but their implementation is a key focus in the course. All quiz questions have zero connection to the topic, library or application, but are exclusively focused on the exact details of the preceding labs, like "which of these smooth identical looking concrete pictures was the third sample of the fifth batch from the test set?". The dataset sizes in the labs are also absurd, unzipping literally broke the provided environment multiple times, taking hours at a time to complete, just to crash in the end. The training times are just as absurd, spent the last 4 hours training a model on the provided dataset with the required parameters for a single epoch, and it's been stuck on the last 57 seconds for over 30 minutes, unlikely to finish. Just pain, and practically no use use.

Everything depends on AI & ML now. Before enrolling in Coursera’s IBM AI Engineering Professional Certificate Course, I had zero knowledge about AI & DL. After completing the course, I am able to use these applications properly. It proved to be a total career game-changer.

Coursera has proven to be a wonderful experience. I have taken classes from other organizations, but there’s no comparison to Coursera.

Highly relevant examples were given which made me get a clear idea of the concepts explained. Deeply appreciate Coursera & IBM for designing this course very well.

Kudos to Instructors Dr Joseph Santarcangelo & Dr Alex Aklson, Ph.D. for providing such a wonderful learning Journey.

This is an excellent course to culminate the IBM AI Engineering Professional Certificate. Just a friendly reminder, unless someone has a baseline knowledge of Applied Data Science, starting this Professional Certificate without completing the IBM Data Science Professional Certificate first will be quite challenging. Well done, IBM! To all the instructors, much appreciated for sharing all your industry experiences and knowledge. Thank you!

I got a chance to put what I learnt into practice and the idea of choosing between Keras Track or PyTorch Track was very beautiful. I can suggest another track for TensorFlow, making it a choice between choosing from 3 tracks instead. That would feel more complete.