The codes should have be explained ...

Too general for practical usage

good as an introduction

Fairly shallow.

I first took Deep Learning Specialization by Andrew and then took NLP specialization by Younes Mourri and then this course. One difference I noticed that Andrew explained all the stuff by himself in detailed 8-10 min video and here these in these two coursers, the two instructor explained concept in 1-2 min video and left the remaining concept to learned by ourselves through notebooks. Andrew put much more effort than these two guys.

Did a good job of explaining some of the terms and processes involved in using AI for medical diagnosis, but the flow and organization of the course were really poor and the methods taught were not general enough to be able to extrapolate to use in new ways outside of the course.

There is no reply or response to discussion forums from the instructors and assessment of the assignments are always zero so I can not pass to week two even my assignment's outputs are matched with the correct ones .

A far too brief overview of AI applications in medical diagnosis - only really covers image analysis and even then is cursory at best. Disappointing as I have found the other deeplearning.ai courses quite helpful.

Not useful. Probability distributions are not intuitive mostly.

This course relays on "add one line" code too often.

T​he course was quite shallow, and the actual challenges of model selection, training or building appropriate augmentation steps were pre-built and not discussed in any detail.

The coding challenges were using badly outdated package versions, for which documentation does not exist anymore and which do not represent best practice usage of the libraries involved.

O​n top of that, the coding challenges expect a very specific solution, while not considering equivalent implementations as correct (case in point: In the week 3 coding challenge, I used np.transpose where the challenge used np.moveaxis. I prefer transpose since it clearly and explicitly states where _all_ the other axis go, while moveaxis makes that change of state implicitly.)

Finally, the grading of the last coding challenge does not respect the special cases that are explicitly mentioned in the excercise itself. The "standardize" function to be implemented explicitly mentions the possibility of a slice having a zero standard deviation and the pre-coded framework handles this special case correctly. However, if one changes the selection of the slice in the cell before, which the user is encouraged to do, it is possible to obtain an empty slice. The grader expects a unit standard deviation though, without checking this edge case.

T​he shallow content and lackluster excercises, as well as the common mistakes in the presentation videos (sometimes corrected by a "question" popup during playback) do not give the impression this course was prepared well.

I have done several of AI courses including the TensorFlow. While the TensorFlow course, gives you a neat and excellent hands on on how to build a network from scratch or implement easily a CNN such as Inception V3, this course make you confused as what sort of aim it follows. Overall confusing and not useful. Though you find some good stuff in the videos but the design and strategy of the course is meaningless.

L​ectures are short, mainly focused on programming details (how to subsample and image or how to calculate an error). The assignments do not help understand the AI part of the medical diagnosis. It can be considered as an intro course for the AI for MD.

No for medical students. Terrible time loosing

Very basic