I learned many things from this course. However, I think in some points it could have been instructed much better. But all in all, it is a very worthy course for the price offered. Thanks a lot!
It was really great learning with coursera and I loved the course. The way faculty teaches here is just awesome as they are very much clear and helped a lot while learning this coursea
교육 기관: Amalka W•
Course covers scalerble deep learning concepts
교육 기관: Ruan L D•
Sometimes great, some lectures not too good.
교육 기관: Andrey O•
Part with DeepLearning4J is not very good...
교육 기관: Deleted A•
Really Helpful course for AI Enthusiasts
교육 기관: Mobassir H•
pytorch instructor was the best <3
교육 기관: Jaime A L•
One of the best courses ive taken
교육 기관: Valerio N•
Very Complete course.
교육 기관: Aarti Y•
It was nice
교육 기관: Tobias H•
교육 기관: Pierre-Matthieu P•
I was pretty disppointed overall.
Pros : we see many types of tools and get to use some of them in the programming assignments. I feel like I now have a general knowledge of the field. I particularly liked the aspects of scaling and deploying models in production.
Cons : This honestly feels more like a rough draft than a finished and polished course. I would have liked a consolidated overview of all these tools, their pros and cons, etc. Some tools and techniques were explained in literaly 15 min(!) and in some cases simply walked through a tutorial from the tool's website (!!). A programming assignment was broken through not being updated for more recent spark versions. Some videos mentioned a non-existent programming assignment (I assume they were reused from an internal IBM training session), etc. The comparison with say Andrew Ng's course on ML is cruel.
교육 기관: Appan P•
Even though this course covers quite a bit of breath - in terms of implementation frameworks, there is scope of improving the presentation material. It will help a lot if the neural network models and the data transformations are explained using pictures.
Also, the one of the videos in the sequence of videos on LSTM for time-series forecasting (week3) talks about comparing performance of MSE and MAE but I could not find any such video on performance comparison.
Also, the assignments are quite simple and wish they had more steps for the student to "fill-up".
There is not much info on deploying the model and online evaluation of its performance. At least one video on how to do it in IBM data cloud will be helpful.
교육 기관: Jakob S•
The course covers some very interesting and important concepts, however on a very low level. The reason for this might simply be the lack of time; one cannot properly cover methods for AI image processing, NLP, etc. in such limited space. I also had mixed feelings about the exercises: It is very nice to see applications of the tools discussed in the lectures, but unfortunately the exercises are so simple that they can be easily finished without really understanding the code.
교육 기관: Manas S•
The course instructors are very experienced and knowledgeable but the teaching part has not been done very well. The assignments were not up to the mark, and an attempt to included too many topics in a very concise format was made. Some topics like Feed-forward NN in Keras were covered very well but most other things were a disappointment.
교육 기관: Jose L M G•
Lo hago, el curso es muy bueno en cuanto al uso de la plataforma watson, pero falla en explicar los fundamentos principales con animaciones, ejemplo, el curso de pytorch de udacity enseña eso muy bien. En lo demas esta bien, pero al no contar con elementos visuales de ayuda en laclase de LSTM se hace tediosa.
교육 기관: Edward J•
Not the best of the courses. Lots of content which jumps around. There is very little practical work so I'm not sure what I have learned there. However, it does give a good overview of the possibilities. The speakers were very clear but there needed to be more engaging examples and user input opportunities.
교육 기관: Jeet D•
The course is very resource heavy, i.e. it has great intuitive resources, but the learning experience was very poor. Some of the instructors were very sparse with the material contents, some just brushed over the contents without much explanation and.
The quality of the course has to be improved.
교육 기관: Sonja T•
Good material. Hard to understand the instructors' English. Not professionally presented. Assignments are too easy, and we didn't get good, meaningful practice. Quizzes often address information that either the instructor failed to present well, if at all, or made mistakes on.
교육 기관: Julián M•
You can learn several things from this course but you need to know Neural Networks and Deep Learning in advance. The content looks a bit disorganized but still pretty useful for day to day Deep learning implementations. Really cool the System ML integration with Keras.
교육 기관: Daniel P•
Too much focus on IBM platform, good overview on Keras/SystemML/DL4J though, some presentations could have been better prepared and implemented. Overall an average Coursera course and not a particularly great experience to work through the material.
교육 기관: Eugene N•
Something happened to the free 1CPU 4GB python environment on IBM watson studio. It is unavailable and so I had to struggle with Skills Network Labs instead. Please can this be checked?
교육 기관: Mohsen F•
it was not an advance course and course content and programming assignment is too easy. the whole course could bet finished in couple of days. video and presentation quality is poor.
교육 기관: Nhu N•
This course contain very good introduction to AI and DeepLearning. Although it assumed alot of knowledge already from the students in order to follow the content effectively.
교육 기관: Robert F•
Fairly okay course. Lectures were real hurried and high level. Had it not been for my Math and CS background I would not have gotten most of the material.
교육 기관: Ceren A•
Several lectures were superficial. I feel like I need to put a lot more time on my on to understand how to build a proper neural network model.
교육 기관: Michael M•
The programming assignments were easy. I wish it had some more difficult assignments requiring you to code your own neural networks