Deep Learning: Convolutional Neural Networks in Python (Udemy.com)

The course description was perfect, and the course itself delivered what I needed. I'm a very conceptual learner - I pick up concepts and then strengthen my understanding of them through application and practice. This is very much how The Lazy Programmer presents the material. I love the maxim, "learn the concept, not the syntax." Most importantly, the course helps you learn how to learn about Convolution Neural Networks in Deep Learning based on the depth of The Lazy Programmer's experience. Good job.

Taking the course is worthwhile. It is a conglomeration of videos that I record at various times. It's easy to follow, it doesn't get tangled up in complicated mathematics. It answers the questions and that adds up to a lot of points. Sometimes he uses different nomenclature for something he's explained before, but as I said before, since it's easy to follow, it's no problem. You see the CNNs and the layers are explained but only the forward propagation. Although it explains different ways to update a weight in a matrix: SGD adam, etc. I feel that This course need the explanation of the backpropagation chain rule in ANN and CNN.

It is a phenomenal course - I have really enjoyed the mathematical lectures to give a bit of background into the WHY - as well as excellent detail on why we use the functional Keras API, which in my other courses I haven't really learnt why there is a switch. Finally we are doing some pretty impressive models on good benchmark datasets, rather than solely playing around on MNIST. This is a nice change, and means I feel more confident to pick up other benchmark datasets, or to collect my own. Would highly recommend this course and instructer!

Complementing my CS and Data Science double major in college, this course was perfect. It covers a broad range of material, explaining concepts in a way that is both comprehensive and easy to understand, and presenting both the intuition behind important concepts and their calculus / data science deeper reasoning (which mostly complements and expands upon material primarily taught in data science or mathematics majors). I thoroughly enjoyed this course, and I feel much more confident pursuing my data science major, my career path, and my newfound passion towards artificial intelligence. I would say this course would be applicable to a relatively broad range of students; particularly those with passions for mathematics, computer science, or development, and a medium current level of depth into the CS, DS, and/or ML fields. It also provides opportunities to inspire students to try out what they learn on their own projects. In other words, it is both comprehensive and well-rounded, and guided me towards both a general understanding and a will to further explore essential concepts. However, be attentive that this does require significant self-motivation (the enjoyment of learning new skills for your betterment and curiosity) to succeed, as this is a video-based course.

The technical lessons are good and very easy to understand for a new learner, although I would have liked to go more in-depth at times (e.g. choosing the hyperparameters, diagnosing an issue with the architecture or the training, talking about network optimizations such as quantization or pruning, etc). The course have a lot of additionnal sections at the end of it, which are just repeating what was said in the already fairly long introduction : a little trimming would improve the experience. Still the best course I have taken so far on Udemy.

It's great that the course has the review and in-depth sections for people who didn't take the earlier courses, or just need to refresh their memory. Convolution is explained really well in several different ways, and the code and exercises are a great start for anyone who wants to do CNNs. One thing that I would love to see is for the NLP section to be longer, or at least give more details about the convolution for text part. Also, for those of us who didn't watch the tutorial for RNNs, the NLP section starts a bit abruptly and some details are a bit confusing. Overall, a great course to take.

This is one of the more thorough courses on CNNs I have taken. As usual, Lazy Programmer does a deeper dive into the mathematics behind the technology than most online instructors out there, but you won't be doing any differentiation by hand like you did for the Softmax function in the Deep Learning course. Most of the code is written in TensorFlow, so you also won't be writing your own neural network from scratch. Parts of the course feel a bit disconnected; I suspect some lectures were peeled from the RNN course and then got Frankenstein-ed into this one. Overall, it's a great intro into how CNNs work.

It was quite an intense course and I needed to interruped a lot to learn more in depth about some of the mentioned topics. Nevertheless, I like it very much to be forced to step deeper into these topics to get a better unterstanding. Good would have been to know more about the learning structure mentioned on deeplearningcourses.com before. I liked very much the differeent perspectives on explaining the problems. I think I learned a lot.

This has been another superb course. I'm really glad to have taken this course and I certainly recommend it. That said, in comparison to the previous courses, it feels a bit incomplete. The mathematical derivation of the back-propagation part of the convolution is noticeably missing. Since most courses abstract most of the mathematical part, this only removes the last little half-star from the rating to reach the "beyond expectations" part. It's still an AMAZING course and you should take it.

Yes, I admit I was lacking some elements (although having the mathematical foundations it was easy to move through) but other parts were almost trivial. Actually, looking at the coding is ok...but having some comments and variables that have descriptive names is better, to relate what I am seeing to the theory described below (also, seeing typos as it is being entered while listening is distracting). Having the comments ahead of time would let me review some feature that has been discussed and to refer back to relevant sections if he refers to previous section of the code (also - focusing on methods/method names would be good rather than details of the function).