I have really loved this specialisation, but I want to be able to apply my knowledge as soon as possible. So far, I feel like I have learnt a lot of theory behind DL, but if asked to make a NN from scratch on a raw dataset, I will be at sea. Could you give me any advice about how I can learn all the other things that go into making a DNN, like data pre-processing, working with various DL frameworks and libraries etc?
Also, specific to CNN, is there any beginner friendly literature that I can try to imitate so that I can grasp the concepts better? I know about MNIST’s handwritten numbers data set, and I will try to implement a CNN on it once I finish this course, but if anyone can suggest other exercises that would be great too.
did you have a look at the recommendations here?
It is a great idea to apply your knowledge to new problems! That’s the best way to make sure you understand the course material and also to learn all the other things that you mention that aren’t covered by the course. @arosacastillo has given you a link that will give you lots of books and articles that cover deep learning and related issues. The MNIST dataset is a great thing to use for experimentation. Once you get done with that another source of problems to work on would be to look at the Kaggle website. They’ve got a huge number of “challenges” that provide you with input datasets and ask you to build systems to solve various types of classification or identification problems.