I did not get a clear understanding on the architecture of neural networks, their deployment in real world, concepts about gradient descent.
If you have just come to Module 2, you have just run a model like a blackbox. We will next go into each component and do it in detail. For example, model design, data preparation, optimization (gradient descent) are all individual modules which you will explore in depth later.
I am more used to traditional ppt, pdf video based study. I want to understand books for examination.
Here are a few links if you prefer the traditional good read. Note that since we are exploring a practical course with continuous ongoing developments, our notes are more of "documentation pages" rather than textbooks. However here are a few which gives you concepts about the basics.
Text Book- gives you a general background: https://www.deeplearningbook.org/
More helpful resources for your course are provided below
Videos- A more practical oriented guide : https://www.youtube.com/playlist?list=PL_lsbAsL_o2CTlGHgMxNrKhzP97BaG9ZN
Problem statements often don’t specify which factor should be prioritized—for example, edge deployment vs. high accuracy vs. real-time performance. So I’m unsure how to decide which factor should carry the most weight when comparing and selecting the best solution.
Even if nothing is mentioned, the decision depends on the requirement, how critical is the problem? how accurate does my answer need to be, how quickly does my model need to run. For example, if you are solving a plant disease classification task, you have the luxury of using a larger model as its not a time critical problem. However, if you are doing autonomous navigation and your model is slow, it will be a disaster. So the decisions are problem dependent. You decide the criticality, the accuracy required, the speed vs accuracy tradeoff allowed.
I need a lot more practice
Here are some basic tutorials for you. You may go through the general flow and treat some units as blackboxes to be explored later.
https://docs.pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html#
https://docs.pytorch.org/tutorials/beginner/transfer_learning_tutorial.html
https://www.kaggle.com/code/geekysaint/solving-mnist-using-pytorch