If Neural Network is a Black Box for You, This is What You Need to Realize (Part 2)

Hello, I have been inactive for quite a while due to my hectic and busy schedule. P.S. I am a fourth-year student (:. My last article on this matter turned out to be successful. Therefore, I decided to write the second part on this. It was difficult to come up with some ideas to discuss; therefore, it is mostly related to CNNs and Transformers. If you haven’t read the previous one, I strongly recommend you to read it first to understand my thoughts.

I hope this piece finds its readers. If you have anything to add, please feel free to reach out.

If Neural Network is a Black Box for You, This is What You Need to Realize.

If Neural Network is a Black Box for You, This is What You Need to Realize (Part 3)

Why deeper and wider CNNs?

We usually say that deeper networks have better performance and generally can learn more useful features. But let’s look at a very simplified example and try to see it ourselves:

Image by the author