How to Decide Neural Network Architecture?
There are a lot of explanations elsewhere, here I’d like to share some example questions in an interview setting.
When building a neural network model, how do you decide what should be the architecture of your model?
Here are some tips for readers’ reference:
The short answer is nobody knows, so you’ll have to test it.
There are two things to take into consideration:
First, you should try several (e.g.10) different architectures, train your model for each design and compare the performance on your validation dataset. You can figure out what’s the right network size this way.
Second, assess how complex your problem is and determine your network complexity accordingly. Typically we use deeper networks with highly complex problems.
Let’s check out how Stanford Lecturer Kian Katanforoosh explains it!
Happy practicing!
Thanks for reading my newsletter. You can follow me on Linkedin or Twitter @Angelina_Magr !
Note: There are different angles to answer an interview question. The author of this newsletter does not try to find a reference that answers a question exhaustively. Rather, the author would like to share some quick insights and help the readers to think, practice and do further research as necessary.
Source of video: Stanford CS229 Lecture 11 (Autumn 2018) — Introduction to Neural Networks by Kian Katanforoosh
Good reads: Blog. The mostly complete chart of Neural Networks, explained
Blog. Rules-of-thumb for building a Neural Network
