When should we use a
deep neural network or just a
Because in the course we are said in the assignment when to use this or that, but in a real-world situation how can we make our choice?
- It will depend on the task (Logistic regression won’t be powerful enough for things like NLP, object localization)
- Depend on what performance are you looking at (what accuracy are you looking for)
- Depend on what inference speeds are you looking at (regression can be lot faster)
- Depend on hardware limitations (IOT basic module won’t be able to handle bigger DL models)
and so on