Hello! I hope you are doing well.
I’ve completed this course and now wondering about creating an ANN model from scratch (not using TensorFlow). I’ve tried for two days, taking course repeatedly, but I still failed to build ANN from zero.
For example, for linear regression, the steps are to define:
- compute_cost
- compute_gradient
- gradient_descent
- run_gradient_descent
How can we create ANN like this? Optional lab uses the w and b values from the TF while explaining ANN using numpy: Dense layer, etc. Kindly guide me on this, I will be highly indebted to you.
Regards,
Saif Ur Rehman.
1 Like
It would be worth considering taking the Deep Learning Specialization next. It is a more advanced course and Prof Ng will show you how to build a Deep ANN in python/numpy there. If you take the entire series, it also covers ConvNets (CNN) in C4 and Sequence Models (RNN) in C5. In all three cases, Prof Ng will show you how to build them directly in python and then how to use them in TF.
1 Like
Thank you, sir. Just need to confirm one thing: do I need to take the unsupervised course too, to understand deep learning (ANN)?
I’m not sure what you mean by the “unsupervised course”, but DLS only talks about Neural Networks to solve Supervised Learning problems (with labelled data).
By unsupervised, I mean the third course of MLS. But I got your point. No need to take it to understand DL. Thanks a million, sir.
Sorry, as you can tell, I haven’t taken MLS yet. I agree that the MLS course on unsupervised learning is independent of DLS. You could do the two in either order and you would be fine.