Andrew has made some brilliant points but what really got my respect was in Course 2, Week 3, ethics lesson. He mentioned walking away from projects that were financially sound but “Killed the project on ethical grounds because it made the world worse off!” Kudos to Andrew! We should all aspire to those ethics!
4 Likes
Thanks for your message.
Which course are you referring to?
1 Like
The 2nd course in the Machine Learning Specialization.
I’d definitely recommend although i don’t share his love of gradient descent!
2 Likes
@oldwolfster why we do it this way, and why it is introduced early on, will make a lot more sense if you decide to move on to the Deep Learning Specialization.
The datasets are just much too large, and the number of nodes involved in a neural network, too many for an explicit solution. We are then representing highly dimensional spaces, and have to try and minimize our equation/loss, somehow.
So, at least with the present state of computing… This is the best way we’ve come up with (so far), to be able to accomplish that.
2 Likes
Gradient descent is like wearing a blindfold and feeling which way to go with a probing cane. 
1 Like
“Don’t worry about it.”
I agree except it seems a bit more difficult than just tapping down the wall.