Shell.ai Hackathon

Challenge Over

Optimize loss function

3 years ago

we have a loss L= a*COSTcd +b*COSTdm +c*COSTif

I far as I can think that we can reduce a loss (a*COSTcd +b*COSTdm) by Gradient D but any idea how to deal with the 3rd loss term?

Any idea relevant to this?

Replies (1)

Kirushikesh DB

I don't understand how you will use the gradient descent algorithm. this is a constraint satisfaction problem, gradient descent gives an optimal solution that may not always satisfy the given constraints.

3 years ago