Multivariable calculus in gradient descent for neural networks compared to regression

[high turret]

How does the multivariable calculus differ in gradient descent for neural networks compared to regression?

Am I correct in saying the multivariable calculus is more specifically partial derivates or are there more areas of multivariable calculus used?

[limpid pond]

Specifically about math part, I’m going to ping @zealous hamlet instead. But when it comes to comparing regression models and neural networks, it sounds like comparing apples to oranges. For example, you can use neural networks to solve regression D problems. Logistic regression is a linear model used to solve classification problems. Could you explain what is regression in your specific context?

[high turret]

hi thanks for the reply
i just finished Andrew Ng's course and he went over how to do gradient descent in neural networks, linear regression and logistic regression. my teacher wants me to write a couple sentences about what i've learned from it so i was going to say "I saw that partial derivatives and vectors are used to perform gradient descent in regression models and neural networks". is this statement correct or am i mistaking partial derivates for something else?

he told me to not make it too long XD so it just needs to be a quick summary

[limpid pond]

I never went through his course. Without the context, it just reads more like mixing up some terminologies. But I also think it’s better to just stick true to what you think you’ve learned rather than thinking there’s a correct answer. As in, seriously reflect on your learning, what’s clear, what’s not really clear yet, and summarize on that reflection, is much more profound than coming up with a correct statement

[robust prawn]

I am really rusty on the math parts specifically but it seems to me the biggest difference is using backpropagation vs working more directly with the gradients?

[high turret]

thank you

[high turret]

so linear regression uses gradients more whereas neural networks use back propagation?