Who am I?

I am Prakhar Srivastava.

Junior year, CS undergrad, currently working in the domain of Deep learning.

Social handle: @prakharcode

Since I’m still a learner, I would love if you heckle me and not harass me.

What will this talk cover?

MATHEMATICS!

and basic code implementation

Where it is used?

Huh, dude every neural netwok!

Let's Begin

So what is Backpropogation?

Backpropogation is bascially backward pass of gradient in a Neural Network, generated by cost function's derivatives.

Literally!

So according to definition we need:

• Neural Network

• A (cost) function to get gradients from

•  Linear Algebra

Output function

NEURAL NETWORK

COST FUNCTION

SOFTMAX