Why are neural networks so effective?

Visuals to demonstrate how a neural network classifies a set of data. Thanks for watching!
Support me on Patreon! https://patreon.com/vcubingx
Source Code: https://github.com/vivek3141/dl-visualization

Here's the course I referred to in the video. I am not affiliated with NYU.
https://www.youtube.com/playli....st?list=PLLHTzKZzVU9

Sinusoids as activation functions:
https://openreview.net/forum?id=Sks3zF9eg
https://vsitzmann.github.io/siren/

Here's the distill.pub article:
https://distill.pub/2020/grand-tour/

Special thanks to Alfredo Canziani and Nikhil Maserang for reviewing the video.

And also thanks to Grant Sanderson himself for giving me some manim tips!

I've been active on twitter, follow me here!
https://twitter.com/vcubingx

Join my discord server!
https://discord.gg/Kj8QUZU

These animation in this video was made using 3blue1brown's library, manim:
https://github.com/3b1b/manim

Music is from GameChops (Route 113, Azalea Town, Ecruteak City

Follow me!
Website: https://vcubingx.com
Twitter: https://twitter.com/vcubingx
Github: https://github.com/vivek3141
Instagram: https://instagram.com/vcubingx
Patreon: https://patreon.com/vcubingx

What does a Neural Network *actually* do? Visualizing Deep Learning, Chapter 2

0:00 Intro
0:18 Recap of Part 1
1:57 Introducing the dataset
2:52 Structure of the Neural Network we’ll be using
3:34 What is softmax?
5:52 Input space decision boundaries
6:24 Modifying the Neural Network to visualize what it’s doing
7:36 Out-of-domain boundaries
8:46 sin(x) as an activation function
9:30 Neuron planes
11:57 Softmax surfaces
13:20 MNIST Transformation
13:42 Outro