Neural networks without calculus: how neurons work?
This is my submission to the #some3 contest.
Neural networks are some of the hottest topics in artificial intelligence, but they are usually taught in a way that assumes knowledge of calculus. I will attempt to cover neural networks and their training without relying on calculus or any other advanced math concepts. In this video we introduce and train a neuron, which is the building block for neural networks.
⏱️TIMESTAMPS
0:00 - Intro
0:26 - Neuron
2:20 - Aggregating inputs
3:37 - Activation functions
5:42 - Bias
7:30 - Training the neuron
9:26 - Prediction and outro
The video was made using the community version of the manim library https://docs.manim.community/en/stable/ (originally by https://www.3blue1brown.com/).
Further resources:
Neural network: https://pyimagesearch.com/2021/05/06/...
Backpropagation: https://en.wikipedia.org/wiki/Backpro...
Logistic function: https://xaktly.com/LogisticFunctions....
Technical details omitted in the video:
Gradient descent: • What is Gradient Descent?
Learning rate: https://en.wikipedia.org/wiki/Learnin...
Batch learning: https://en.wikipedia.org/wiki/Online_...