Back to Catalog
Neural Networks: From Scratch
No frameworks. No magic. Build your own Deep Learning library from zero using only Math and NumPy.
12 Weeks Difficult
What you'll build
Understand Backpropagation calculus
Build a custom Autograd engine
Implement Dense Layers & Activations
Master Stochastic Gradient Descent
Create a Digit Recognizer (MNIST)
Your Progress
Track your journey through this course.
0% Completed0/12 Modules
Your progress is stored locally on this device. Export it to continue on another device.
Curriculum
Frequently Asked Questions
Why not just use PyTorch?
To master PyTorch, you must understand what it does for you. We build the magic so you can control it.
Is there Calculus?
Yes. We visually explain Derivatives and the Chain Rule. You cannot hide from the math here.
What will we build?
A mini-library like Keras, capable of training on real image data.