←
1
2
3
Layer by Layer
Stack layer on layer and solve harder puzzles.
🎯
Step 1 of 3
What you'll learn in this level
- What the blueprint of a neural net is
- Why more layers can learn trickier patterns
- Why more layers are still not automatically better
Nets Made of Layers
Deep Learning is all about layers! The more layers a network has, the more complex patterns it can learn.
Think of it like a sandwich. You need the right ingredients (layers) in the right order to make it delicious (smart).
Kinds of Layers
- 1
Input Layer
It takes in the raw data. In a photo that means the pixels.
- 2
Hidden Layers
This is where the real work happens. These layers look for features and find patterns.
- 3
Output Layer
It gives out the final answer.
- 4
The blueprint
The way you order the layers is the blueprint. Experts call that the architecture.
Ready? Now you design a net yourself.
Guess first, then experiment
You give your net 20 layers instead of 2. Does it solve the task better for sure now?