Neural Network & Learning Visualization
Learning Parameters
📊 Current Step
Ready to start backpropagation
💥 Loss Function
Mean Squared Error (MSE): L = ½(y_pred - y_true)²
Current Loss: 0.00
🌊 Gradient Flow
Watch how gradients flow backward through the network
⚖️ Weight Matrix
🔢 Input → Hidden Weights
📊 Hidden → Output Weights
⛓️ Chain Rule in Action
The chain rule allows us to compute gradients for nested functions:
∂L/∂w = ∂L/∂y × ∂y/∂z × ∂z/∂w
Where:
- L = Loss function
- y = Network output
- z = Weighted sum before activation
- w = Weight parameter