C5: Loss Function Landscapes
1D Curves
2D Landscapes
Configuration
Loss Function
MSE (Regression)
MAE (Regression)
Binary Cross-Entropy (Class)
Hinge Loss (SVM)
Prediction
Target Value (y)
1.0
y=0
y=1
y=0.5
Predicted (ŷ)
0.00
Training Landscape
Surface of a simple linear model:
y = w*x + b
fitted to 3 data points.
Run Gradient Descent
Reset Weights
Learning Rate
0.1
Loss L(ŷ, y):
0.000
Gradient ∇L:
0.000
Description of current state...
High Loss
Low Loss
Current Weights