Loss Functions
AI · Intermediate
Definition
A loss function measures how wrong predictions are; training minimizes it.
The choice of loss defines what the model optimizes.
Diagram
Flowchart
Example: MSE & cross-entropy
import math
y, p = [1, 0], [0.9, 0.2]
mse = sum((a - b) ** 2 for a, b in zip(y, p)) / 2
ce = -math.log(0.9)
print(round(mse, 3), round(ce, 3))Output: 0.025 0.105
Key points
- Cross-entropy pairs with softmax/sigmoid
- MAE is robust to outliers
Interview questions
Why not MSE for classification?
Cross-entropy gives stronger gradients for confident wrong predictions.