Mean Squared Error (MSE)
A regression loss function and evaluation metric that calculates the average of squared differences between predicted and actual values, penalizing larger errors more heavily.
Formula
MSE = (1/n) * sum((predicted - actual)^2). The squaring makes MSE more sensitive to outliers than MAE. RMSE (root MSE) is often reported for easier interpretation in original units.
Usage
Default loss for regression problems. Also used in image reconstruction, denoising, and as a component of more complex losses. When outliers should be penalized heavily, MSE is preferred over MAE.