MSE

Predict a continuous value with ŷ = wx + b. Derive the MSE loss and compute one gradient descent step from scratch.

MSE for regression, cross-entropy for classification, and the TD error loss in DQN — how loss functions guide neural network training.