Cost Function

A function that quantifies the discrepancy between a model's predicted values and the ground-truth labels, providing the mathematical objective for neural network training. Common forms include Mean Squared Error (MSE) and cross-entropy loss.