Model Performance Measurements

October 12, 2026 deep-learning

Handwritten notes on evaluating deep learning models: loss functions, accuracy metrics, confusion matrices, ROC curves, and regularization techniques.

These notes cover how to measure and improve deep learning model performance:

Loss Functions: Cross-entropy, MSE, hinge loss — choosing the right objective for your task.
Classification Metrics: Accuracy, precision, recall, F1 score, and when each matters most.
Visualization Tools: Confusion matrices, ROC-AUC curves, precision-recall plots, and learning curves.
Regularization: Dropout, batch normalization, early stopping, and their effect on generalization.