Fundamentals of AI and ML
Model Fit and Bias-Variance
Model fit is about whether a model captures useful patterns without memorizing noise. AIF-C01 questions often contrast overfitting, underfitting, high bias, and high variance.
Concepts
- Overfitting means a model memorizes training noise and performs poorly on unseen data.
- Underfitting means a model is too simple or insufficiently trained to capture the signal.
- High bias often causes underfitting; high variance often causes overfitting.
- Regularization, more data, early stopping, and simpler models can reduce overfitting.
Exam tips
- Overfitting performs well on training data but poorly on unseen data.
- Underfitting misses the signal because the model or training setup is too simple.
- Regularization, more data, simpler models, and early stopping can reduce overfitting.