Machine Learning (ML) Learning Curve

A Machine Learning (ML) Learning Curve is a graphical plot that represents a ML model's learning performance against the amount of learning/training.

• AKA: Training Curve.
• Context:
  • It can visually demonstrate a model's improvement in predictive performance as it learns from a gradually increasing amount of training data.
  • It can help in diagnosing problems with learning such as overfitting or underfitting by comparing training and validation curves.
  • It can (often) be utilized to determine if adding more training data will likely improve model performance.
  • ...
• Example(s):
  • One illustrates how the accuracy of a Support Vector Machine (SVM) model increases as the size of the training dataset increases.
  • One illustrates the decrease in error rate of a Neural Network as the number of epochs (complete passes through the training dataset) increases.
  • ...
• Counter-Example(s):
  • A Receiver Operating Characteristic (ROC) Curve, which plots the true positive rate against the false positive rate at various threshold settings.
  • A Precision-Recall Curve, which shows the trade-off between precision and recall for different threshold.
• See: Cross-Validation Holdout Method, Plot (Graphics), Mathematical Optimization, Loss Function, Computational Learning Theory; Generalization Performance; Logistic Regression; Overfitting.

---

---

References

2024

• (Wikipedia, 2024) ⇒ https://en.wikipedia.org/wiki/Learning_curve_(machine_learning) Retrieved:2024-2-22.
  • In machine learning, a learning curve (or training curve) plots the optimal value of a model's loss function for a training set against this loss function evaluated on a validation data set with same parameters as produced the optimal function. Synonyms include error curve, experience curve, improvement curve and generalization curve. More abstractly, the learning curve is a curve of (learning effort)-(predictive performance), where usually learning effort means number of training samples and predictive performance means accuracy on testing samples. The machine learning curve is useful for many purposes including comparing different algorithms, choosing model parameters during design, adjusting optimization to improve convergence, and determining the amount of data used for training.

2011

• (Perlich, 2011) ⇒ Claudia Perlich. (2011). “Learning Curves in Machine Learning.” In: (Sammut & Webb, 2011) p.577
  • QUOTE: A learning curve shows a measure of predictive performance on a given domain as a function of some measure of varying amounts of learning effort. The most common form of learning curves in the general field of machine learning shows predictive accuracy on the test examples as a function of the number of training examples as in Fig. 1.

---