Hands-On Artificial Intelligence for Cybersecurity

Algorithm learning curves

A useful tool for identifying a component between the Bias and the Variance is important in determining the generalization error of an algorithm. This is the learning curve, through which the predictive performance of the algorithm is compared with the amount of training data. This way, it is possible to evaluate how the training score and the testing score of an algorithm vary as the training dataset changes:

(Image Credits: Wikipedia https://commons.wikimedia.org/wiki/File:Variance-bias.svg)

If the training score and the testing score tend to converge when the training dataset grows (as shown in the preceding diagram), ...