A subset of articial intelligence methods which 'learn' through training on real-world data sets. The model is then tested on a 'test' data set and initially assessed through an accuracy measure. The approach is specifically termed 'supervised learning'.

Supervised learning comprise both a variety of simple regression methods as well 'deep learning' appraoches such as Tensorflow. 'Deep learning' approaches are generally run under a GPU rather than a CPU and require specialist archetecture. 'Unsupervised learning' could be considered machine learning, but would be traditionally termed multi-variate statistics such as PCA.