15.3 Classical ML: What it does (2)

  • All ML algorithms/models: Automatically find transformations that turn data into more-useful representations for a given task (e.g., coordinate changes or linear projections)
  • ML (technically): searching for useful representations of some input data, within a predefined space of possibilities, using guidance from a feedback signal
  • Classic ML algorithms (e.g., logistic regression): usually not creative in finding transformations; they just search through predefined set of operations, called a hypothesis space
  • So what is deep learning? Why is it special?6


Chollet, Francois, and J J Allaire. 2018. Deep Learning with R. 1st ed. Manning Publications.

  1. Further reading…Chollet and Allaire (2018, Ch. 1.1.3)↩︎