deep learning book ch5

5.1 learning algorithms

machine learning tasks:

 1) classification
  f: Rⁿ -> {1, ..., k}

 2) classification with missing inputs

 3) regression
  f: Rⁿ -> R

 4) Transcription
  ex) optical character recognition, speech recognition

 5) Machine Tranlation

 6) Structured output
  output is a vector
  subsumes transcription and translation tasks
  ex) pixel-wise segmentation

 7) Anomaly detection
  ex) credit card fraud detection
  이상한 것을 잡아냄.

 8) Synthesis and sampling
  ex) speech synthesis, tacotron

 9) Imputation of missing value
  algorithm is given a example x ∈ Rⁿ  but with some entries x_i of x missing. algorithm provides a prediction of the values of the missing entries

 10) Denoising
   ex) image super resolution

 11) Density estimation or probability mass function estimation
   machine learning algorithm is asked to learn a function p_model : Rⁿ -> R, where p_model(x) can be interpreted as a pdf or pmf on the space that the examples were drawn from.


machine learning Experiences:

 1) Unsupervised learning
 2) Supervised learning
 3) reinforcement learning