dm.cs.tu-dortmund.de/mlbits/intro-kinds/
Different Kinds of ML – Lecture Notes
Digit recognition:
\(x\) =scan image,
\(y\in \{ 0,1,2,3,\ldots ,9\}\)
➜
Sales prediction:
\(x\) =time series,
\(y\in \mathbb {R}\) value at time \(t+1\)
Regression
Usually easy to evaluate on the training [...] reserved unless otherwise noted.
Supervised Learning:
Given
a training set \(D=\{ (x_i,y_i)\vert i=1\ldots N\}\)
Find
a good approximation function of \(f\) such that \(f(x_i)\approx y_i\) even for [...] \(\sum _i C(f(x_i), y_i)\) .
Unsupervised Learning
Given
an unlabeled data set \(D=\{ x_i\vert i=1\ldots N\}\)
Find
a plausible, compact model \(M\) that explains the observed data well
Contrary to …
