MACHINE LEARNING TECHNIQUES - LASA
MACHINE LEARNING TECHNIQUES - LASA
MACHINE LEARNING TECHNIQUES - LASA
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
99<br />
5.8 Support Vector Regression<br />
In SVM, we have consider a mapping from the input data X onto a binary output y =± 1. Support<br />
Vector regression (SVR) extends the principle of SVM to allow a mapping f to a real value<br />
output:<br />
f : X → °<br />
( )<br />
x→ y = f x<br />
(5.58)<br />
SVR starts from the standard linear regression model and applies it in feature space. Assume a<br />
X →φ<br />
X ∈ H. Then, SVR seeks a linear mapping in<br />
projection of the data into a feature space ( )<br />
feature space of the form:<br />
( ) φ ( )<br />
f x = w, x + b, w∈H,<br />
b∈° (5.59)<br />
To recall, SVM approximates the classification problem by choosing a subset of data points, the<br />
support vectors, to support the decision function. This sparsification of the training dataset is the<br />
key strength of the algorithm. When considering non-separable datasets, SVM introduced a slack<br />
variable to give room for imprecise classification. SVR proceeds similarly and tries to find the<br />
optimal number of support vector while allowing for imprecise fitting. The allowed imprecision of<br />
the fitting through f is measured by a parameter ε ≥ 0 and is measured through an e -loss<br />
function:<br />
{ ε}<br />
( ) ( )<br />
y− f x = max 0, y− f x − ,<br />
(5.60)<br />
ε<br />
Points with a non-zero e -loss function lie outside the e-insensitive tube that surrounds the<br />
function f , see Figure 5-10.<br />
Figure 5-10: Effect of a non-linear regression through SVR. The tightness of the e-insensitive tube around<br />
the regression signal varies along the state space as an effect of the distance in feature space between the<br />
support vectors (the support vector are plain circles). Datapoints within the e-insensitive tube do not<br />
influence the regression model and are indicated with un-filled circles.<br />
© A.G.Billard 2004 – Last Update March 2011