Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Known results New results Risk bounds for linear regression
 

Summary: Known results New results
Risk bounds for linear regression
Jean-Yves Audibert1,2 & Olivier Catoni3
1. Imagine - Université Paris Est,
2. Willow - CNRS/ENS/INRIA
3. DMA - CNRS/ENS
October 2009
Known results New results
Framework
Least squares regression
Training data = n input-output pairs :
Z1 = (X1, Y1), . . . , Zn = (Xn, Yn)
A new input X comes
General goal: predict the corresponding output Y
Probabilistic assumption :
Z = (X, Y), Z1, . . . , Zn i.i.d.
from some unknown distribution P
Prediction function: f : X R
Risk: R(f) = E [Y - f(X)]2
Known results New results

  

Source: Audibert, Jean-Yves - Département d'Informatique, École Normale Supérieure

 

Collections: Computer Technologies and Information Sciences