Sciweavers

378 search results - page 2 / 76
» New Bounds and Approximations for the Error of Linear Classi...
Sort
View
TIT
2008
123views more  TIT 2008»
13 years 7 months ago
Geometric Upper Bounds on Rates of Variable-Basis Approximation
In this paper, approximation by linear combinations of an increasing number n of computational units with adjustable parameters (such as perceptrons and radial basis functions) is ...
Vera Kurková, Marcello Sanguineti
NIPS
2001
13 years 8 months ago
(Not) Bounding the True Error
We present a new approach to bounding the true error rate of a continuous valued classifier based upon PAC-Bayes bounds. The method first constructs a distribution over classifier...
John Langford, Rich Caruana
COCO
2009
Springer
106views Algorithms» more  COCO 2009»
14 years 1 months ago
Improved Approximation of Linear Threshold Functions
We prove two main results on how arbitrary linear threshold functions f(x) = sign(w · x − θ) over the n-dimensional Boolean hypercube can be approximated by simple threshold f...
Ilias Diakonikolas, Rocco A. Servedio
JAT
2007
100views more  JAT 2007»
13 years 7 months ago
A Sobolev-type upper bound for rates of approximation by linear combinations of Heaviside plane waves
Quantitative bounds on rates of approximation by linear combinations of Heaviside plane waves are obtained for sufficiently differentiable functions f which vanish rapidly enough...
Paul C. Kainen, Vera Kurková, Andrew Vogt
ICML
2010
IEEE
13 years 8 months ago
Boosting Classifiers with Tightened L0-Relaxation Penalties
We propose a novel boosting algorithm which improves on current algorithms for weighted voting classification by striking a better balance between classification accuracy and the ...
Noam Goldberg, Jonathan Eckstein