1473 search results - page 126 / 295 » Permutation Tests for Classification |
TNN
2008
14 years 9 months ago
2008
Abstract--The stationarity requirement for the process generating the data is a common assumption in classifiers' design. When such hypothesis does not hold, e.g., in applicat...
ICML
2001
IEEE
15 years 10 months ago
2001
IEEE
We present an improved bound on the difference between training and test errors for voting classifiers. This improved averaging bound provides a theoretical justification for popu...
FLAIRS
2010
15 years 8 days ago
2010
We propose a generic framework that uses utility in decision making to drive the data mining process. We use concepts from meta-learning and build on earlier work by Elovici and B...
COLING
2008
14 years 11 months ago
2008
In this paper we propose a new distance function (rank distance) designed to reflect stylistic similarity between texts. To assess the ability of this distance measure to capture ...
ICPR
2006
IEEE
15 years 11 months ago
2006
IEEE
We propose a competitive finite mixture of neurons (or perceptrons) for solving binary classification problems. Our classifier includes a prior for the weights between different n...