Sciweavers

4636 search results - page 94 / 928
» Modelling and Validation with VipTool
Sort
View
ICML
2002
IEEE
14 years 8 months ago
Is Combining Classifiers Better than Selecting the Best One
We empirically evaluate several state-of-theart methods for constructing ensembles of heterogeneous classifiers with stacking and show that they perform (at best) comparably to se...
Saso Dzeroski, Bernard Zenko
FLAIRS
2009
13 years 5 months ago
Unit Testing for Qualitative Spatial and Temporal Reasoning
Researchers in commonsense, qualitative spatial and temporal reasoning (QSTR) provide flexible and intuitive methods for reasoning about vague and uncertain information including ...
Carl P. L. Schultz, Robert Amor, Hans W. Guesgen
NIPS
1998
13 years 9 months ago
Dynamically Adapting Kernels in Support Vector Machines
The kernel-parameter is one of the few tunable parameters in Support Vector machines, controlling the complexity of the resulting hypothesis. Its choice amounts to model selection...
Nello Cristianini, Colin Campbell, John Shawe-Tayl...
NCA
2007
IEEE
13 years 7 months ago
Handling of incomplete data sets using ICA and SOM in data mining
Based on independent component analysis (ICA) and self-organizing maps (SOM), this paper proposes an ISOM-DH model for the incomplete data’s handling in data mining. Under these ...
Hongyi Peng, Siming Zhu
TNN
2010
143views Management» more  TNN 2010»
13 years 2 months ago
Using unsupervised analysis to constrain generalization bounds for support vector classifiers
Abstract--A crucial issue in designing learning machines is to select the correct model parameters. When the number of available samples is small, theoretical sample-based generali...
Sergio Decherchi, Sandro Ridella, Rodolfo Zunino, ...