Graphics Reference
In-Depth Information
Table 8.4
Hybridizations with other learning approaches and ensembles
Description
Reference
First approach for nested generalized examples learning (hyperrectangle
learning): EACH
[
138
]
Experimental review on nested generalized examples learning
[
162
]
Unification of rule induction with instance-based learning: RISE
[
50
]
Condensed nearest neighbour (CNN) ensembles
[
5
]
Inflating instances to obtain rules: INNER
[
114
]
Bagging for lazy learning
[
174
]
Evolutionary ensembles for classifiers selection
[
142
]
Ensembles for weighted IS
[
71
]
Boostrapping for KNN
[
148
]
Evolutionary optimization in hyperrectangles learning
[
67
]
Evolutionary optimization in hyperrectangles learning for imbalanced
problems
[
69
]
Review of ensembles for data preprocessing in imbalanced problems
[
60
]
Boosting by warping of the distance metric for KNN
[
121
]
Evolutionary undersampling based on ensembles for imbalanced prob-
lems
[
61
]
Table 8.5
Scaling-up and distributed approaches
Description
Reference
Recursive subdivision of prototype reduction methods for tackling large data sets
[
91
]
Stratified division of training data sets to improve the scaling-up of PS methods
[
20
]
Usage of KD-trees for prototype reduction schemes
[
120
]
Distributed condensation for large data sets
[
6
]
Divide-and-conquer recursive division of training data for speed-up IS
[
81
]
Division of data based of ensembles with democratic voting for IS
[
70
]
Usage of stratification for scaling-up evolutionary algorithms for IS
[
41
]
Distributed implementation of the stratification process combined with k-means for
IS
[
33
]
Scalable divide-and-conquer based on bookkeeping for instance and feature selection [
74
]
Scaling-up IS based on the parallelization of small subsets of data
[
82
]
Table
8.6
collects the developments in data complexity related with IS found in
the specialized literature.
8.6 Experimental Comparative Analysis in Prototype Selection
The aim of this section is to show all the factors and issues related to the experimental
study. We specify the data sets, validation procedure, parameters of the algorithms,
performance metrics and PS methods involved in the analysis. The experimental
