ISBN-10:
3642039987
ISBN-13:
9783642039980
Pub. Date:
09/30/2009
Publisher:
Springer Berlin Heidelberg
Applications of Supervised and Unsupervised Ensemble Methods / Edition 1

Applications of Supervised and Unsupervised Ensemble Methods / Edition 1

by Oleg Okun

Hardcover

Current price is , Original price is $179.99. You
Select a Purchase Option (2010)
  • purchase options
    $121.33 $179.99 Save 33% Current price is $121.33, Original price is $179.99. You Save 33%.
  • purchase options

Product Details

ISBN-13: 9783642039980
Publisher: Springer Berlin Heidelberg
Publication date: 09/30/2009
Series: Studies in Computational Intelligence , #245
Edition description: 2010
Pages: 268
Product dimensions: 6.20(w) x 9.40(h) x 0.90(d)

Table of Contents

An Ensemble Pruning Primer Grigorios Tsoumakas Ioannis Partalas Ioannis Vlahavas 1

1 Introduction 1

2 Background 2

2.1 Producing the Models 2

2.2 Combining the Models 3

3 A Taxonomy of Ensemble Pruning Methods 4

4 Ranking-Based Methods 4

5 Clustering-Based Methods 5

6 Optimization-Based Methods 6

6.1 Genetic Algorithms 6

6.2 Semi-definite Programming 6

6.3 Hill Climbing 7

7 Other Methods 10

7.1 Statistical Procedures 10

7.2 Reinforcement Learning 10

7.3 Boosting 11

8 Conclusions 11

References 12

Evade Hard Multiple Classifier Systems Battista Biggio Giorgio Fumera Fabio Roli 15

1 Introduction 15

2 Related Work 17

2.1 Previous Works on Multiple Classifiers for Security Applications 17

2.2 A Theoretical Framework for Adversarial Classification Problems 19

3 Are Multiple Classifier Systems Harder to Evade? 21

3.1 Adding Features to a Classification System 22

3.2 Splitting Features across an Ensemble of Classifiers 24

4 A Case Study in Spam Filtering 28

4.1 Adding Features to a Spam Filter 29

4.2 Splitting the Features of a Spam Filter across an Ensemble of Classifiers 32

5 Conclusions 37

References 37

A Personal Antispam System Based on a Behaviour-Knowledge Space Approach Francesco Gargiulo Antonio Penta Antonio Picarieilo Antonio Picariello Carlo Sansone 39

1 Introduction 39

2 Related Work 40

3 System Architecture 42

4 Textual Features 43

4.1 Semantic Features 43

4.2 Syntactic Features 46

5 Image Features 47

5.1 Visual Features 47

5.2 OCR-Based Features 49

6 Combining Text-Based and Image-Based Classifiers 50

7 Experimental Results 52

8 Conclusion 55

References 56

Weighted Decoding ECOC for FacialAction Unit Classification Terry Windeatt 59

1 Introduction 59

2 Ensembles and Bootstrapping 61

3 Error-Correcting Output Coding ECOC 63

3.1 Motivation 64

3.2 ECOC Algorithm and OOB Estimate 65

3.3 Coding Strategies and Errors 66

3.4 Weighted Decoding 68

4 Dataset and Feature Extraction 69

5 Experiments on Cohn-Kanade Database 71

6 Discussion 74

7 Conclusion 75

References 75

Prediction of Gene Function Using Ensembles of SVMs and Heterogeneous Data Sources Matteo Re Giorgio Valentini 79

1 Introduction 79

2 Methods 80

2.1 Linear Weighted Combination with Linear and Logarithmic Weights 81

2.2 Decision Templates 82

3 Experimental Setup 83

3.1 Heterogeneous Biomolecular Datasets 83

3.2 The Functional Catalogue (FunCat) 85

3.3 Base Learners Tuning and Generation of Optimized Classifiers 86

4 Results 86

5 Discussion 88

6 Conclusions 89

References 90

Partitioner Trees for Classification: A New Ensemble Method Georg Krempl Vera Hofer 93

1 Introduction 93

2 Partitioner Trees 95

2.1 Algorithm 95

2.2 Discussion 99

3 Experiments 104

3.1 Experimental Setup 104

3.2 Results 105

3.3 Conclusion 111

References 112

Disturbing Neighbors Diversity for Decision Forests Jesús Maudes Juan J. Rodríguez César García-Osorio 113

1 Introduction 113

2 Method 115

3 Results 119

4 Lesion Study 124

5 Conclusion 131

References l32

Improving Supervised Learning with Multiple Clusterings Cédric Wemmert Germain Forestier Sébastien Derivaux 135

1 Introduction 135

2 Related Works 136

3 Description of the Method 138

3.1 Improving Supervised Classification with Clustering 138

3.2 The Proposed Method 140

4 Experiments 142

4.1 Artificial Benchmark Evaluation 142

4.2 Real Data Evaluation 144

5 Conclusion 148

References 148

The Neighbors Voting Algorithm and Its Applications Gabriele Lombardi Elena Casiraghi Paola Campadelli 151

1 Introduction 151

2 The Tensor Voting Framework 152

3 The Neighbors Voting Algorithm 155

3.1 Statistical Interpretation 156

4 Applications of the Neighbors Voting Algorithm 157

4.1 Point Clustering 157

4.2 Classification 158

4.3 Image Inpainting with NV 160

5 Results 161

5.1 Clustering Results 161

5.2 Classification Results 165

5.3 Inpainting Results 170

6 Conclusions and Future Work 170

References 172

Clustering Ensembles with Active Constraints Muna Al-Razgan Carlotta Domeniconi 175

1 Introduction 175

2 Related Work 176

3 Locally Adaptive Clustering 177

4 Selecting Informative Constraints 179

5 Chunklet Graph 181

6 Chunklet Assignment 182

7 Constrained-Weighted Bipartite Partitioning Algorithm (C-WBPA) 182

8 Empirical Evaluation 185

8.1 Analysis of the Results 188

9 Conclusions 188

References 189

Verifiable Ensembles of Low-Dimensional Submodels for Multi-class Problems with Imbalanced Misclassification Costs Sebastian Nusser Clemens Otte Werner Hauptmann 191

1 Introduction 191

2 The Binary Ensemble Framework 193

2.1 Decision Tree-Like Ensemble Model 194

2.2 Non-Hierarchical Ensemble Model 194

2.3 An Illustrative Example 195

3 Multi-class Extensions of Binary Classifiers 197

4 The Multi-class Ensemble Framework 199

4.1 Ensemble of Multi-class Submodels 199

4.2 Hierarchical Separate-and-Conquer Ensemble 200

4.3 One-versus-Rest Ensemble 201

4.4 An Illustrative Example (Cont'd) 202

5 Experiments 203

5.1 Binary Classification Problems 205

5.2 Multi-class Classification Problems 206

5.3 Comparison of the Ensemble Methods 206

5.4 Comparison of Different Feature Selection Methods 208

6 Conclusions 210

References 210

Independent Data Model Selection for Ensemble Dispersion Forecasting Angelo Ciaramella Giulio Giunta Angela Riccio Stefano Galmarini 213

1 Introduction 213

2 The 'Median Model' Approach 215

3 Negentropy-Based Hierarchical Agglomeration 218

3.1 Kullback-Leibler Divergence 218

3.2 Negentropy Information 218

3.3 Agglomerative Approach 219

4 Experimental Results 219

5 Conclusions 229

References 229

Integrating Liknon Feature Selection and Committee Training Erinija Pranckeviciene 233

1 Introduction 233

2 Computational Paradigm and Parameters 234

2.1 Liknon-Based Feature Selection 235

2.2 Banana Example: Classification in Nonlinear Class Separation 238

3 The Benchmark of NIPS2003 Feature Selection Challenge 240

3.1 Perforinance, Size and Purity of the Feature Subsets in the Benchmark 240

4 NN3 Committee and Liknon Feature Profiles in the Benchmark 242

5 Conclusion 248

References 248

Evaluating Hybrid Ensembles for Intelligent Decision Support for Intensive Care Pedro Gago Manuel Filipe Santos 251

1 Introduction 251

2 Previous Work 252

3 INTCare System 254

3.1 Description of the Agents 254

4 Problem Description 256

4.1 Data Description 257

5 Related Work 257

6 Experimental Setting 258

7 Results 260

8 Discussion 262

9 Conclusion 263

References 263

Index 267

Customer Reviews

Most Helpful Customer Reviews

See All Customer Reviews