Multilabel Classification : Problem Analysis, Metrics and Techniques
This book offers a comprehensive review of multilabel techniques widely used to classify and label texts, pictures, videos and music in the Internet. A deep review of the specialized literature on the field includes the available software needed to work with this kind of data. It provides the user w...
Saved in:
| Main Authors | , , , |
|---|---|
| Format | eBook Book |
| Language | English |
| Published |
Cham
Springer
2016
Springer International Publishing AG Springer International Publishing |
| Edition | 1 |
| Subjects | |
| Online Access | Get full text |
| ISBN | 9783319411101 3319411101 |
| DOI | 10.1007/978-3-319-41111-8 |
Cover
Table of Contents:
- 4.3 Binary Classification Based Methods -- 4.3.1 OVO Versus OVA Approaches -- 4.3.2 Ensembles of Binary Classifiers -- 4.4 Multiclass Classification-Based Methods -- 4.4.1 Labelsets and Pruned Labesets -- 4.4.2 Ensembles of Multiclass Classifiers -- 4.5 Data Transformation Methods in Practice -- 4.5.1 Experimental Configuration -- 4.5.2 Classification Results -- 4.6 Summarizing Comments -- References -- 5 Adaptation-Based Classifiers -- 5.1 Overview -- 5.2 Tree-Based Methods -- 5.2.1 Multilabel C4.5, ML-C4.5 -- 5.2.2 Multilabel Alternate Decision Trees, ADTBoost.MH -- 5.2.3 Other Tree-Based Proposals -- 5.3 Neuronal Network-Based Methods -- 5.3.1 Multilabel Back-Propagation, BP-MLL -- 5.3.2 Multilabel Radial Basis Function Network, ML-RBF -- 5.3.3 Canonical Correlation Analysis and Extreme Learning Machine, CCA-ELM -- 5.4 Vector Support Machine-Based Methods -- 5.4.1 MODEL-x -- 5.4.2 Multilabel SVMs Based on Ranking, Rank-SVM and SCRank-SVM -- 5.5 Instance-Based Methods -- 5.5.1 Multilabel kNN, ML-kNN -- 5.5.2 Instance-Based and Logistic Regression, IBLR-ML -- 5.5.3 Other Instance-Based Classifiers -- 5.6 Probabilistic Methods -- 5.6.1 Collectible Multilabel Classifiers, CML and CMLF -- 5.6.2 Probabilistic Generic Models, PMM1 and PMM2 -- 5.6.3 Probabilistic Classifier Chains, PCC -- 5.6.4 Bayesian and Tree Naïve Bayes Classifier Chains, BCC and TNBCC -- 5.6.5 Conditional Restricted Boltzmann Machines, CRBM -- 5.7 Other MLC Adaptation-Based Methods -- 5.8 Adapted Methods in Practice -- 5.8.1 Experimental Configuration -- 5.8.2 Classification Results -- 5.9 Summarizing Comments -- References -- 6 Ensemble-Based Classifiers -- 6.1 Introduction -- 6.2 Ensembles of Binary Classifiers -- 6.2.1 Ensemble of Classifier Chains, ECC -- 6.2.2 Ranking by Pairwise Comparison, RPC -- 6.2.3 Calibrated Label Ranking, CLR -- 6.3 Ensembles of Multiclass Classifiers
- 6.3.1 Ensemble of Pruned Sets, EPS -- 6.3.2 Random k-Labelsets, RAkEL -- 6.3.3 Hierarchy of Multilabel Classifiers, HOMER -- 6.4 Other Ensembles -- 6.5 Ensemble Methods in Practice -- 6.5.1 Experimental Configuration -- 6.5.2 Classification Results -- 6.5.3 Training and Testing Times -- 6.6 Summarizing Comments -- References -- 7 Dimensionality Reduction -- 7.1 Overview -- 7.1.1 High-Dimensional Input Space -- 7.1.2 High-Dimensional Output Space -- 7.2 Feature Space Reduction -- 7.2.1 Feature Engineering Approaches -- 7.2.2 Multilabel Supervised Feature Selection -- 7.2.3 Experimentation -- 7.3 Label Space Reduction -- 7.3.1 Sparseness and Dependencies Among Labels -- 7.3.2 Proposals for Reducing Label Space Dimensionality -- 7.3.3 Experimentation -- 7.4 Summarizing Comments -- References -- 8 Imbalance in Multilabel Datasets -- 8.1 Introduction -- 8.2 Imbalanced MLD Specificities -- 8.2.1 How to Measure the Imbalance Level -- 8.2.2 Concurrence Among Imbalanced Labels -- 8.3 Facing Imbalanced Multilabel Classification -- 8.3.1 Classifier Adaptation -- 8.3.2 Resampling Techniques -- 8.3.3 The Ensemble Approach -- 8.4 Multilabel Imbalanced Learning in Practice -- 8.4.1 Experimental Configuration -- 8.4.2 Classification Results -- 8.5 Summarizing Comments -- References -- 9 Multilabel Software -- 9.1 Overview -- 9.2 Working with Multilabel Data -- 9.2.1 Multilabel Data File Formats -- 9.2.2 Multilabel Data Repositories -- 9.2.3 The mldr.datasets Package -- 9.2.4 Generating Synthetic MLDs -- 9.3 Exploratory Analysis of MLDs -- 9.3.1 MEKA -- 9.3.2 The mldr Package -- 9.4 Conducting Multilabel Experiments -- 9.4.1 MEKA -- 9.4.2 MULAN -- 9.4.3 The RunMLClassifier Utility -- 9.5 Summarizing Comments -- References -- Glossary
- Intro -- Preface -- Contents -- Acronyms -- 1 Introduction -- 1.1 Overview -- 1.2 The Knowledge Discovery in Databases Process -- 1.3 Data Preprocessing -- 1.4 Data Mining -- 1.4.1 DM Methods Attending to Available Data -- 1.4.2 DM Methods Attending to Target Objective -- 1.4.3 DM Methods Attending to Knowledge Representation -- 1.5 Classification -- 1.5.1 Binary Classification -- 1.5.2 Multiclass Classification -- 1.5.3 Multilabel Classification -- 1.5.4 Multidimensional Classification -- 1.5.5 Multiple Instance Learning -- References -- 2 Multilabel Classification -- 2.1 Introduction -- 2.2 Problem Formal Definition -- 2.2.1 Definitions -- 2.2.2 Symbols -- 2.2.3 Terminology -- 2.3 Applications of Multilabel Classification -- 2.3.1 Text Categorization -- 2.3.2 Labeling of Multimedia Resources -- 2.3.3 Genetics/Biology -- 2.3.4 Other Application Fields -- 2.3.5 MLDs Repositories -- 2.4 Learning from Multilabel Data -- 2.4.1 The Data Transformation Approach -- 2.4.2 The Method Adaptation Approach -- 2.4.3 Ensembles of Classifiers -- 2.4.4 Label Correlation Information -- 2.4.5 High Dimensionality -- 2.4.6 Label Imbalance -- 2.5 Multilabel Data Tools -- References -- 3 Case Studies and Metrics -- 3.1 Overview -- 3.2 Case Studies -- 3.2.1 Text Categorization -- 3.2.2 Labeling of Multimedia Resources -- 3.2.3 Genetics/Biology -- 3.2.4 Synthetic MLDs -- 3.3 MLD Characteristics -- 3.3.1 Basic Metrics -- 3.3.2 Imbalance Metrics -- 3.3.3 Other Metrics -- 3.3.4 Summary of Characterization Metrics -- 3.4 Multilabel Classification by Example -- 3.4.1 The ML-kNN Algorithm -- 3.4.2 Experimental Configuration and Results -- 3.5 Assessing Classifiers Performance -- 3.5.1 Example-Based Metrics -- 3.5.2 Label-based Metrics -- References -- 4 Transformation-Based Classifiers -- 4.1 Introduction -- 4.2 Multilabel Data Transformation Approaches