| Preface | p. xi |
| Authors | p. xiii |
| Symbol Description | p. xv |
| Data of High Dimensionality and Challenges | p. 1 |
| Dimensionality Reduction Techniques | p. 3 |
| Feature Selection for Data Mining | p. 8 |
| A General Formulation for Feature Selection | p. 8 |
| Feature Selection in a Learning Process | p. 9 |
| Categories of Feature Selection Algorithms | p. 10 |
| Degrees of Supervision | p. 10 |
| Relevance Evaluation Strategies | p. 11 |
| Output Formats | p. 12 |
| Number of Data Sources | p. 12 |
| Computation Schemes | p. 13 |
| Challenges in Feature Selection Research | p. 13 |
| Redundant Features | p. 14 |
| Large-Scale Data | p. 14 |
| Structured Data | p. 14 |
| Data of Small Sample Size | p. 15 |
| Spectral Feature Selection | p. 15 |
| Organization of the Book | p. 17 |
| Univariate Formulations for Spectral Feature Selection | p. 21 |
| Modeling Target Concept via Similarity Matrix | p. 21 |
| The Laplacian Matrix of a Graph | p. 23 |
| Evaluating Features on the Graph | p. 29 |
| An Extension for Feature Ranking Functions | p. 36 |
| Spectral Feature Selection via Ranking | p. 40 |
| SPEC for Unsupervised Learning | p. 41 |
| SPEC for Supervised Learning | p. 42 |
| SPEC for Semi-Supervised Learning | p. 42 |
| Time Complexity of SPEC | p. 44 |
| Robustness Analysis for SPEC | p. 45 |
| Discussions | p. 54 |
| Multivariate Formulations | p. 55 |
| The Similarity Preserving Nature of SPEC | p. 56 |
| A Sparse Multi-Output Regression Formulation | p. 61 |
| Solving the L2,1-Regularized Regression Problem | p. 66 |
| The Coordinate Gradient Descent Method (CGD) | p. 69 |
| The Accelerated Gradient Descent Method (AGD) | p. 70 |
| Efficient Multivariate Spectral Feature Selection | p. 71 |
| A Formulation Based on Matrix Comparison | p. 80 |
| Feature Selection with Proposed Formulations | p. 82 |
| Connections to Existing Algorithms | p. 83 |
| Connections to Existing Feature Selection Algorithms | p. 83 |
| Laplacian Score | p. 84 |
| Fisher Score | p. 85 |
| Relief and ReliefF | p. 86 |
| Trace Ratio Criterion | p. 87 |
| Hilbert-Schmidt Independence Criterion (HSIC) | p. 89 |
| A Summary of the Equivalence Relationships | p. 89 |
| Connections to Other Learning Models | p. 91 |
| Linear Discriminant Analysis | p. 91 |
| Least Square Support Vector Machine | p. 95 |
| Principal Component Analysis | p. 97 |
| Simultaneous Feature Selection and Extraction | p. 99 |
| An Experimental Study of the Algorithms | p. 99 |
| A Study of the Supervised Case | p. 101 |
| Accuracy | p. 101 |
| Redundancy Rate | p. 101 |
| A Study of the Unsupervised Case | p. 104 |
| Residue Scale and Jaccard Score | p. 104 |
| Redundancy Rate | p. 105 |
| Discussions | p. 106 |
| Large-Scale Spectral Feature Selection | p. 109 |
| Data Partitioning for Parallel Processing | p. 111 |
| MPI for Distributed Parallel Computing | p. 113 |
| MPI_BCAST | p. 114 |
| MPI_SCATTER | p. 115 |
| MPI_REDUCE | p. 117 |
| Parallel Spectral Feature Selection | p. 118 |
| Computation Steps of Univariate Formulations | p. 119 |
| Computation Steps of Multivariate Formulations | p. 120 |
| Computing the Similarity Matrix in Parallel | p. 121 |
| Computing the Sample Similarity | p. 121 |
| Inducing Sparsity | p. 122 |
| Enforcing Symmetry | p. 122 |
| Parallelization of the Univariate Formulations | p. 124 |
| Parallel MRSF | p. 128 |
| Initializing the Active Set | p. 130 |
| Computing the Tentative Solution | p. 131 |
| Computing the Walking Direction | p. 131 |
| Calculating the Step Size | p. 132 |
| Constructing the Tentative Solution | p. 133 |
| Time Complexity for Computing a Tentative Solution | p. 134 |
| Computing the Optimal Solution | p. 134 |
| Checking the Global Optimality | p. 137 |
| Summary | p. 137 |
| Parallel MCSF | p. 139 |
| Discussions | p. 141 |
| Multi-Source Spectral Feature Selection | p. 143 |
| Categorization of Different Types of Knowledge | p. 145 |
| A Framework Based on Combining Similarity Matrices | p. 148 |
| Knowledge Conversion | p. 150 |
| KSIMFEA→KSIMSAM | p. 151 |
| KFUNFEA, KINTFEA → KSIMSAM | p. 152 |
| MSFS: The Framework | p. 153 |
| A Framework Based on Rank Aggregation | p. 153 |
| Handling Knowledge in KOFS | p. 155 |
| Internal Knowledge | p. 155 |
| Knowledge Conversion | p. 156 |
| Ranking Using Internal Knowledge | p. 157 |
| Relevance Propagation with KRELint, FEA | p. 157 |
| Relevance Voting with KFUNint, FEA | p. 157 |
| Aggregating Feature Ranking Lists | p. 158 |
| An EM Algorithm for Computing | p. 159 |
| Experimental Results | p. 160 |
| Data and Knowledge Sources | p. 160 |
| Pediatric ALL Data | p. 160 |
| Knowledge Sources | p. 160 |
| Experiment Setup | p. 161 |
| Performance Evaluation | p. 162 |
| Empirical Findings | p. 164 |
| Discussion of Biological Relevance | p. 166 |
| Discussions | p. 167 |
| References | p. 171 |
| Index | p. 191 |
| Table of Contents provided by Ingram. All Rights Reserved. |
