Semi-supervised learning : background, applications and future directions /

Detalles Bibliográficos
Clasificación:Libro Electrónico
Otros Autores: Zhong, Guoqiang (Editor ), Huang, Kaizhu (Editor )
Formato: Electrónico eBook
Idioma:Inglés
Publicado: New York : Nova Science Publishers, [2018]
Colección:Education in a competitive and globalizing world series.
Temas:
Supervised learning (Machine learning)
Apprentissage supervisé (Intelligence artificielle)
COMPUTERS > General.
Acceso en línea:Texto completo
Tabla de Contenidos:
  • Intro; SEMI-SUPERVISED LEARNINGBACKGROUND, APPLICATIONSAND FUTURE DIRECTIONS; SEMI-SUPERVISED LEARNINGBACKGROUND, APPLICATIONSAND FUTURE DIRECTIONS; CONTENTS; PREFACE; Introduction to This Book; Target Audience; Acknowledgments; Chapter 1CONSTRAINED DATASELF-REPRESENTATIVE GRAPHCONSTRUCTION; Abstract; 1. Introduction; 2. Constrained Data Self-Representative GraphConstruction; 3. Kernelized Variants; 3.1. Hilbert Space; 3.2. Column Generation; 4. Performance Evaluation; 4.1. Label Propagation; 4.1.1. Gaussian Random Fields; 4.1.2. Local and Global Consistency; 4.2. Experimental Results
  • 4.2.1. Comparison among Several Graph Construction Methods4.2.2. Stability of the Proposed Method; 4.2.3. Sensitivity to Parameters; 4.2.4. Computational Complexity and CPU Time; Acknowledgments; Conclusion; References; Chapter 2INJECTING RANDOMNESS INTO GRAPHS:AN ENSEMBLE SEMI-SUPERVISEDLEARNING FRAMEWORK; Abstract; 1. Introduction; 2. Background; 2.1. Graph-Based Semi-Supervised Learning; 2.2. Ensemble Learning and Random Forests; 2.3. Anchor Graph; 3. Random Multi-Graphs; 3.1. Problem Formulation; 3.2. Algorithm; 3.3. Graph Construction; 3.4. Semi-Supervised Inference
  • 3.5. Inductive Extension3.6. Randomness as Regularization; 4. Experiments; 4.1. Data Sets; 4.2. Experimental Results; 4.3. Impact of Parameters; 4.4. Hyperspectral Image Classification; Acknowledgments; Conclusion; References; Chapter 3LABEL PROPAGATION VIA KERNELFLEXIBLE MANIFOLD EMBEDDING; Abstract; 1. Introduction; 2. RelatedWork; 2.1. Semi-Supervised Discriminant Analysis; 2.2. Semi-Supervised Discriminant Embedding; 2.3. Laplacian Regularized Least Square; 2.4. Review of the Flexible Manifold Embedding Framework; 3. Kernel FlexibleManifold Embedding; 3.1. The Objective Function
  • 3.2. Optimal Solution3.3. The Algorithm; 3.4. Difference between KFME and Existing Methods; 3.4.1. Difference between KFME and FME; 3.4.2. Difference between KFME and Other Methods; 4. Experimental Results; 4.1. Datasets; 4.2. Method Comparison; 4.3. Results Analysis; 4.4. Stability with Respect to Graph; Acknowledgments; Conclusion; References; Chapter 4FAST GRAPH-BASED SEMI-SUPERVISEDLEARNING AND ITS APPLICATIONS; Abstract; 1. Introduction; 2. Related Work; 2.1. Scalable Graph-Based SSL/TL Methods; 2.2. Scalable Graph Construction Methods; 2.3. Robust Graph-Based SSL/TL Methods
  • 3. Minimum Tree Cut Method3.1. Notations; 3.2. The Proposed Method; 3.3. The Tree Labeling Algorithm; 3.4. Generate a Spanning Tree from a Graph; 4. Insensitiveness to Graph Construction; 5. Experiments; 5.1. Data Set; 5.1.1. UCI Data Set; 5.1.2. Image; 5.1.3. Text; 5.2. Graph Construction; 5.3. Accuracy; 5.4. Speed; 5.5. Robustness; 5.6. Effect of Different Spanning Tree and Ensemble of MultipleSpanning Trees; 6. Applications in Text Extraction; 6.1. Interactive Text Extraction in Natural Scene Images; 6.2. Document Image Binarization; Conclusion and FutureWork; References

Ejemplares similares