Semisupervised Learning for Computational

The rapid advancement in the theoretical understanding of statistical and machine learning methods for semisupervised learning has made it difficult for nonspecialists to keep up to date in the field. Providing a broad, accessible treatment of the theory as well as linguistic applications, Semisupervised Learning for Computational Linguistics offers self-contained coverage of semisupervised methods that includes background material on supervised and unsupervised learning.The book presents a brief history of semisupervised learning and its place in the spectrum of learning methods before moving on to discuss well-known natural language processing methods, such as self-training and co-training. It then centers on machine learning techniques, including the boundary-oriented methods of perceptrons, boosting, support vector machines (SVMs), and the null-category noise model. In addition, the book covers clustering, the expectation-maximization (EM) algorithm, related generative methods, and agreement methods. It concludes with the graph-based method of label propagation as well as a detailed discussion of spectral methods.Taking an intuitive approach to the material, this lucid book facilitates the application of semisupervised learning methods to natural language processing and provides the framework and motivation for a more systematic study of machine learning.

Introduction	p. 1
A brief history	p. 1
Probabilistic methods in computational linguistics	p. 1
Supervised and unsupervised training	p. 2
Semisupervised learning	p. 3
Semisupervised learning	p. 4
Major varieties of learning problem	p. 4
Motivation	p. 6
Evaluation	p. 7
Active learning	p. 8
Organization and assumptions	p. 8
Leading ideas	p. 8
Mathematical background	p. 10
Notation	p. 11
Self-training and Co-training	p. 13
Classification	p. 13
The standard setting	p. 13
Features and rules	p. 14
Decision lists	p. 16
Self-training	p. 18
The algorithm	p. 19
Parameters and variants	p. 20
Evaluation	p. 23
Symmetry of features and instances	p. 25
Related algorithms	p. 27
Co-Training	p. 28
Applications of Self-Training and Co-Training	p. 31
Part-of-speech tagging	p. 31
Information extraction	p. 33
Parsing	p. 35
Word senses	p. 36
WordNet	p. 36
Word-sense disambiguation	p. 38
Taxonomic inference	p. 40
Classification	p. 43
Two simple classifiers	p. 43
Naive Bayes	p. 43
k-nearest-neighbor classifier	p. 45
Abstract setting	p. 48
Function approximation	p. 48
Defining succes	p. 50
Fit and simplicity	p. 52
Evaluating detectors and classifiers that abstain	p. 53
Confidence-rated classifiers	p. 53
Measures for detection	p. 54
Idealized performance curves	p. 57
The multiclass case	p. 59
Binary classifiers and ECOC	p. 62
Mathematics for Boundary-Oriented Methods	p. 67
Linear separators	p. 67
Representing a hyperplane	p. 67
Eliminating the threshold	p. 69
The point-normal form	p. 70
Naive Bayes decision boundary	p. 72
The gradient	p. 74
Graphs and domains	p. 74
Convexity	p. 76
Differentiation of vector and matrix expressions	p. 79
An example: linear regression	p. 81
Constrained optimization	p. 83
Optimization	p. 83
Equality constraints	p. 84
Inequality constraints	p. 87
The Wolfe dual	p. 91
Boundary-Oriented Methods	p. 95
The perceptron	p. 97
The algorithm	p. 97
An example	p. 99
Convergence	p. 100
The perceptron algorithm as gradient descent	p. 101
Game self-teaching	p. 103
Boosting	p. 105
Abstention	p. 110
Semisupervised boosting	p. 111
Co-boosting	p. 113
Support Vector Machines (SVMs)	p. 114
The margin	p. 114
Maximizing the margin	p. 116
The nonseparable case	p. 119
Slack in the separable case	p. 121
Multiple slack points	p. 123
Transductive SVMs	p. 125
Training a transductive SVM	p. 127
Null-category noise model	p. 129
Clustering	p. 131
Cluster and label	p. 131
Clustering concepts	p. 132
Objective	p. 132
Distance and similarity	p. 133
Graphs	p. 136
Hierarchical clustering	p. 137
Self-training revisited	p. 139
k-means clustering	p. 139
Pseudo relevance feedback	p. 140
Graph mincut	p. 143
Label propagation	p. 146
Clustering by propagation	p. 146
Self-training as propagation	p. 147
Co-training as propagation	p. 150
Bibliographic notes	p. 152
Generative Models	p. 153
Gaussian mixtures	p. 153
Definition and geometric interpretation	p. 153
The linear discriminant decision boundary	p. 156
Decision-directed approximation	p. 159
McLachlan's algorithm	p. 162
The EM algorithm	p. 163
Maximizing likelihood	p. 163
Relative frequency estimation	p. 164
Divergence	p. 166
The EM algorithm	p. 169
Agreement Constraints	p. 175
Co-training	p. 175
The conditional independence assumption	p. 176
The power of conditional independence	p. 178
Agreement-based self-teaching	p. 182
Random fields	p. 184
Applied to self-training and co-training	p. 184
Gibbs sampling	p. 186
Markov chains and random walks	p. 187
Bibliographic notes	p. 192
Propagation Methods	p. 193
Label propagation	p. 194
Random walks	p. 196
Harmonic functions	p. 198
Fluids	p. 203
Flow	p. 203
Pressure	p. 205
Conservation of energy	p. 209
Thomson's principle	p. 210
Computing the solution	p. 213
Graph mincuts revisited	p. 215
Bibliographic notes	p. 220
Mathematics for Spectral Methods	p. 221
Some basic concepts	p. 221
The norm of a vector	p. 221
Matrices as linear operators	p. 222
The column space	p. 222
Eigenvalues and eigenvectors	p. 224
Definition of eigenvalues and eigenvectors	p. 224
Diagonalization	p. 225
Orthogonal diagonalization	p. 226
Eigenvalues and the scaling effects of a matrix	p. 227
Matrix norms	p. 227
The Rayleigh quotient	p. 228
The 2 x 2 case	p. 230
The general case	p. 232
The Courant-Fischer minimax theorem	p. 234
Bibliographic notes	p. 236
Spectral Methods	p. 237
Simple harmonic motion	p. 237
Harmonics	p. 237
Mixtures of harmonics	p. 239
An oscillating particle	p. 241
A vibrating string	p. 243
Spectra of matrices and graphs	p. 251
The spectrum of a matrix	p. 252
Relating matrices and graphs	p. 253
The Laplacian matrix and graph spectrum	p. 256
Spectral clustering	p. 257
The second smallest eigenvector of the Laplacian	p. 257
The cut size and the Laplacian	p. 259
Approximating cut size	p. 260
Minimizing cut size	p. 262
Ratiocut	p. 263
Spectral methods for semisupervised learning	p. 265
Harmonics and harmonic functions	p. 265
Eigenvalues and energy	p. 267
The Laplacian and random fields	p. 268
Harmonic functions and the Laplacian	p. 270
Using the Laplacian for regularization	p. 272
Transduction to induction	p. 274
Bibliographic notes	p. 275
Bibliography	p. 277
Index	p. 301
Table of Contents provided by Ingram. All Rights Reserved.

What is included with this book?

The New copy of this book will include any supplemental materials advertised. Please check the title of the book to determine if it should include any access cards, study guides, lab manuals, CDs, etc.

The Used, Rental and eBook copies of this book are not guaranteed to include any supplemental materials. Typically, only the book itself is included. This is true even if the title states it includes any access cards, study guides, lab manuals, CDs, etc.

FREE SHIPPING

Amazon no longer offers textbook rentals. We do!

Semisupervised Learning for Computational Linguistics

Summary

Table of Contents

Supplemental Materials