| Preface |
| Introduction / 1： |
| Applications and problems / 1.1： |
| Definitions and terminology / 1.2： |
| Cross-validation / 1.3： |
| Learning scenarios / 1.4： |
| Outline / 1.5： |
| The PAC Learning Framework / 2： |
| The PAC learning model / 2.1： |
| Guarantees for finite hypothesis sets - consistent case / 2.2： |
| Guarantees for finite hypothesis sets - inconsistent case / 2.3： |
| Generalities / 2.4： |
| Deterministic versus stochastic scenarios / 2.4.1： |
| Bayes error and noise / 2.4.2： |
| Estimation and approximation errors / 2.4.3： |
| Model selection / 2.4.4： |
| Chapter notes / 2.5： |
| Exercises / 2.6： |
| Rademacher Complexity and VC-Dimension / 3： |
| Rademacher complexity / 3.1： |
| Growth function / 3.2： |
| VC-dimension / 3.3： |
| Lower bounds / 3.4： |
| Support Vector Machines / 3.5： |
| Linear classification / 4.1： |
| SVMs - separable case / 4.2： |
| Primal optimization problem / 4.2.1： |
| Support vectors / 4.2.2： |
| Dual optimization problem / 4.2.3： |
| Leave-one-out analysis / 4.2.4： |
| SVMs - non-separable case / 4.3： |
| Margin theory / 4.3.1： |
| Kernel Methods / 4.5： |
| Positive definite symmetric kernels / 5.1： |
| Definitions / 5.2.1： |
| Reproducing kernel Hilbert space / 5.2.2： |
| Properties / 5.2.3： |
| Kernel-based algorithms / 5.3： |
| SVMs with PDS kernels / 5.3.1： |
| Representer theorem / 5.3.2： |
| Learning guarantees / 5.3.3： |
| Negative definite symmetric kernels / 5.4： |
| Sequence kernels / 5.5： |
| Weighted transducers / 5.5.1： |
| Rational kernels / 5.5.2： |
| Boosting / 5.6： |
| AdaBoost / 6.1： |
| Bound on the empirical error / 6.2.1： |
| Relationship with coordinate descent / 6.2.2： |
| Relationship with logistic regression / 6.2.3： |
| Standard use in practice / 6.2.4： |
| Theoretical results / 6.3： |
| VC-dimension-based analysis / 6.3.1： |
| Margin-based analysis / 6.3.2： |
| Margin maximization / 6.3.3： |
| Game-theoretic interpretation / 6.3.4： |
| Discussion / 6.4： |
| On-Line Learning / 6.5： |
| Prediction with expert advice / 7.1： |
| Mistake bounds and Halving algorithm / 7.2.1： |
| Weighted majority algorithm / 7.2.2： |
| Randomized weighted majority algorithm / 7.2.3： |
| Exponential weighted average algorithm / 7.2.4： |
| Perceptron algorithm / 7.3： |
| Winnow algorithm / 7.3.2： |
| On-line to batch conversion / 7.4： |
| Game-theoretic connection / 7.5： |
| Multi-Class Classification / 7.6： |
| Multi-class classification problem / 8.1： |
| Generalization bounds / 8.2： |
| Uncombined multi-class algorithms / 8.3： |
| Multi-class SVMs / 8.3.1： |
| Multi-class boosting algorithms / 8.3.2： |
| Decision trees / 8.3.3： |
| Aggregated multi-class algorithms / 8.4： |
| One-versus-all / 8.4.1： |
| One-versus-one / 8.4.2： |
| Error-correction codes / 8.4.3： |
| Structured prediction algorithms / 8.5： |
| Ranking / 8.6： |
| The problem of ranking / 9.1： |
| Generalization bound / 9.2： |
| Ranking with SVMs / 9.3： |
| RankBoost / 9.4： |
| Margin bound for ensemble methods in ranking / 9.4.1： |
| Bipartite ranking / 9.5： |
| Boosting in bipartite ranking / 9.5.1： |
| Area under the ROC curve / 9.5.2： |
| Preference-based setting / 9.6： |
| Second-stage ranking problem / 9.6.1： |
| Deterministic algorithm / 9.6.2： |
| Randomized algorithm / 9.6.3： |
| Extension to other loss functions / 9.6.4： |
| Regression / 9.7： |
| The problem of regression / 10.1： |
| Finite hypothesis sets / 10.2： |
| Rademacher complexity bounds / 10.2.2： |
| Pseudo-dimension bounds / 10.2.3： |
| Regression algorithms / 10.3： |
| Linear regression / 10.3.1： |
| Kernel ridge regression / 10.3.2： |
| Support vector regression / 10.3.3： |
| Lasso / 10.3.4： |
| Group norm regression algorithms / 10.3.5： |
| On-line regression algorithms / 10.3.6： |
| Algorithmic Stability / 10.4： |
| Stability-based generalization guarantee / 11.1： |
| Stability of kernel-based regularization algorithms / 11.3： |
| Application to regression algorithms: SVR and KRR / 11.3.1： |
| Application to classification algorithms: SVMs / 11.3.2： |
| Dimensionality Reduction / 11.3.3： |
| Principal Component Analysis / 12.1： |
| Kernel Principal Component Analysis (KPCA) / 12.2： |
| KPCA and manifold learning / 12.3： |
| Isomap / 12.3.1： |
| Laplacian eigenmaps / 12.3.2： |
| Locally linear embedding (LLE) / 12.3.3： |
| Johnson-Lindenstrauss lemma / 12.4： |
| Learning Automata and Languages / 12.5： |
| Finite automata / 13.1： |
| Efficient exact learning / 13.3： |
| Passive learning / 13.3.1： |
| Learning with queries / 13.3.2： |
| Learning automata with queries / 13.3.3： |
| Identification in the limit / 13.4： |
| Learning reversible automata / 13.4.1： |
| Reinforcement Learning / 13.5： |
| Learning scenario / 14.1： |
| Markov decision process model / 14.2： |
| Policy / 14.3： |
| Definition / 14.3.1： |
| Policy value / 14.3.2： |
| Policy evaluation / 14.3.3： |
| Optimal policy / 14.3.4： |
| Planning algorithms / 14.4： |
| Value iteration / 14.4.1： |
| Policy iteration / 14.4.2： |
| Linear programming / 14.4.3： |
| Learning algorithms / 14.5： |
| Stochastic approximation / 14.5.1： |
| TD(0) algorithm / 14.5.2： |
| Q-learning algorithm / 14.5.3： |
| SARSA / 14.5.4： |
| TD(λ) algorithm / 14.5.5： |
| Large state space / 14.5.6： |
| Conclusion / 14.6： |
| Linear Algebra Review / A： |
| Vectors and norms / A.1： |
| Norms / A.1.1： |
| Dual norms / A.1.2： |
| Matrices / A.2： |
| Matrix norms / A.2.1： |
| Singular value decomposition / A.2.2： |
| Symmetric positive semidefinite (SPSD) matrices / A.2.3： |
| Convex Optimization / B： |
| Differentiation and unconstrained optimization / B.1： |
| Convexity / B.2： |
| Constrained optimization / B.3： |
| Probability Review / B.4： |
| Probability / C.1： |
| Random variables / C.2： |
| Conditional probability and independence / C.3： |
| Expectation, Markov's inequality, and moment-generating function / C.4： |
| Variance and Chebyshev's inequality / C.5： |
| Concentration inequalities / D： |
| Hoeffding's inequality / D.1： |
| McDiarmid's inequality / D.2： |
| Other inequalities / D.3： |
| Binomial distribution: Slud's inequality / D.3.1： |
| Normal distribution: tail bound / D.3.2： |
| Khintchine-Kahane inequality / D.3.3： |
| Notation / D.4： |
| References |
| Index |
| Preface |
| Introduction / 1： |
| Applications and problems / 1.1： |
図書
