Machine Learning, 2nd Edition

Index

a

• accuracy 8, 270
• activation function 81
• activation rate 83, 362
• active learning 284
• AdaBoost 7, 143
• adversarial machine learning 285
• affine function 16, 359
• aggregated classifier 129
• Akaike’s information criterion (AIC) 173
• approximation method 9
• arcing, see boosting
• Aristotle 131
• Armstrong, J. Scott 283
• atomic learners 131
• AUC 275
• Austen, Jane 267
• autoencoder
  • deep 258
  • single-hidden-layer 258

b

• backpropagation 86, 215
  • backward propagation 215
  • forward propagation 215
• bagging 138, 172
• ball 69, 236
• bandwidth parameter
  • of a histogram 65, 189
  • of a kernel density estimate 61
• Bartholomew, David J. 15
• base learners 131
• Bayes
  • -ian 9, 46, 270
  • -ian information criterion (BIC) 174
  • classifier 12, 45, 47, 269, 292
  • error rate 47, 129
  • factor 269
  • naive 68
  • regression 122
  • risk 47, 122
• Bellman, Richard E. 235
• Benjamini–Hochberg procedure 279
• Benjamini–Hochberg–Yekutieli procedure 279
• BFGS algorithm 77, 205
• bias
  • contribution to risk 36, 122
  • in neural networks 82, 119
  • in software 50–51
  • in training data 49, 51–52
  • of a predictor 125, 129
  • of an estimator 20, 122, 136
• bias-variance trade-off 20, 37, 121–130, 136, 164
  • general loss 125–130
  • squared-error loss 121–125
• BIC, see Bayesian information criterion
• bold driver algorithm 206
• Bonferroni correction 278
• boosting
  • AdaBoost 146
  • by resampling (arcing) 143
  • by reweighting 143
  • general 143
  • gentle 153
  • gradient 151
  • LogitBoost 150
• bootstrap
  • aggregation 138
  • in random forest 141
  • in risk estimation 171, 172
  • sample 138
  • sample size 138
• Box, George E. ...