Classification Methods For Internet Applications

Preface......Page 6
Contents......Page 9
1.1 Spam Filtering......Page 13
1.1.1 The Road of Spam......Page 14
1.1.3 Spam Filters......Page 15
1.1.4 Image Spam......Page 17
1.1.6 Spam Filtering as a Classification Task......Page 18
1.2 Recommender Systems......Page 19
1.2.1 Purpose of Recommender Systems......Page 20
1.2.3 Content Based and Collaborative Recommenders......Page 22
1.2.4 Conversational Recommender Systems......Page 27
1.2.5 Social Networks and Recommendations......Page 28
1.2.8 Recommendation as a Classification Task......Page 29
1.3.1 Opinion Mining—Subjectivity, Affect and Sentiment Analysis......Page 30
1.3.2 Sentiment Analysis Systems......Page 32
1.3.3 Open Challenges......Page 36
1.3.5 Sentiment Analysis as a Classification Task......Page 37
1.4 Example-Based Web Search......Page 38
1.4.1 Example-Based Search and Object Annotations......Page 39
1.4.2 Example-Based Search in Text Documents......Page 41
1.4.3 Example-Based Search in General Objects......Page 42
1.4.4 Scribble and Sketch Input......Page 43
1.4.5 Multimedia and Descriptors......Page 44
1.4.6 Example Based Search as a Classification Task......Page 48
1.5 Malware Detection......Page 49
1.5.1 Static Malware Analysis......Page 50
1.5.2 Dynamic Malware Analysis......Page 52
1.5.3 Hybrid Malware Analysis......Page 54
1.5.4 Taxonomy of Malware......Page 55
1.5.5 Malware Detection as a Classification Task......Page 58
1.6.1 A Brief History of IDS......Page 59
1.6.2 Common Kinds of Attacks......Page 61
1.6.3 Host-Based Intrusion Detection......Page 62
1.6.4 Network-Based Intrusion Detection......Page 64
1.6.5 Other Types......Page 71
References......Page 72
2.1 Classifiers and Classes......Page 81
2.1.1 How Many Classes for Spam Filtering?......Page 83
2.2 Measures of Classifier Performance......Page 84
2.2.1 Performance Measures in Binary Classification......Page 87
2.3 Linear Separability......Page 89
2.3.1 Kernel-Based Mapping of Data  into a High-Dimensional Space......Page 91
2.4 Classifier Learning......Page 95
2.4.1 Classifier Overtraining......Page 96
2.4.2 Semi-supervised Learning......Page 97
2.4.3 Spam Filter Learning......Page 101
2.5 Feature Selection for Classifiers......Page 104
2.6.1 Clustering......Page 107
2.6.2 Regression......Page 108
2.6.3 Clustering and Regression in Recommender Systems......Page 110
2.6.4 Clustering in Malware Detection......Page 112
References......Page 114
3.1 Typology of Classification Methods......Page 116
3.2 Classification Based on k-Nearest Neighbours......Page 118
3.2.1 Distances Between Feature Vectors......Page 119
3.2.2 Using k-nn Based Classifiers in Recommender Systems......Page 123
3.2.3 Using k-nn Based Classifiers in Malware and Network Intrusion Detection......Page 128
3.3.1 Bayes Classifiers......Page 129
3.3.2 Bayes Spam Filters......Page 133
3.3.3 Bayes Classification in Sentiment Analysis......Page 134
3.3.4 Logit Method......Page 136
3.3.5 Using Bayes Classifiers and Logit Method in Recommender Systems......Page 140
3.4 Classifiers Estimating Features Distribution......Page 142
3.4.1 Linear and Quadratic Discriminant Analysis......Page 143
3.4.2 Discriminant Analysis in Spam Filtering......Page 144
3.5 Classification Based on Artificial Neural Networks......Page 147
3.5.1 Perceptrons for Linearly Separable Classes......Page 149
3.5.2 Multilayer Perceptrons for General Classes......Page 152
3.5.3 Deep Neural Networks Suitable for Classification......Page 156
3.5.4 ANN-Based Classification in Spam Filtering......Page 161
3.5.5 ANN-Based Classification in Recommender Systems......Page 162
3.5.6 Artificial Neural Networks in Sentiment Analysis......Page 165
3.5.7 ANN-Based Classification in Intrusion Detection......Page 168
References......Page 170
4.1 Relationship of Generalization to the Separating Margin......Page 175
4.2.1 Optimization Task for the Minimal Expected Error......Page 178
4.2.2 Support Vectors and Their Role......Page 180
4.3 Support Vector Machines for Non-linear Classification......Page 182
4.3.1 Extension to Multiple Classes......Page 183
4.3.2 Extension Including Noise Tolerance......Page 184
4.3.3 SVM in Spam Filtering......Page 185
4.3.4 SVM in Recommender Systems......Page 187
4.3.5 SVM in Malware and Network Intrusion Detection......Page 188
4.4 Active Learning and Its Relevance to SVM Classifiers......Page 195
4.4.1 Examples of SVM-Based Active Learning in Internet Applications......Page 197
References......Page 205
5 Aiming at Comprehensibility......Page 207
5.1 Classifier Comprehensibility and Formal Logic......Page 208
5.1.1 Classification Rules in Boolean and Fuzzy Logic......Page 209
5.1.2 Observational Calculus......Page 211
5.1.3 Observational Rules Based on Probability Estimation......Page 213
5.1.4 Observational Rules Based on Hypotheses Testing......Page 215
5.1.5 Classification Rules in Spam Filtering......Page 219
5.1.6 Classification Rules in Recommender Systems......Page 221
5.1.7 Classification Rules in Malware and Network Intrusion Detection......Page 224
5.2 Classification Rules from Genetic Algorithms......Page 230
5.2.1 Main GA Approaches to Classification Rules Learning......Page 231
5.2.2 Genetic Operations for Classification Rules......Page 232
5.3 Rules Obtained from Classification Trees......Page 234
5.3.1 Classification Tree Learning......Page 236
5.3.2 Classification Tree Pruning......Page 239
5.3.3 Classification Trees in Spam Filtering......Page 241
5.3.4 Classification Trees in Recommender Systems......Page 242
5.3.5 Classification Trees in Malware and Network Intrusion Detection......Page 247
References......Page 253
6.1 Connecting Classifiers into a Team......Page 257
6.1.1 Aggregation Functions......Page 259
6.1.2 Classifier Confidence......Page 261
6.1.4 Teams as an Approach to Semi-supervised Learning......Page 262
6.1.5 Main Methods for Team Construction......Page 263
6.1.6 Using Teams of Classifiers with Multimedia Data......Page 266
6.1.7 Teams of Classifiers in Malware and Network Intrusion Detection......Page 267
6.2 Ensembles of Trees—Random Forests......Page 272
6.2.1 Kinds of Random Forests......Page 273
6.2.2 Random Forest Learning......Page 275
6.2.3 Random Forests in Spam Filtering......Page 277
6.2.4 Random Forests in Recommender Systems......Page 278
6.2.5 Random Forests in Malware and Network Intrusion Detection......Page 282
References......Page 287
7 Conclusion......Page 290