Machine Learning and Data Mining in Pattern Recognition: 14th International Conference, MLDM 2018, New York, NY, USA, July 15-19, 2018, Proceedings 1

This two-volume set LNAI 10934 and LNAI 10935 constitutes the refereed proceedings of the 14th International Conference on Machine Learning and Data Mining in Pattern Recognition, MLDM 2018, held in New York, NY, USA in July 2018. The 92 regular papers presented in this two-volume set were carefully reviewed and selected from 298 submissions. The topics range from theoretical topics for classification, clustering, association rule and pattern mining to specific data mining methods for the different multi-media data types such as image mining, text mining, video mining, and Web mining. Preface Organization Contents – Part I Contents -- Part II MaxMin Linear Initialization for Fuzzy C-Means 1 Introduction 2 Related Works 3 MaxMin Linear Fuzzy Clustering Initialization Method 4 Transformed Standardized Fuzzy Difference Validity Index 5 Experimental Validation 5.1 Datasets 5.2 Experimental Settings 5.3 Experimental Results 6 Conclusion and Perspectives References Automatic Rail Flaw Localization and Recognition by Featureless Ultrasound Signal Analysis Abstract 1 Introduction 2 Rail Flaws Localization via Ultrasound B-Scan Signals Analysis 2.1 Representation of Ultrasound Testing Results in B-Scan Form 2.2 Ultrasound B-Scan as Multi-component Discrete Signal 2.3 B-Scan Convergence Procedure 2.4 B-Scan Segmentation Procedure 3 B-Scan Segments Comparison 3.1 B-Scan Elements Comparison 3.2 DTW-Based B-Scan Segments Comparison 4 Rail Flaws Recognition Experiments 5 Conclusion Acknowledgements References Social Media Sentiment Analysis Based on Domain Ontology and Semantic Mining Abstract 1 Introduction 2 Related Work 3 Construction of Domain Ontology in Social Media 3.1 Extracting Topic Characteristics of Terms 3.2 Clustering Algorithm of Topic Items 4 Identification of Topic-Opinion Pairs Based on Domain Ontology 4.1 Identification of Subjective Sentence 4.2 Extraction of Key Subjective Sentence and Relations 5 Post Sentiment Analysis Based on Domain Ontology 6 Experiments and Results 7 Discussion References Mining Associations in Large Graphs for Dynamically Incremented Marked Nodes 1 Introduction 2 Developing the Algorithm 2.1 Problem Complexity 2.2 Proposed Method for Addition of New Marked Nodes 3 Results 4 Conclusion References A New Global Foreground Modeling and Local Background Modeling Method for Video Analysis 1 Introduction 2 A New Feature Vector for Feature Representation 3 Local Background Modeling Using a Single Gaussian Density 4 Novel Global Foreground Modeling 5 Foreground and Background Classification 6 Experiments 7 Conclusions References A Deep Learning Based Automatic Severity Detector for Diabetic Retinopathy 1 Introduction 2 Related Works 3 Approach 3.1 The Dataset, Hardware, Software 3.2 Computational Considerations 3.3 Preprocessing 3.4 Data Augmentation 3.5 Network Architecture 4 Results and Discussion 5 Conclusion References Compact Representation of Documents Using Terms and Termsets 1 Introduction 2 Experimental Methodology 3 Construction of Document Vectors 4 Experiments 5 Conclusions References PATENet: Pairwise Alignment of Time Evolving Networks Abstract 1 Introduction 2 Preliminaries 2.1 The Smith-Waterman (SW) Sequence Alignment Algorithm 3 PATENet 3.1 Alternative Substitution Matrix Construction 3.2 From SW to PATENet 3.3 OSN Alignment Score 4 Experiments 4.1 Empirical Design 4.2 Synthetic Data Generation 4.3 Results 5 Discussion 5.1 Additional Considerations and Future Directions 5.2 Generalizations 6 Conclusion Acknowledgments References Automated Identification of Potential Conflict-of-Interest in Biomedical Articles Using Hybrid Deep Neural Network Abstract 1 Introduction 2 Related Works 3 Conflict-of-Interest: Issues and Challenges 4 Proposed Method 4.1 Preprocessing: Sentence Splitting 4.2 Input Vector Representation 4.3 HDNN Architecture 5 Classification Experiments 5.1 Ground-Truth Dataset and Tools 5.2 Experimental Results 6 Conclusions Acknowledgment References Hierarchical Bayesian Classifier Combination 1 Introduction 2 Bayesian Classifier Combination 2.1 Bayesian Model Averaging 2.2 Bayesian Methods in Classifier Combination 3 Hierarchical Bayesian Combination Model (HBCC) 3.1 The Classifier Combination Model 3.2 The Hierarchical Bayesian Combination of Classifier Model (HBCC) 4 Experiments 4.1 Experiment Setup 4.2 Results and Discussion 5 Conclusion References Optimizing Support Vector Regression with Swarm Intelligence for Estimating the Concrete Compression Strength 1 Introduction 2 Background 2.1 Support Vector Regression 2.2 Fish School Search 2.3 Artificial Bee Colony 2.4 Particle Swarm Optimization 3 Proposal of the Hybrid Algorithms 4 Experimental Setup 5 Results 6 Conclusion References Automated Contradiction Detection in Biomedical Literature 1 Introduction 2 Related Work 3 Dataset 4 Methods 4.1 Identification of Abstract Claims 4.2 Contradiction Detection 5 Results 5.1 Claim Extraction Results 5.2 Contradiction Detection Results 6 Conclusions and Future Work References Following the Common Thread Through Word Hierarchies 1 Introduction 2 Related Work 3 Taxonomy Generation 3.1 An Algorithm for Taxonomy Extraction 3.2 Taxonomy Evolution 4 Example 5 Conclusion References Document Clustering Using Local and Universal Knowledge Abstract 1 Introduction 2 The Proposed Methods 2.1 First Method: Combining the Similarities 2.2 Second Method: Concatenating the Feature Vectors 3 Experiments and Discussion 3.1 Datasets 3.2 Experiments 4 Summary and Future Work References Tweet Classification Using Sentiment Analysis Features and TF-IDF Weighting for Improved Flu Trend Detection 1 Introduction 2 Related Work 3 Methodology 3.1 Corpus 3.2 Preprocessing 3.3 Feature Extraction 3.4 Training and Testing 3.5 Performance Metrics 4 Discussion and Results 5 Conclusion References Adaptive Adjacency Kanerva Coding for Memory-Constrained Reinforcement Learning 1 Introduction 2 Related Work 2.1 Sarsa Algorithm 2.2 Kanerva-Based Learning 3 Adaptive Adjacency Kanerva Coding 3.1 Prototype Adjacencies 3.2 Adaptively Changing Adjacencies 4 Experimental Evaluation 4.1 Evaluate Performance with Mountain Car Domain 4.2 Evaluate Performance with Hunter-Prey Domain 5 Conclusion References Predicting Drug Target Interactions Based on GBDT 1 Introduction 2 Methods 2.1 DTIs 2.2 Features of Drug 2.3 Features of Target 2.4 GBDT 2.5 Feature Engineering 3 Experiment 3.1 Validation 3.2 IDs of Drug and Target 3.3 The Proportion of Negative Samples 3.4 Competing Methods 4 Conclusion References From Black-Box to White-Box: Interpretable Learning with Kernel Machines 1 Introduction 2 Related Work 3 Problem Setting 3.1 Kernel Machines 3.2 Random Fourier Features 4 Our Approach 4.1 Kernel Function 4.2 Phase 1: Learning Dense Models 4.3 Phase 2: Learning Sparse Models 4.4 Interpretation 5 Experiments 5.1 Synthetic Data 5.2 Benchmark Data 5.3 Visualization 6 Conclusion References A Two-List Framework for Accurate Detection of Frequent Items in Data Streams Abstract 1 Introduction 2 Relevant Prior Algorithms 2.1 Space-Saving Algorithm 2.2 Filtered Space-Saving Algorithm 2.3 Efficiently Finding the Smallest Frequency Key 3 Proposed Algorithms 3.1 FSS2L Framework 3.2 FSSA Algorithm 3.3 Optimizations that Can Be Added to FSS and FSSA 3.4 Adapting the Ratio of List0 Size to List1 Size in FSS2L 4 Space and Time Complexity of Considered Algorithms 5 Experimental Results 5.1 Experiments on Artificial Data 5.2 Experiments on Real Data 6 Conclusion References Evaluating Frequent-Set Mining Approaches in Machine-Learning Problems with Several Attributes: A Case Study in Healthcare Abstract 1 Introduction 2 Background 3 Method 3.1 Data 3.2 Association-Rule Mining 3.2.1 Apriori Algorithm 3.3 Machine-Learning Algorithms 3.3.1 ZeroR 3.3.2 Decision Tree 3.3.3 Naïve Bayes 3.3.4 Logistic Regression 3.3.5 Support Vector Machines 3.4 Model Calibration 3.4.1 Apriori Implementation 3.4.2 Machine Learning Combined with Apriori 4 Results 4.1 Apriori Algorithm 4.2 Machine-Learning Algorithms 5 Discussion and Conclusions Acknowledgement References A Comparative Study to the Bank Market Prediction Abstract 1 Introduction 2 Related Works 3 Dataset Description 4 Proposed Method 5 Results and Discussion 6 Conclusion References Deep Metric Learning for Sequential Data Using Approximate Information 1 Introduction 2 Deep Metric Learning Using Approximate Information 2.1 Triplet LSTM 2.2 Objective 2.3 Jaccard Distance 2.4 Method 3 Experimental Evaluation 3.1 Baseline 3.2 Model 3.3 Triplet Selection 3.4 Experiment - Metric Learning Using Approximate Information 3.5 Experiment - Classification with Augmentation 3.6 Datasets 3.7 Data Pre-processing 3.8 Hold Out Sets 3.9 Evaluation 3.10 Hyper Parameters and Training Details 4 Results and Discussion 5 Related Work 6 Conclusion and Future Work References Machine Learning Applied to Point-of-Sale Fraud Detection Abstract 1 Introduction 2 Challenges and Related Work 3 Scope 3.1 Fraud Types 3.2 Fraud Types 4 Data, Methods and Tools 4.1 Data Profile 4.2 Data Pre-processing 4.3 Fraud Class Labeling 4.4 Algorithm Selection 4.5 Principal Component Analysis 4.6 Tools 4.7 Definitions 5 Research Results 5.1 Statistical Analysis 5.2 Feature Selection and Engineering 5.3 Algorithm Parameter Tuning 5.4 Algorithm Results 6 Conclusion 7 Future Research Acknowledgement References Prefix and Suffix Sequential Pattern Mining 1 Introduction 2 Sequential Pattern Mining 3 Prefix/Suffix Pattern Mining 3.1 Prefix/Suffix Sequences 3.2 Prefix/Suffix Projections 3.3 Sequential Prefix/Suffix Mining Problems 3.4 Algorithms 4 Theoretical Analysis 4.1 Reduced Problem Size 4.2 Reduced Search Space 4.3 Mining Several Prefix/Suffix Patterns 5 Empirical Analysis 5.1 Empirical Setup 5.2 Retail 5.3 Click-Streams 5.4 Natural Languages 6 Conclusions and Future Work References Long Short-Term Memory Recurrent Neural Network for Stroke Prediction Abstract 1 Introduction 2 Deep Learning 3 ICD-10 Complaint Electronic Healthcare Records 4 Prediction of Stroke Using EHRs and Deep Learning 4.1 The Selection of EHRs Based on Risk Factors 4.2 Deep Learning by Using Long Short -Term Memory Recurrent Neural Networks (LSTM-RNN) 5 Evaluation Model 5.1 Data Source 5.2 Predictive Model 6 Result 7 Conclusion and Future Work Acknowledgements References Adversarial Machine Learning: A Literature Review Abstract 1 Introduction 2 Related Works 3 Methodology 3.1 Overview 3.2 Collection 3.3 Data-Set Restrictions 3.4 Evaluation 4 Results 4.1 Pre-sort 4.2 Raw Counts 4.3 Trends 5 Discussion 5.1 Raw Count Data Analysis 5.2 Trends Data Analysis 6 Conclusion References Reverse Engineering Gene Regulatory Networks Using Graph Mining 1 Introduction 1.1 Related Work 2 Methods 2.1 Definitions and Notation 2.2 Candidate Generation and Pattern Discovery 2.3 Frequency Counting 2.4 Proposed Algorithms 2.5 Correctness and Complexity Analysis 3 Experiments and Results 3.1 Datasets 3.2 Experimental Setup 3.3 Experimental Results 4 Conclusions References Spammer Detection via Combined Neural Network 1 Introduction 1.1 Background 1.2 Related Work 2 Features 2.1 Profile Features 2.2 Behavioral Features 3 The Combined Neural Network 3.1 Model Details 3.2 Model Training 4 Experiment and Analysis 4.1 Data and Evaluation Metrics 4.2 Features Effectiveness Analysis 4.3 The Combined Neural Network 4.4 Performance Tracking 5 Conclusion References Pedestrian Detection: Performance Comparison Using Multiple Convolutional Neural Networks 1 Introduction 2 Model Architecture 2.1 Modern Convolutional Detectors 2.2 Feature Extractors 3 Experimental Setup 4 Results 4.1 Detection Time on Test Images on CPU 4.2 Sample Detections 4.3 Total Loss Function 4.4 Learning Rate 5 Conclusion References Automated Machine Learning Algorithm Mining for Classification Problem Abstract 1 Introduction 2 Foundations and Related Work 2.1 Problem Statement 2.2 Algorithm Selection and Hyper-parameter Optimization Methods 2.3 Hyper-parameter Search 3 Methodology 3.1 Identification of Essential Hyper-parameters 3.2 Algorithm Evaluation and Selection 4 Experimental Evaluation 4.1 Corpora and Setup 4.2 Experimental Results 5 Conclusion and Future Work Acknowledgments References Enhancing Outlier Detection by an Outlier Indicator Abstract 1 Introduction 2 Related Work 3 The Proposed Enhancer for Outlier Mining 3.1 A Simple Idea 3.2 Some Definitions 3.3 Our Proposed Outlier Detection Algorithm 4 Experiments and Results 4.1 Performance of Our Algorithm on Synthetic Data 4.2 Performance of Our Algorithm on Real Data 5 Conclusions Acknowledgment References A Semi-supervised Approach to Discover Bivariate Causality in Large Biological Data 1 Introduction 2 Related Work 3 Pairwise Semi-supervised Causal Inference 3.1 Supervised Causal Inference with Inverse Regression 3.2 Supervised Causality Discovery with Distance Correlation 3.3 Method 4 Experiments on Benchmark Data Sets 4.1 Cause-Effect Pairs 4.2 Network Reconstruction 4.3 Runtime Results for Network Reconstruction 5 Effects of Metformin on Human Gut Composition 6 Conclusions References Risk Scores Learned by Deep Restricted Boltzmann Machines with Trained Interval Quantization 1 Introduction 2 Related Work 3 Problem Statement 4 Deep Restricted Boltzmann Machines 5 Quantized Deep Restricted Boltzmann Machines 6 Trained Interval Quantization 7 Experiments 8 Conclusion References A New Approach for Tuned Clustering Analysis Abstract 1 Introduction 2 Related Work 3 The Approach 3.1 The Concept 3.2 The Pseudo Code of the Tuned Clustering Analysis 4 Experiments 4.1 Objectives 4.2 Medical Overview 4.3 The Echo Heart Measurements Database 4.4 The Implementation 5 Results 6 Discussion 7 Conclusion and Further Research References Correction to: A New Approach for Tuned Clustering Analysis Correction to: Chapter “A New Approach for Tuned Clustering Analysis” in: P. Perner (Ed.): Machine Learning and Data Mining in Pattern Recognition, LNCS 10934, https://doi.org/10.1007/978-3-319-96136-1_34 Author Index