Advances in Knowledge Discovery and Data Mining

Name: Advances in Knowledge Discovery and Data Mining | 22nd Pacific-Asia Conference, PAKDD 2018, Melbourne, VIC, Australia, June 3-6, 2018, Proceedings, Part I
Brand: Springer
Price: 96.29 EUR
Availability: OnlineOnly

22nd Pacific-Asia Conference, PAKDD 2018, Melbourne, VIC, Australia, June 3-6, 2018, Proceedings, Part I

Dinh Phung Vincent S. Tseng Geoffrey I. Webb Bao Ho Mohadeseh Ganji Lida Rashidi(Herausgeber*in)

Springer (Verlag)

Erschienen am 18. Juni 2018

XXVII, 701 Seiten

E-Book

PDF mit Wasserzeichen-DRM

Systemvoraussetzungen

978-3-319-93034-3 (ISBN)

96,29 €inkl. 7% MwSt.

Systemvoraussetzungen

für PDF mit Wasserzeichen-DRM

E-Book Einzellizenz

Als Download verfügbar

Beschreibung

Weitere Details

Weitere Ausgaben

Inhalt

Intro
PC Chairs' Preface
General Chairs' Preface
Organization
Contents - Part I
Classification and Supervised Machine Learning
Classifier Risk Estimation Under Limited Labeling Resources
1 Introduction
2 Problem Formulation
3 Estimation Methods
3.1 Simple Random Sampling
3.2 Stratified Sampling
3.3 Allocation Methods for Stratified Sampling
3.4 Comparison of Variances
3.5 Stratification Methods
4 Experiments and Results
4.1 Proportional and Equal Allocation
4.2 Optimal Allocation
4.3 Dependence on True Accuracy
5 Discussions and Conclusions
References
Social Stream Classification with Emerging New Labels
1 Introduction
2 Related Work
3 Preliminaries
4 The Proposed Framework
4.1 NL-Forest: Training Process
4.2 NL-Forest: Deployment
4.3 NL-Forest: Model Update
4.4 Model Complexity
5 Experiment
5.1 Experimental Setup
5.2 Simulated Data Stream
5.3 Real Data Stream
5.4 Sensitivity of Parameters
6 Conclusion
References
Exploiting Anti-monotonicity of Multi-label Evaluation Measures for Inducing Multi-label Rules
1 Introduction
2 Preliminaries
2.1 Multi-label Rule Learning
2.2 Bipartition Evaluation Functions
3 Properties of Multi-label Evaluation Measures
4 Algorithm for Learning Multi-label Head Rules
5 Evaluation
6 Related Work
7 Conclusions
References
Modeling Label Interactions in Multi-label Classification: A Multi-structure SVM Perspective
1 Introduction
2 A Quick Review of Existing Work
3 Multi-structure SVM
4 Dual MSSVM and an Efficient Optimization Algorithm
5 Experiments
6 Conclusion
References
Sentiment Classification Using Neural Networks with Sentiment Centroids
1 Introduction
2 Related Work
2.1 Sentiment Features Learning
2.2 Neural Networks for Sentiment Classification
3 Our Approach
3.1 Text Sequence Encoder Models
3.2 Sentiment Centriods Constraint
4 Experiments
4.1 Datasets
4.2 Evaluation Metrics
4.3 Training Settings
4.4 Sentence-Level Classification
4.5 Document-Level Classification
4.6 The Effect of Sentiment Centroids
5 Conclusion and Future Work
References
Random Pairwise Shapelets Forest
1 Introduction
2 Random Pairwise Shapelets Forest
2.1 Providing More Information by Combination
2.2 Proposed Algorithm
3 Decomposed Mean Decrease Impurity
4 Experiment and Evaluation
4.1 Experimental Setup
4.2 Predictive Performance
4.3 Computational Performance
5 Case Studies
5.1 GunPoint
5.2 ArrowHead
6 Conclusion
References
A Locally Adaptive Multi-Label k-Nearest Neighbor Algorithm
1 Introduction
1.1 Background
1.2 Motivation
1.3 Paper Organization
2 Related Work
3 Methodology
4 Experiment
4.1 Experiment Setup
4.2 Results
5 Conclusion
References
Classification with Reject Option Using Conformal Prediction
1 Introduction
2 Conformal Classifiers
3 Error Probabilities Using Posterior Information
3.1 Getting Rid of
4 Experiments
5 Concluding Remarks
References
Target Learning: A Novel Framework to Mine Significant Dependencies for Unlabeled Data
1 Introduction
2 Bayesian Network Classifiers
3 Target Learning
4 Experimental Study
5 Conclusion
6 Code
References
Automatic Chinese Reading Comprehension Grading by LSTM with Knowledge Adaptation
1 Introduction
2 Model for Automatic Open-Ended Chinese Reading Comprehension Grading
2.1 Negative Sampling Based Continuous Bag-of-Words Embedding
2.2 Knowledge Adaptation for Continuous Bag-of-Word Embedding
2.3 Recurrent Layer
2.4 Fully-Connected Layer with Softmax Activation
3 Experiments
3.1 Data Sets and Preprocessing
3.2 Baselines
3.3 Results and Analysis
3.4 Parameter Sensitivity
4 Related Work
5 Conclusions
References
Data Mining with Algorithmic Transparency
1 Introduction
2 Related Work
3 Problem Definition
4 The Reverse Engineering Approximate Learning (REAL) Model
4.1 Minimum Cost and Complexity Sampling
4.2 Direct Hypothesis Formation
4.3 Indirect Hypothesis Formation
5 Experimental Results
5.1 Experiments on UCI Datasets
6 Conclusions and Future Work
References
Cost-Sensitive Reference Pair Encoding for Multi-Label Learning
1 Introduction
2 Preliminary
3 Proposed Approach
4 Active Learning for CSMLC
5 Experiments
6 Conclusion
References
Fuzzy Integral Optimization with Deep Q-Network for EEG-Based Intention Recognition
1 Introduction
2 Preliminaries
3 Methodology
3.1 Local Spatio-Temporal Information Extraction
3.2 Global Temporal Information Extraction
3.3 Choquet Integral with Deep Q-Network
4 Experiments
4.1 Dataset and Model Implementation
4.2 Compared Algorithms
4.3 Experimental Result
5 Conclusion
References
Heterogeneous Domain Adaptation Based on Class Decomposition Schemes
1 Introduction
2 Problem Formulation
3 Class Decomposition Schemes and Coding Matrices
4 Class Code Alignment Algorithm
4.1 Detailed Description
4.2 CCA Classification of Target Instances
5 Experiments
5.1 Settings of the CCA Algorithm
5.2 Baseline Classifiers
5.3 Experiments on the Office Dataset
5.4 Experiments on the Wikipedia Dataset
5.5 Experiments on the Multiple Feature Dataset
5.6 Results and Discussions
6 Conclusion
References
A Deep Neural Spoiler Detection Model Using a Genre-Aware Attention Mechanism
1 Introduction
2 Related Work
2.1 Spoiler Detection
2.2 Attention Mechanism
3 Dataset Analysis
3.1 Dataset Description
3.2 Spoiler Characteristics Analysis
4 Our Approach
4.1 Task Formulation
4.2 Genre Encoder
4.3 Sentence Encoder
4.4 Binary Classifier
5 Experimental Evaluation
5.1 Experimental Setup
5.2 Results
6 Conclusion
References
Robust Semi-Supervised Learning on Multiple Networks with Noise
1 Introduction
2 Problem Formulation
2.1 Graph-Based Semi-supervised Learning
2.2 Objective Formulation
2.3 Graph Weights Interpreted
3 iMUNE Algorithm
3.1 Complexity Analysis
4 Evaluation
4.1 Experiment Setup
4.2 Parameters
4.3 Evaluation Results
5 Conclusion
References
-Distance Weighted Support Vector Regression
1 Introduction
2 Background
2.1 Recent Progress in SV Theory
3 The Proposed -DWSVR
3.1 The Formulation of -DWSVR
3.2 The Regression of Medium Problems with Kernel Functions
3.3 The Regression of Larger Problems
4 Experiments
4.1 Experimental Setup
4.2 Results and Discussion
4.3 Parameter Effects
4.4 Time Cost
References
Healthcare, BioInformatics and Related Topics (Application)
Corrosion Prediction on Sewer Networks with Sparse Monitoring Sites: A Case Study
1 Introduction
2 Case Study Background
3 Preliminaries
3.1 Related Work on Sewer Corrosion
3.2 Brief Introduction to Gaussian Process
4 Methodology
4.1 Gaussian Process Based Prediction Model
4.2 Factor Estimation
4.3 Corrosion Prediction
5 Case Study
5.1 Evaluation
5.2 Discussion
6 Conclusion
References
CAPED: Context-Aware Powerlet-Based Energy Disaggregation
1 Introduction
2 Related Work
3 Preliminary
3.1 Problem Definition
3.2 Description of a Typical Training Dataset
4 Proposed Approach: CAPED
4.1 Learning Powerlets
4.2 Estimation of Context-Aware Occurrence Probability
4.3 Context-Aware Signal Decoding
5 Experiments
5.1 Experimental Setup
5.2 Performance Discussion
5.3 Parameter Study
6 Conclusion
References
Rolling Forecasting Forward by Boosting Heterogeneous Kernels
1 Introduction
2 Background: Network Traffic and Device Configuration
3 Related Work
4 Solution
4.1 Overall Process
4.2 Kernel Design
5 Experiments
5.1 Data
5.2 Setup
5.3 Comparison Results
5.4 Contribution of Kernels
6 Conclusion
References
IDLP: A Novel Label Propagation Framework for Disease Gene Prioritization
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Notations
2.3 Dual Label Propagation on Heterogeneous Network
2.4 Improved Dual Label Propagation on Heterogeneous Network
3 Results
3.1 Baselines
3.2 Experimental Settings
3.3 Evaluation
3.4 Accuracy Evaluation
4 Robustness Evaluation of IDLP
5 Conclusion
References
Deep Learning for Forecasting Stock Returns in the Cross-Section
Abstract
1 Introduction
2 Related Works
3 Data and Methodology
3.1 Dataset for MSCI Japan Universe
3.2 Problem Definition
3.3 Training and Prediction
3.4 Performance Measures
3.5 Compared Models
4 Experimental Results
4.1 Shallow Versus Deep Neural Networks
4.2 Comparison with Support Vector Regression and Random Forests
4.3 Ensemble
4.4 Long-Short Portfolio Strategy
5 Conclusions
References
Vine Copula-Based Asymmetry and Tail Dependence Modeling
1 Introduction
2 Preliminaries
2.1 Vine Copula
2.2 Tail Dependencies
3 Our Weighted Partial Regular Vine Model
3.1 Partial Regular Vine Construction
3.2 Vine Structure Selection
3.3 Bivariate Copula Selection
3.4 Marginal Distribution Specification and Parameter Estimation
4 Case Study
4.1 Data and Marginal Distribution Specification
4.2 Regular Vine Copula Structure Specification and Tail Dependence Analysis
4.3 Out-of-Sample Performance Analysis
5 Conclusion and Future Work
References
Detecting Forged Alcohol Non-invasively Through Vibrational Spectroscopy and Machine Learning
1 Introduction
2 Background
2.1 Spectroscopy
2.2 Classification
3 Data
4 Experimental Setup
5 Results
5.1 Leave-one-bottle-out Cross Validationa
5.2 Classifying the Bottleb
5.3 PCA Transformsc
6 Conclusions
References
Research and Application of Mapping Relationship Based on Learning Attention Mechanism
1 Introduction
2 Problem Definition
2.1 Definition
2.2 Evaluation Method
2.3 Problem Definition
3 Definition of Model
3.1 Encoder
3.2 Decoder
3.3 Attention Mechanism
4 Experiments and Results
4.1 Datasets
4.2 Results and Analysis
5 Conclusions and Future Work
References
Human Identification via Unsupervised Feature Learning from UWB Radar Data
1 Introduction
2 Related Work
3 Characteristics of the UWB Data
4 Unsupervised Feature Learning and Classification for Human Identification
4.1 Convolutional Extraction
4.2 Pre-Processing
4.3 Unsupervised Feature Learning
4.4 Feature Transformation
4.5 Classification
5 Experiments
5.1 Data Collection and Pre-Processing
5.2 Effect of Whitening and Patch Size
5.3 Effect of Number of Feature Bases
5.4 Final Classification Results
6 Conclusion and Future Work
References
Prescriptive Analytics Through Constrained Bayesian Optimization
1 Introduction
2 Framework
2.1 Prescriptive Analytics
2.2 Computing c
2.3 Optimization
3 Experiments
3.1 Constrained Bayesian Optimization vs Genetic Algorithm
3.2 Iris Dataset
3.3 Application to Policy Design for Better Community Health
4 Conclusion
References
Neighborhood Constraint Matrix Completion for Drug-Target Interaction Prediction
1 Introduction
2 Materials and Method
2.1 Materials
2.2 Notation and Problem Description
2.3 Neighborhood Constraint Matrix Completion
3 Results and Discussion
3.1 Comparison Methods
3.2 Experimental Settings
3.3 Performance Results
4 Conclusion
References
Detecting Hypopnea and Obstructive Apnea Events Using Convolutional Neural Networks on Wavelet Spectrograms of Nasal Airflow
1 Introduction
2 Related Work
3 Data
3.1 Data Preprocessing and Preparation
4 Method
4.1 Signal Normalization
4.2 CNN Design
4.3 1-D CNN: Nasal Signal
4.4 2-D CNN: Nasal Signal Spectrogram
5 Results and Discussion
6 Conclusion
References
Deep Ensemble Classifiers and Peer Effects Analysis for Churn Forecasting in Retail Banking
1 Introduction
2 Related Work
3 Data
3.1 Feature Engineering
4 Methods
4.1 Deep Convolutional Neural Networks
4.2 Deep Ensemble Classifier
5 Results
6 Conclusion
References
GBTM: Graph Based Troubleshooting Method for Handling Customer Cases Using Storage System Log
1 Introduction
2 Background and Dataset
2.1 Event Message System (EMS) Logs
2.2 Data Filtering
2.3 Graph Construction from EMS Log
3 Troubleshooting Methodology
3.1 GBST Algorithm
3.2 Clustering
3.3 Set Expansion
3.4 Creation of NEPCS and AEPCS
3.5 Ranking Modules and PCS Construction
4 Experimental Setup
4.1 Evaluation Procedure
4.2 Evaluation Metrics
4.3 Baseline Models
5 Evaluation
5.1 Direct Validation
5.2 Indirect Validation
5.3 Comparative Study Across Metrics
5.4 Stability of GBTM
6 Conclusion
References
Fusion of Modern and Tradition: A Multi-stage-Based Deep Network Approach for Head Detection
1 Introduction
2 Deep Motion Information Network
2.1 Motion Image and Its Representation
2.2 Deep Motion Proposals Network
2.3 Coarse-Fine Multi-level CNN for Head Detection
3 Results and Discussion
4 Conclusion
References
Learning Treatment Regimens from Electronic Medical Records
1 Introduction
2 Related Work
3 Methods
3.1 Data Collection and Preprocessing
3.2 Data Representation and Patient Clustering
3.3 Treatment Period Identification
3.4 Learning Group Treatment Regimens
4 Experimental Evaluation
4.1 Experimental Design
4.2 Results
4.3 Evaluation
5 Discussion
6 Conclusion
References
Human, Behaviour and Interactions (Application)
Mining POI Alias from Microblog Conversations
1 Introduction
2 Related Works
3 Selecting Toponym Candidates
4 Finding Compatible Toponyms
4.1 Compatibility Measures
4.2 Tuning Factor Weights
5 Experimental Results
5.1 Dataset and Evaluation Metric
5.2 Results and Discussion
6 Conclusion
References
DyPerm: Maximizing Permanence for Dynamic Community Detection
1 Introduction
2 Related Work
3 Methodology
3.1 The DyPerm Algorithm
4 Experimental Results
4.1 Datasets
4.2 Baseline Methods
4.3 Comparative Evaluation
4.4 Run-Time Analysis
5 Conclusion
References
Mining User Behavioral Rules from Smartphone Data Through Association Analysis
1 Introduction
2 Association Rules: A Background
3 Redundancy in Association Rules
4 Our Approach
4.1 Association Generation Tree (AGT)
5 Experiments
5.1 Dataset
5.2 Evaluation Results
6 Conclusion and Future Work
References
A Context-Aware Evaluation Method of Driving Behavior
1 Introduction
2 Related Work
3 Preliminary
3.1 Problem Description
3.2 Driving Data and Driving Events
3.3 Driving Contexts
4 Analysis of Driving Contexts
5 Evaluation Method
5.1 Fundamental Idea
5.2 Weighting Event in Driving Context
5.3 Evaluating Driving Performance
6 System Design
6.1 Pre-Processing
6.2 Evaluation
7 Experimental Study
7.1 Effectiveness of the Normalizing Method
7.2 Effectiveness of the Evaluation Method
8 Conclusion
References
Measurement of Users' Experience on Online Platforms from Their Behavior Logs
1 Introduction
2 Related Work and Defining Experience Value
3 Framework
4 Learning Experience Values
4.1 Rule-Based Method
4.2 Value Iteration Method
5 Experimentation
6 Results and Discussion
7 Conclusion
References
Mining Human Periodic Behaviors Using Mobility Intention and Relative Entropy
1 Introduction
2 Related Work
2.1 Human Periodic Behavior Mining
2.2 Periodicity Detection
3 Mobility Intention Based Period Behaviors Mining
3.1 Mobility Intention Extraction
3.2 Period Identification
4 Experiment and Analysis
4.1 Periodicity Detection on Synthetic Time Series Data
4.2 Performance Evaluation Using Real Dataset
4.3 Location Prediction on Real Datasets
5 Conclusions
References
Context-Uncertainty-Aware Chatbot Action Selection via Parameterized Auxiliary Reinforcement Learning
1 Introduction
2 Related Work
3 Preliminaries
3.1 Value-Based DRL
3.2 Policy-Based DRL
4 User Simulator
4.1 Data Preparation
4.2 Dialogue Simulation
5 Proposed Model
5.1 Parameterized A3C
5.2 Auxiliary Tasks
5.3 PA4C Model
6 Experiments
6.1 Baseline Models and Hyperparameters
6.2 Results
7 Conclusion
References
Learning Product Embedding from Multi-relational User Behavior
1 Introduction
2 Related Work
2.1 Product Information Network
2.2 Network Embedding
3 Multi-relational Product Network Embedding
3.1 Problem Definition
3.2 Learning Embedding
3.3 Model Optimization
3.4 Algorithm Complexity Analysis
4 Experiments
4.1 Data Sets and Experimental Setup
4.2 Visualizations
4.3 Label Classification
4.4 Parameter Sensitivity and Scalability
5 Conclusions
References
Vulnerability Assessment of Metro Systems Based on Dynamic Network Structure
1 Introduction
2 Related Work
3 Vulnerability Assessment Framework
3.1 Attack Strategies
3.2 Subgrpah Centrality
3.3 Vulnerability Metrics
4 Experimental Analysis and Discussion
4.1 Data Preprocessing
4.2 Topological Properties of Metro Network
4.3 Vulnerability Analysis and Travel Patterns
5 Conclusion
References
Visual Relation Extraction via Multi-modal Translation Embedding Based Model
1 Introduction
2 Related Work
2.1 Knowledge Graph Embedding
2.2 Visual Relation Detection
3 Method
4 Objects Detection Module
5 Visual Phrase Attention Module
6 Translation Embedding Module
7 Experiments and Analysis
7.1 Datasets and Metrics
7.2 Comparison with State-of-the-Art
7.3 Zero-Shot Learning
8 Conclusion
References
Anomaly Detection and Analytics
Sub-trajectory- and Trajectory-Neighbor-based Outlier Detection over Trajectory Streams
1 Introduction
2 Overview
2.1 Preliminary
2.2 Problem Formulation
3 Methodology
3.1 Feature Extraction
3.2 Distance Function
4 Detection Framework
4.1 The Basic Framework
4.2 The Optimized Framework
5 Experiments
5.1 Effectiveness Evaluation
5.2 Efficiency Evaluation
6 Conclusion
References
An Unsupervised Boosting Strategy for Outlier Detection Ensembles
1 Introduction
2 Related Work
3 Boosting for Ensemble Selection
3.1 Construction of the Target Vector
3.2 Weights and Ensemble Diversity
3.3 Boosting Procedure
4 Experiments and Evaluation
4.1 Datasets
4.2 Ensemble Members
4.3 Competitors and Settings
4.4 Results
5 Conclusion
References
DeepAD: A Generic Framework Based on Deep Learning for Time Series Anomaly Detection
1 Introduction
2 Related Work
3 DeepAD Framework
3.1 Time Series Forecasting (TSF)
3.2 Merging Predictions (MP)
3.3 Anomaly Detector (AD)
4 Evaluation
4.1 Dataset
4.2 Evaluation Metrics
4.3 Results
5 Conclusion
References
Anomaly Detection Technique Robust to Units and Scales of Measurement
1 Introduction
2 Preliminaries and Related Work
3 New Method Robust to Units and Scales of Measurement
4 Empirical Evaluation
4.1 Synthetic Datasets
4.2 Benchmark Datasets
5 Concluding Remarks
References
Automated Explanations of User-Expected Trends for Aggregate Queries
1 Introduction
2 Related Work
3 Problem Formulation
3.1 Prospective Trend Problem
4 UTE Architecture
4.1 Naive Splitter
4.2 Basic Merger (BM)
4.3 UTE Splitting Approaches
4.4 XTrend Transformation
4.5 Xtrend Merging
5 Experiments
5.1 Datasets
5.2 Comparing Splitting Algorithms
5.3 Comparing Merging Algorithms
6 Conclusions
References
Social Spammer Detection: A Multi-Relational Embedding Approach
1 Introduction
2 Preliminaries
2.1 Formulating Multi-Relational Spammer Detection
2.2 Feature Design from Multi-Relational Data
3 Methodology
3.1 Multi-Relational Embedding
3.2 Parameter Estimation
4 Experimental Results
4.1 Experimental Setup
4.2 Performance Comparison
5 Related Work
6 Conclusions
References
Opinion Mining and Sentiment Analysis
Learning to Rank Items of Minimal Reviews Using Weak Supervision
1 Introduction
2 Literature Review
3 L2RI: Learning to Rank Items with Weak Supervision
3.1 Learning to Rank Items Model
3.2 A Rank-Oriented Loss Function
3.3 Generating Ranking Scores for Weak Supervision
3.4 The Features
4 Experiments
4.1 Experiment Setup
4.2 Results
5 Conclusion and Future Work
References
Multimodal Mixture Density Boosting Network for Personality Mining
1 Introduction
2 Preliminary
3 Methodology
3.1 DCA Feature Fusion Layer
3.2 Mixture Density Network
3.3 Dynamic Cascade Boosting Network
4 Experiment
4.1 Datasets
4.2 Features Extraction
4.3 Evaluation
4.4 Results
5 Conclusions
References
Identifying Singleton Spammers via Spammer Group Detection
1 Introduction
2 Related Work
3 The Proposed Approach: SSGD
3.1 Inferring Hidden Reviewer-Product Associations
3.2 Finding and Ranking Spammer Groups
4 Experiment Setup
5 Results and Discussion
5.1 Recall and Precision for Singleton Spammer Detection
5.2 Qualitative Analysis of Detected Spammer Groups
6 Conclusions
References
Adaptive Attention Network for Review Sentiment Classification
1 Introduction
2 Related Work
3 Adaptive Attention Network
3.1 Two-Layer AAN Architecture
3.2 AAN for Review Modeling
4 Experiments
4.1 Comparison with Baselines
4.2 Impact of User and Product Embeddings
4.3 Impact of Adaptive Attention Mechanism
5 Conclusion
References
Cross-Domain Sentiment Classification via a Bifurcated-LSTM
1 Introduction
2 Related Works
2.1 Cross-Domain Sentiment Classification
2.2 Multi-task Learning
3 Recurrent Neural Network Models for Text Classification
3.1 Long Short-Term Memory
3.2 Text Classification with LSTM
4 Bifurcated-LSTM for Cross-Domain Sentiment Classification
4.1 Dataset Augmentation
4.2 Bifurcated-LSTM
4.3 Orthogonality Constraints
4.4 Training and Testing
5 Experiments Setting
5.1 Dataset
5.2 Dataset Augmentation
5.3 Hyperparameters
6 Performance Evaluation of Bifurcated-LSTM
6.1 Performance Evaluation
6.2 Performance Comparison
7 Conclusion and Future Work
References
Author Index

Systemvoraussetzungen

Als PDF speichern Als Link merken