Machine Learning and Knowledge Discovery in Databases : Applied Data Science and Demo Track / / edited by Gianmarco De Francisci Morales [and five others]

• Record Nr.: UNINA9910746295803321
• Pubbl/distr/stampa: Cham, Switzerland : , : Springer, , [2023]; ©2023
• ISBN: 3-031-43427-7
• Edizione: [First edition.]
• Descrizione fisica: 1 online resource (745 pages)
• Collana: Lecture Notes in Computer Science Series ; ; Volume 14174
• Disciplina: 006.3
• Soggetti: Data mining; Databases; Machine learning
• Lingua di pubblicazione: Inglese
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Nota di bibliografia: Includes bibliographical references and index.
• Nota di contenuto: Intro -- Preface -- Organization -- Invited Talks Abstracts -- Neural Wave Representations -- Physics-Inspired Graph Neural Networks -- Mapping Generative AI -- Contents - Part VI -- Applied Machine Learning -- Rectifying Bias in Ordinal Observational Data Using Unimodal Label Smoothing -- 1 Introduction -- 2 Related Work -- 3 Unimodal Label Smoothing Based on the Geometric Distribution -- 3.1 Motivation -- 3.2 Basic Unimodal Label Smoothing -- 3.3 Class-Wise Unimodal Label Smoothing Using a Smoothing Relation -- 3.4 Unimodal Smoothing Heuristics for Prescriptive Machine Learning -- 4 Evaluation -- 4.1 Relation and Priors Based Smoothing Results -- 4.2 Time Based Smoothing Results -- 5 Conclusion -- References -- Class-Conditional Label Noise in Astroparticle Physics -- 1 Introduction -- 2 Binary Classification in Astroparticle Physics -- 2.1 Source Detection -- 2.2 Noisy Labels of Real Telescope Data -- 3 Related Work on Class-Conditional Label Noise -- 3.1 Class Imbalance in CCN -- 3.2 Other Types of Label Noise -- 4 Partially-Known Class-Conditional Label Noise -- 5 Experiments -- 5.1 Baseline Methods -- 5.2 Merits of PK-CCN: Methodology -- 5.3 Merits of PK-CCN: Results on Conventional Imbalanced Data -- 5.4 Case Study: Detection of the Crab Nebula -- 6 Conclusion and Outlook -- References -- A Baseline ; Generative Probabilistic Model for Weakly Supervised Learning -- 1 Introduction -- 2 Related Work -- 3 Model Formulation -- 3.1 Labelling Functions -- 3.2 Factor Analysis -- 3.3 Weakly Supervision with Factor Analysis -- 3.4 Other Probabilistic Generative Latent Variable Models -- 4 Datasets -- 5 Experiments -- 6 Benefits of the Model -- 7 Limitations -- 8 Discussion and Future Direction -- References -- DyCOD - Determining Cash on Delivery Limits for Real-Time E-commerce Transactions via Constrained Optimisation Modelling -- 1 Introduction.; 2 System View -- 3 Problem Formulation -- 4 Data Collection Design -- 5 Dynamic COD Algorithm -- 5.1 Customer Score Generation -- 5.2 Customer Segmentation -- 5.3 Limit Allocation -- 6 Real-Time Inference -- 7 Results -- 7.1 Offline Evaluation -- 7.2 Online A/B Experiment -- 8 Related Work -- 9 Conclusion -- 10 Future Scope -- References -- Computational Social Sciences -- Pattern Mining for Anomaly Detection in Graphs: Application to Fraud in Public Procurement -- 1 Introduction -- 2 Related Work -- 3 Problem Formulation -- 4 PANG Framework -- 4.1 Description of the Framework -- 4.2 Assessment on Benchmarks -- 5 Public Procurement Use Case -- 5.1 Extraction of the Graph Dataset -- 5.2 Results on Public Procurement Data -- 6 Conclusion -- References -- Aspect-Based Complaint and Cause Detection: A Multimodal Generative Framework with External Knowledge Infusion -- 1 Introduction -- 2 Related Studies -- 3 Dataset Extension -- 3.1 Annotator Details -- 3.2 Annotation Phase and Dataset Analysis -- 4 Methodology -- 4.1 Problem Formulation -- 4.2 Redefining AbCC Task as Multimodal Text-to-Text Generation Task -- 4.3 Multimodal Generative Aspect-Based Complaint and Cause Detection Framework (MuGACD) -- 5 Experiments and Results -- 5.1 Baselines -- 5.2 Experimental Setup -- 5.3 Results and Discussion -- 5.4 Error Analysis -- 6 Conclusion -- References -- Sequence-Graph Fusion Neural Network for User Mobile App Behavior Prediction -- 1 Introduction -- 2 Related Work -- 2.1 Traditional Methods -- 2.2 Sequence-Based Deep Learning Methods -- 2.3 Graph-Based Deep Learning Methods -- 3 Problem Definition -- 4 Methodology -- 4.1 Framework Overview -- 4.2 Sequence Block: Learning the App Switch Patterns -- 4.3 BipGraph Block: Learning the User-App Similarity -- 4.4 HyperGraph Block: Learning the Correlations in Hyperedges -- 4.5 Optimization -- 5 Experiment.; 5.1 Setup -- 5.2 Performance Evaluation -- 5.3 Ablation Study -- 5.4 Case Study -- 6 Conclusion -- References -- DegUIL: Degree-Aware Graph Neural Networks for Long-Tailed User Identity Linkage -- 1 Introduction -- 2 Related Work -- 3 Preliminaries -- 3.1 Problem Formulation -- 3.2 Graph Neural Networks -- 4 The Proposed Framework: DegUIL -- 4.1 Uncovering Absent Neighborhood -- 4.2 Removing Noisy Neighborhood -- 4.3 Adaptive Aggregation -- 4.4 Training Loss -- 4.5 Characteristics of DegUIL -- 5 Experiments -- 5.1 Experimental Settings -- 5.2 Result -- 6 Conclusion -- References -- Ex-ThaiHate: A Generative Multi-task Framework for Sentiment and Emotion Aware Hate Speech Detection with Explanation in Thai -- 1 Introduction -- 2 Related Works -- 3 Ex-ThaiHate Dataset Development -- 3.1 Data Annotation -- 4 Methodology -- 4.1 Redefining Explainable HSD Task as Text-to-Text Generation Task -- 4.2 Sequence-to-Sequence Learning (Seq2Seq) -- 5 Experimental Results and Analysis -- 5.1 Experimental Settings and Baselines Setup -- 5.2 Findings from Experiments -- 5.3 Error Analysis -- 6 Conclusion and Future Works -- References -- Deep Serial Number: Computational Watermark for DNN Intellectual Property Protection -- 1 Introduction -- 2 Embedding Deep Serial Number in DNNs -- 2.1 Requirements for ; Serial Number Watermarking -- 2.2 The Proposed DSN Framework -- 2.3 Entangled Watermark Embedding -- 2.4 Serial Number Space -- 3 Experiments -- 3.1 Experimental Setups -- 3.2 Prediction Distortion Analysis -- 3.3 Prediction Reliability Analysis -- 3.4 Attacking Robustness Analysis -- 4 A Case Study on PDF OCR Model -- 5 Related Work -- 5.1 Digital Watermarks for DNNs -- 6 Conclusions and Future Work -- 7 Limitations and Ethical Statement -- References.; How Early Can We Detect? Detecting Misinformation on Social Media Using User Profiling and Network Characteristics -- 1 Introduction -- 2 Related Works -- 3 Problem Definition -- 4 CIPHER: Methodology -- 4.1 Network Construction -- 4.2 User Causal Model -- 4.3 Temporal Characteristics -- 4.4 Linguistic Pattern Analysis -- 5 Experimental Evaluations -- 6 Conclusion -- References -- Boosting the Performance of Deployable Timestamped Directed GNNs via Time-Relaxed Sampling -- 1 Introduction -- 2 Time Relaxed Directed - GNN -- 2.1 Graph Neural Networks -- 2.2 Proposed: TRD-GNN -- 2.3 Adaptation for Mini-Batch Training -- 2.4 Comparison with Temporal Graph Networks -- 3 Experiments -- 3.1 Experimental Setup -- 3.2 Performance Comparison (EQ1) -- 3.3 Analysis of TRD-GNN (EQ2) -- 3.4 Time Comparison (EQ3) -- 4 Related Work -- 5 Application to Industry -- 6 Conclusion -- References -- Semi-Supervised Social Bot Detection with Initial Residual Relation Attention Networks -- 1 Introduction -- 2 Related Work -- 3 Problem Definition and Preliminaries -- 3.1 Problem Definition -- 3.2 Homophily and Heterophily -- 4 Human-Bot Network Analysis -- 5 SIRAN -- 5.1 Overview of SIRAN -- 5.2 Heterophily-Aware Attention Mechanism -- 5.3 Initial Residual Connection -- 5.4 Training and Optimization -- 6 Experiments -- 6.1 Experimental Setup -- 6.2 Overall Results -- 6.3 Ablation Study -- 6.4 Robustness Study -- 7 Conclusion and Future Work -- References -- Finance -- On Calibration of Mathematical Finance Models by Hypernetworks -- 1 Introduction -- 2 Related Work -- 2.1 Mathematical Finance Models for Option Pricing -- 2.2 Machine Learning Based Calibration Methods -- 3 Preliminaries -- 3.1 Options -- 3.2 Pricing Models -- 3.3 Calibration -- 4 Methodology -- 4.1 Hypernetwork -- 4.2 Pseudo Data Generation -- 5 Experiments -- 5.1 Datasets -- 5.2 Experimental Settings.; 5.3 Results Analysis -- 5.4 Ablation Study -- 6 Conclusion -- 7 Ethical Implications -- References -- PU GNN: Chargeback Fraud Detection in P2E MMORPGs via Graph Attention Networks with Imbalanced PU Labels -- 1 Introduction -- 2 Related Work -- 2.1 Fraud Detection -- 2.2 Graph Neural Networks -- 2.3 Imbalanced Positive and Unlabeled Learning -- 3 Problem Definition -- 4 Proposed Method -- 4.1 Player Behavior Modeling -- 4.2 Graph Attention Networks -- 4.3 Imbalanced Positive and Unlabeled Learning -- 4.4 Loss, Training and Inference -- 5 Experiments -- 5.1 Experimental Setup -- 5.2 Performance Evaluation -- 5.3 Ablation Study -- 6 Conclusion and Feature Work -- References -- BCAD: An Interpretable Anomaly Transaction Detection System Based on Behavior Consistency -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Concepts of Behavior -- 3.2 Overview of BCAD -- 3.3 Representation Learning -- 3.4 Behavior Consistency Contrastive Learning -- 3.5 Attribute Consistency Contrastive Learning -- 4 Experiment -- 4.1 Dataset -- 4.2 Experiment Settings -- 4.3 Model Performance -- 4.4 Ablation Study -- 4.5 Interpretability -- 4.6 Ethics -- 5 Conclusion -- References -- Advancing Fraud Detection Systems Through Online Learning -- 1 Introduction -- 2 Background and Related Work -- 3 Threat Model -- 3.1 Banking Fraud -- 4 Fraud Detection Systems -- 5 Online Learning Approach -- 6 Experimental Evaluation -- 6.1 Dataset -- 6.2 Fraud Strategy Modeling -- 6.3 Competing Models -- 6.4 Performance Metric -- 6.5 Experiment 1: ; Alternation of Fraud Strategies -- 6.6 Experiment 2: Adversarial Alternation of Fraud Strategies -- 7 Conclusion -- References -- Hardware and Systems -- Continual Model-Based Reinforcement Learning for Data Efficient Wireless Network Optimisation -- 1 Introduction -- 2 Related Work -- 3 Methods -- 3.1 Description of the Dataset.; 3.2 Problem Formulation.