Advances in Knowledge Discovery and Data Mining : 28th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2024, Taipei, Taiwan, May 7–10, 2024, Proceedings, Part I / edited by De-Nian Yang, Xing Xie, Vincent S. Tseng, Jian Pei, Jen-Wei Huang, Jerry Chun-Wei Lin.

Format:

Book

Author/Creator:

Yang, De-Nian.

Contributor:

Xie, Xing.

Tseng, Vincent S.

Pei, Jian.

Huang, Renwei.

Lin, Jerry Chun-Wei.

Series:

Lecture Notes in Artificial Intelligence, 2945-9141 ; 14645

Language:

English

Subjects (All):

Artificial intelligence.

Algorithms.

Education--Data processing.

Education.

Computer science--Mathematics.

Computer science.

Signal processing.

Computer networks.

Artificial Intelligence.

Design and Analysis of Algorithms.

Computers and Education.

Mathematics of Computing.

Signal, Speech and Image Processing.

Computer Communication Networks.

Local Subjects:

Artificial Intelligence.

Design and Analysis of Algorithms.

Computers and Education.

Mathematics of Computing.

Signal, Speech and Image Processing.

Computer Communication Networks.

Physical Description:

1 online resource (406 pages)

Edition:

1st ed. 2024.

Place of Publication:

Singapore : Springer Nature Singapore : Imprint: Springer, 2024.

Summary:

The 6-volume set LNAI 14645-14650 constitutes the proceedings of the 28th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2024, which took place in Taipei, Taiwan, during May 7–10, 2024. The 177 papers presented in these proceedings were carefully reviewed and selected from 720 submissions. They deal with new ideas, original research results, and practical development experiences from all KDD related areas, including data mining, data warehousing, machine learning, artificial intelligence, databases, statistics, knowledge engineering, big data technologies, and foundations.

Contents:

Intro

General Chairs' Preface

PC Chairs' Preface

Organization

Contents - Part I

Anomaly and Outlier Detection

Spatial-Temporal Transformer with Error-Restricted Variance Estimation for Time Series Anomaly Detection

1 Introduction

2 Related Work

3 Approach

3.1 Problem Formulation

3.2 Proposed Architecture

3.3 Error-Restricted Probability (ERP) Loss

3.4 Anomaly Score

4 Experiments

4.1 Dataset Description

4.2 Baseline Methods

4.3 Experimental Settings

4.4 Overall Results

4.5 Ablation Study

5 Conclusion

References

Multi-task Contrastive Learning for Anomaly Detection on Attributed Networks

2 Problem Definition

3 The Proposed Framework

3.1 Subgraph Sampling Based Data Augmentation

3.2 Context Matching Contrastive Learning

3.3 Link Prediction Contrastive Learning

3.4 Model Training and Anomaly Score Inference

4.1 Experimental Setup

4.2 Result and Analysis

4.3 Ablation Study

4.4 Parameter Study

5 Related Works

6 Conclusions

SATJiP: Spatial and Augmented Temporal Jigsaw Puzzles for Video Anomaly Detection

2 Related Works

3 Problem Formulation: Frame-Level VAD

4 Proposal: SATJiP

4.1 Preliminary

4.2 Masked Temporal Jigsaw Puzzles (MTJiP)

5 Experiments

5.1 Datasets and Evaluation Metric

5.2 Implementation Details

5.3 Comparison in Detecting Accompanying Anomalies (AA)

5.4 Comparison in Detecting Diverse Video Anomalies

5.5 Ablation Study

5.6 VAD Examples

6 Conclusion

STL-ConvTransformer: Series Decomposition and Convolution-Infused Transformer Architecture in Multivariate Time Series Anomaly Detection

2.1 Prediction-Based Models

2.2 Reconstruction-Based Models.

2.3 Transformers for Time Series Analysis

3 Methodology

3.2 Overall Architecture

3.3 Data Preprocessing

3.4 Decomposition Block

3.5 Local-Transformer Encoder and Decoder

3.6 Loss Function and Anomaly Score

TOPOMA: Time-Series Orthogonal Projection Operator with Moving Average for Interpretable and Training-Free Anomaly Detection

3 TOPOMA: Our Proposed Anomaly Detector

3.2 Moving Average of Orthogonal Projection Operators

3.3 Adaptive Choice of Anomaly Score Thresholds

3.4 Complexity Analysis

4 Results and Discussion

4.1 Synthetic Data

4.2 Real-World Data

Latent Space Correlation-Aware Autoencoder for Anomaly Detection in Skewed Data

3 Problem Formulation

3.1 Robust Hybrid Error with MD in Latent Space

3.2 Objective Function

4.1 Datasets

5 Hyperparameter Sensitivity

SeeM: A Shared Latent Variable Model for Unsupervised Multi-view Anomaly Detection

2 Proposed Model for Multi-view Anomaly Detection

2.1 The SeeM Model and Its Inference

2.2 Complexity Analysis

2.3 Anomaly Score

3 Experiments

3.1 Datasets and Baselines

3.2 Multi-view Anomaly Detection Performance

3.3 Latent Dimension Analysis

3.4 Non-linear Projections

3.5 A Use Case with Real-World Multi-view Data

4 Related Work

Classification

QWalkVec: Node Embedding by Quantum Walk

2 Preliminaries

2.1 Notations

2.2 Quantum Walks on Graphs

3 Related Works

3.1 Problems

4 Proposed Method: QWalkVec

4.1 Algorithm

5 Evaluations.

5.1 Experimental Settings and Dataset

5.2 Overall Results

Human-Driven Active Verification for Efficient and Trustworthy Graph Classification

2.1 Human-in-the-loop Machine Learning

2.2 Deep Learning for Case-Based Reasoning

2.3 Interpretable Graph Neural Networks

3.1 Problem Formulation and Framework Overview

3.2 Human-Compatible Representation Learning

3.3 Interpretable Predictor

3.4 Prediction Explanation

4.1 Datasets and Baselines

4.2 Implementations and Configurations

4.3 Predictive Performance Comparison

4.4 Benefits of Human-AI Interactions

4.5 User Perception of Prediction Explanations

4.6 Is Instance-Level Feedback Helpful in Any Cases?

5 Discussions of Fairness and Ethical Issues

6 Conclusion and Future Work

SASBO: Sparse Attack via Stochastic Binary Optimization

3 Methods

3.1 Problem Definition

3.2 Sparse Adversarial Attack via Stochastic Binary Optimization

4 Experiments and Results

4.1 Non-targeted Attack

4.2 Targeted Attack

4.3 Visualization

LEMT: A Label Enhanced Multi-task Learning Framework for Malevolent Dialogue Response Detection

3 Method

3.2 Overall Framework

3.3 Utterance Encoder

3.4 Malevolence Shift Detection

3.5 Hierarchy-Aware Label Encoder

3.6 Malevolence Detection in Dialogues

3.7 Multi-task Learning

4.1 Datasets and Evaluation Metrics

4.2 Compared Baselines

4.3 Implementation Details

4.4 Main Results

4.6 Analysis of Malevolence Shift Detection

4.7 Case Study

4.8 Analysis of LLMs

References.

Two-Stage Knowledge Graph Completion Based on Semantic Features and High-Order Structural Features

2 Preliminary

2.1 Knowledge Graph

2.2 Knowledge Graph Completion

2.3 Dynamic Graph Attention Variant GATv2

3.1 Structural Local Contexts Aggregation

3.2 High-Order Connected Contexts Aggregation

3.3 Decoder

4 Experiment

4.1 Datasets and Metrics

4.2 Results and Analysis

5 Conclusions

Instance-Ambiguity Weighting for Multi-label Recognition with Limited Annotations

2.1 Multi-label Recognition with Full Annotations

2.2 Multi-label Recognition with Limited Annotations

3.2 Ambiguity-Aware Instance Weighting

3.3 Total Training Loss

4.1 Experiment Settings

4.2 Results

4.3 Ablation Studies

4.4 Model Analysis

Chaotic Neural Oscillators with Deep Graph Neural Network for Node Classification

1.1 Node Classification

1.2 Graph Neural Network (GNN)

1.3 Chaotic Neural Oscillator (CNO)

2 Methodology

3 Experiment

3.1 Datasets

3.2 Settings and Baselines

3.3 Results

3.4 Ablation Study

4 Conclusion

Adversarial Learning of Group and Individual Fair Representations

3 Preliminaries

4 Methodology

4.1 Problem Statement

4.2 Model

4.3 Theoretical Properties of Loss Functions

4.4 Optimization with Focal Loss

5 Experiments and Analysis

Class Ratio and Its Implications for Reproducibility and Performance in Record Linkage

2.1 Data Partitioning

2.2 Classification and Evaluation

3 Experimental Study

3.2 Results.

4 Discussion and Recommendations

5 Conclusions and Future Work

Clustering

Clustering-Friendly Representation Learning for Enhancing Salient Features

3 Proposed Method

3.1 The Framework of cIDFD

3.2 Loss for Background Feature Extraction

3.3 Loss for Target Feature Extraction

3.4 Two-Stage Learning

4.2 Comparison with Conventional Methods

4.3 Representation Distribution

4.4 Similarity Distribution

ImMC-CSFL: Imbalanced Multi-view Clustering Algorithm Based on Common-Specific Feature Learning

1.1 Motivation

1.2 Contribution

3 Imbalanced Multi-view Clustering Algorithm Based on Common-Specific Feature Learning (ImMC-CSFL)

3.1 Deep Feature Extraction Module

3.2 Common Information Learning Module

3.3 Specific Information Learning Module

3.4 Deep Multi-view Clustering Based on Common-Specific Feature Learning

4.1 Experimental Datasets and Evaluation Criteria

4.2 Methods of Comparison

4.3 Experimental Results

5 Summary

Multivariate Beta Mixture Model: Probabilistic Clustering with Flexible Cluster Shapes

2 Multivariate Beta Mixture Model

2.1 Multivariate Beta Distribution

2.2 MBMM Density Function and Generative Process

2.3 Parameter Learning for the MBMM

2.4 The Similarity Score Between Data Points

3.1 Comparisons on the Synthetic Datasets

3.2 Comparison on the Real Datasets

3.3 Distance Between Data Points

5 Discussion

AutoClues: Exploring Clustering Pipelines via AutoML and Diversification

3 AutoClues

3.1 Formalization

3.2 Implementation.

4 Benchmark Generation and Empirical Evaluation.

ISBN:

9789819722426

981972242X

OCLC:

1432236428