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 II / edited by De-Nian Yang, Xing Xie, Vincent S. Tseng, Jian Pei, Jen-Wei Huang, Jerry Chun-Wei Lin.

Format:

Book

Contributor:

Yang, De-Nian, editor.

Series:

Lecture Notes in Artificial Intelligence, 2945-9141 ; 14646

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 (XXXIV, 459 p. 145 illus., 138 illus. in color.)

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 II

Deep Learning

AdaPQ: Adaptive Exploration Product Quantization with Adversary-Aware Block Size Selection Toward Compression Efficiency

1 Introduction

2 Related Works

3 Preliminary

4 Methodology

4.1 Adaptive Exploration Quantization

4.2 Adversary-Aware Block Size Selection

5 Experiments

6 Conclusion

References

Ranking Enhanced Supervised Contrastive Learning for Regression

2 Related Work

3 Preliminaries

4.1 Motivation

4.2 Ranking Enhanced Supervised Contrastive Learning (RESupCon)

5.1 Datasets

5.2 Baselines and Settings

5.3 Overall Performance

5.4 Comparison on Spearman's Rank Correlation Coefficients

5.5 Parameter Study and Loss Curve

Treatment Effect Estimation Under Unknown Interference

4 Proposed Method: Treatment Effect Estimation Under Unknown Interference

4.1 Covariate Representation Learner

4.2 Graph Structure Learner

4.3 Aggregation Function

4.4 Outcome Predictors and ITE Estimators

5.1 Experiment Settings

5.2 Results

A Identifiability of the Expectation of Potential Outcomes

B HSIC

C Implementation Details

D Ablation Experiments

A New Loss for Image Retrieval: Class Anchor Margin

3 Method

4 Experiments

4.1 Datasets

4.2 Experimental Setup

4.3 Results

5 Conclusion

Personalized EDM Subject Generation via Co-factored User-Subject Embedding

3 Proposed Model

3.1 Retrieve and Re-rank

3.2 Variational Encoder and Bi-directional Selective Encoder.

3.3 User-Subject Co-factor System

3.4 User-Based Decoder

4 Experimental Results

4.1 Quantitative Results

4.2 Effect of Template

5 Conclusions and Future Work

Spatial-Temporal Bipartite Graph Attention Network for Traffic Forecasting

3 Definitions and Problem Statement

3.1 Definitions

3.2 Problem Statement

4.1 Data Inputs and Data Preprocessing

4.2 Encoder Decoder Architecture

4.3 Bipartite Graph Attention Layer

4.4 Heterogeneous Cross Attention Layers

5.1 Experiment Setup

5.2 Comparison of Performance

5.3 Ablation Study

6 Conclusion and Future Works

CMed-GPT: Prompt Tuning for Entity-Aware Chinese Medical Dialogue Generation

3 Datasets

4 Method

4.1 Pre-training Model

4.2 Medical Dialogue Generation Model

5.1 Experimental Setting

5.2 Experimental Results

MvRNA: A New Multi-view Deep Neural Network for Predicting Parkinson's Disease

3 Methods

3.1 Data Representation Based on Multiple Views

3.2 ResNet18 with BWH

3.3 Channel Attention Implemented Using SENet

4 Experiments and Results

4.1 Dataset

4.2 Experimental Settings

4.3 Experimental Results and Analysis

4.4 Ablation Experiment

Path-Aware Cross-Attention Network for Question Answering

3 Task Definition

4.1 Text Encoder and Path Encoder

4.2 Path-Aware Cross-Attention

4.3 Self-learning Based Path Scoring Method

4.4 Learning and Inference

5 Experiment

5.1 Dataset

5.2 Baseline Models

5.3 Main Result

6 Analysis

6.1 Ablation Studies

6.2 Model Interpretability.

6.3 Quantitative Analisis

7 Conclusion

StyleAutoEncoder for Manipulating Image Attributes Using Pre-trained StyleGAN

3 Methodology

3.1 Preliminaries

3.2 StyleAutoEncoder

3.3 Discussion

4.1 Evaluation Metrics

4.2 Models Implementation

4.3 Manipulation of Facial Features

4.4 Evaluation on Animal Faces

SEE: Spherical Embedding Expansion for Improving Deep Metric Learning

3.1 Preliminary

3.2 Spherical Embedding Expansion

4.1 Experiment Setting

4.2 Quantitative Results

4.3 Ablation Studies

Multi-modal Recurrent Graph Neural Networks for Spatiotemporal Forecasting

3.1 Problem Formulation

3.2 Model Design

4.1 Model Baselines

4.2 Primary Results

4.3 Ablation Study

5 Conclusions

7 Appendix

Layer-Wise Sparse Training of Transformer via Convolutional Flood Filling

2 Background and Related Work

2.1 Transformer

2.2 Related Work on Sparse Attention

3 Motivation: Analysis of Sparse Patterns in MHA

4 SPION: Layer-Wise Sparse Attention in Transformer

4.1 Overview of SPION

4.2 Sparsity Pattern Generation with Convolutional Flood Fill Algorithm

5 Experimental Evaluation

5.1 Performance Evaluation

5.2 Computational Complexity Analysis

Towards Cost-Efficient Federated Multi-agent RL with Learnable Aggregation

2 Preliminary

3 Federated MARL with Learnable Aggregation

4 Convergence Analysis

6 Related Work

References.

LongStory: Coherent, Complete and Length Controlled Long Story Generation

2.1 Neural Story Generation

2.2 Recursive Models

2.3 Autometic Metrics

3.1 Task Description

3.2 Long and Short Term Contexts Weight Calibrator(CWC)

3.3 Long Story Structural Positions (LSP)

3.4 Base Pretrained Model

4.1 Experiments Set-Up

4.2 Experimental Results

4.3 Further Analysis

Relation-Aware Label Smoothing for Self-KD

3 Our Approach

3.1 RAS-KD

5 Ablation Study

Bi-CryptoNets: Leveraging Different-Level Privacy for Encrypted Inference

2 Relevant Work

3 Our Bi-CryptoNets

3.1 The Bi-branch of Neural Network

3.2 The Unidirectional Connections

3.3 The Feature Integration

4 Knowledge Distillation for Bi-CryptoNets

Enhancing YOLOv7 for Plant Organs Detection Using Attention-Gate Mechanism

2.1 Attention-Gate Mechanism

3 YOLOv7 with Attention-Gate Mechanism

4.1 Experiment Materials

4.2 Evaluation Metrics

4.3 Experimental Results

On Dark Knowledge for Distilling Generators

3 Theoretical Analysis of Dark Knowledge in Distilling the Generator

3.1 Dark Knowledge of Generators

3.2 Distillation Empirical Risk

3.3 Generalization of the Student Generator

3.4 Impact of Probability Approximation

4 DKtill: Extracting Dark Knowledge for Training Student Generator

4.1 Extracting from Probabilistic Generators

4.2 Extracting from Non-probabilistic Generators

5 Empirical Illustration

5.1 Setting.

5.2 Distilling Probabilistic Generators

5.3 Distilling Non-probabilistic Generators

5.4 Small Generators Through DKtill

RPH-PGD: Randomly Projected Hessian for Perturbed Gradient Descent

2.1 Notation

2.2 Methods to Escape from Saddle Points

2.3 Perturbed Gradient Descent

3 Algorithms

3.1 Randomly Projected Hessian

3.2 Shifted Randomly Projected Hessian

3.3 RPH-PGD

5 Conclusion and Future Work

Transformer based Multitask Learning for Image Captioning and Object Detection

3 Proposed Method

3.1 Objective Function

4 Experimental Setup

5 Results

5.1 Comparison and Analysis

5.2 Ablation Studies

Communicative and Cooperative Learning for Multi-agent Indoor Navigation

3 Cooperative Indoor Navigation Task

3.1 Task Definition

3.2 Multi-agent Indoor Navigation Environment

3.3 Data Collection

4 Cooperative Indoor Navigation Models

4.1 Preliminaries

4.2 Framework

5.1 Benchmarking CIN with MARL Models

5.2 Implementation Details

5.3 Evaluation Metrics

5.4 Quantitative and Qualitative Results

Enhancing Continuous Domain Adaptation with Multi-path Transfer Curriculum

2 Methodology

2.1 Preliminary

2.2 Method Framework

2.3 Wasserstein-Based Transfer Curriculum

2.4 Multi-path Optimal Transport

3 Experimental Results

3.1 Datasets and Experimental Configurations

3.2 Analysis of Wasserstein-Based Transfer Curriculum

3.3 Adaptation Comparison Results

3.4 Ablation Study

4 Conclusion

Graphs and Networks.

Enhancing Network Role Modeling: Introducing Attributed Multiplex Structural Role Embedding for Complex Networks.

Notes:

Includes bibliographical references and index.

ISBN:

9789819722532

9819722535

OCLC:

1492996767