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 III / 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 ; 14647

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 (448 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 III

Interpretability and Explainability

Neural Additive and Basis Models with Feature Selection and Interactions

1 Introduction

2 Generalized Additive Models (GAMs)

2.1 Neural Additive Model (NAM)

2.2 Neural Basis Model (NBM)

3 NAM and NBM with Feature Selection

3.1 Motivation

3.2 Model Architecture

3.3 Implementation Remark

4 Discussion of Model Complexities

5 Experiments

5.1 Experimental Settings

5.2 Baselines

5.3 Results

6 Conclusion

References

Random Mask Perturbation Based Explainable Method of Graph Neural Networks

2 Related Work

3 Problem Statement

4 Explainable Method

4.1 Node Importance Based on Fidelity

4.2 Explanation Sparsity

5.1 Experimental Setup

5.2 Quantitative Experiments

5.3 Ablation Study

5.4 Use Case

RouteExplainer: An Explanation Framework for Vehicle Routing Problem

3 Proposed Framework: RouteExplainer

3.1 Many-to-Many Edge Classifier

3.2 Counterfactual Explanation for VRP

4 Experiments

4.1 Quantitative Evaluation of the Edge Classifier

4.2 Qualitative Evaluation of Generated Explanations

5 Conclusion and Future Work

On the Efficient Explanation of Outlier Detection Ensembles Through Shapley Values

3 Outlier Detection Ensembles

4 The bagged Shapley Values

5 Theoretical Guarantees for the Approximation

6 Experiments

6.1 Quality of the Approximation

6.2 Effectiveness

6.3 Scalability

7 Conclusions

Interpreting Pretrained Language Models via Concept Bottlenecks

2.1 Interpreting Pretrained Language Models.

2.2 Learning from Noisy Labels

3 Enable Concept Bottlenecks for PLMs

3.1 Problem Setup

4 C3M: A General Framework for Learning CBE-PLMs

4.1 ChatGPT-Guided Concept Augmentation

4.2 Learning from Noisy Concept Labels

A Definitions of Training Strategies

B Details of the Manual Concept Annotation for the IMDB Dataset

C Implementation Detail

D Parameters and Notations

E Statistics of Data Splits

F Statistics of Concepts in Transformed Datasets

G More Results on Explainable Predictions

H A Case Study on Test-Time Intervention

I Examples of Querying ChatGPT

Unmasking Dementia Detection by Masking Input Gradients: A JSM Approach to Model Interpretability and Precision

3 Methods

3.1 Jacobian Saliency Map (JSM)

3.2 Jacobian-Augmented Loss Function (JAL)

4.1 Dataset

4.2 Preprocessing

4.3 Multimodal Classification

4.4 Performance Evaluation

5 Conclusion

Towards Nonparametric Topological Layers in Neural Networks

1.1 Background

1.2 Motivation and Challenges

1.3 Contributions

2 Preliminaries and Related Work

2.1 Basics of Topology

2.2 Topological Neural Network

2.3 Functional Spaces for Machine Learning

3 Methodology

4 Evaluation

4.1 Experimental Setup

4.2 Implementation

4.3 Overall Performance

4.4 Learning Rate

4.5 Temporal-Spatial Correlation

Online, Streaming, Distributed Algorithms

Streaming Fair k-Center Clustering over Massive Dataset with Performance Guarantee

1.1 Problem Statement

1.2 Related Work

1.3 Our Contribution

2 A Two-Pass Algorithm with Approximation Ratio 3

2.1 The -Independent Center Set

2.2 The Two-Pass Streaming Algorithm.

3 The Streaming Algorithm with an Approximation Ratio 7

3.1 The Streaming Algorithm for Constructing 1 and 2

3.2 Post-streaming Construction of Center Set C from 12

4 Experimental Results

4.1 Experimental Setting

4.2 Experimental Analysis

Projection-Free Bandit Convex Optimization over Strongly Convex Sets

2.1 Projection-Free OCO Algorithms

2.2 Bandit Convex Optimization

3 Main Results

3.1 Preliminaries

3.2 Our Proposed Algorithm

3.3 Theoretical Guarantees

4.1 Problem Settings

4.2 Experimental Results

Adaptive Prediction Interval for Data Stream Regression

3 Background

4 Adaptive Prediction Interval(AdaPI)

5 Experiments and Results

5.1 Comparison to Interval Forecast

5.2 Comparison Between MVE and AdaPI

6 Conclusions

Probabilistic Guarantees of Stochastic Recursive Gradient in Non-convex Finite Sum Problems

1.1 Related Works

1.2 Our Contributions

1.3 Notation

2 Prob-SARAH Algorithm

3 Theoretical Results

3.1 Technical Assumptions

3.2 Main Results on Complexity

3.3 Proof Sketch

4 Numerical Experiments

4.1 Logistic Regression with Non-convex Regularization

4.2 Two-Layer Neural Network

Rethinking Personalized Federated Learning with Clustering-Based Dynamic Graph Propagation

3.1 Model Overview

3.2 Client Model Clustering

3.3 Dynamic Weighted Graph Construction

3.4 Knowledge Propagation and Aggregation

3.5 Precise Personalized Model Distribution

4 Experiment

4.1 Experiment Setup

4.2 Performance Evaluation

4.3 Ablation Study

4.4 Case Study

4.5 Hyperparameter Study.

5 Conclusion

Unveiling Backdoor Risks Brought by Foundation Models in Heterogeneous Federated Learning

3.1 Threat Model

3.2 FMs Empowered Backdoor Attacks to HFL

4.3 Homogeneous Setting Evaluation

4.4 Case Study: Attack Effectiveness v.s. Public Data Utilization Ratio

4.5 Hyper-Parameter Study: ASR v.s. Poisoning Ratio

Combating Quality Distortion in Federated Learning with Collaborative Data Selection

2 Related Works

3 Proposal

3.2 Design Principle

3.3 Collaborative Sample Selection (CSS)

4.1 Datasets and Experimental Settings

Probabilistic Models and Statistical Inference

Neural Marked Hawkes Process for Limit Order Book Modeling

2 Background

3 Neural Marked Hawkes Process

4 Related Work

How Large Corpora Sizes Influence the Distribution of Low Frequency Text n-grams

2 Background and Related Work

3 The Model

4 Results

4.1 The Corpora Collection

4.2 The Range of k Values for W(k,C

L,n) Prediction

4.3 The Assessment Criteria and Parameter Estimation

4.4 Comparison with Other Models

4.5 Obtained Results

4.6 The Predictions with Growing Corpus Size

5 Conclusions

Meta-Reinforcement Learning Algorithm Based on Reward and Dynamic Inference

2.1 Meta-Reinforcement Learning

2.2 Context-Based Meta-Reinforcement Learning

2.3 Parametric Task Distributions

4 Method

4.1 Reward and Dynamics Inference.

4.2 Meta-Reinforcement Learning Algorithm Based on Reward and Dynamics Inference Encoders

5 Experiment

5.1 Common MuJoCo Environments

5.2 Cartesian Product Combinations of Tasks with Different Goals and Dynamics

6 Discussion

Security and Privacy

SecureBoost+: Large Scale and High-Performance Vertical Federated Gradient Boosting Decision Tree

2 Preliminaries

2.1 Gradient Boosting Decision Tree

2.2 Paillier Homomorphic Encryption

2.3 SecureBoost

2.4 Performance Bottlenecks Analysis for SecureBoost

3 Proposed SecureBoost+ Framework

3.1 Ciphertext Operation Optimization

3.2 Training Mechanism Optimization

4.1 Setup

4.2 Ciphertext Operation Optimization Evaluation

4.3 Training Mechanism Optimization Evaluation

Construct a Secure CNN Against Gradient Inversion Attack

2 Preliminary

2.1 Federated Learning

2.2 Gradient Inversion Attack

2.3 Recursive Gradient Attack on Privacy (R-GAP)

3 Secure Convolutional Neural Networks

4.1 Quantitative Results

4.2 Quantitative Results

5 Related Work

6 Limitation and Conclusion

Backdoor Attack Against One-Class Sequential Anomaly Detection Models

3 Preliminaries

3.1 Deep One-Class Sequential Anomaly Detection

3.2 Mutual Information Maximization

4 Methodology

4.1 Threat Model

4.2 The Proposed Attack

4.3 Post-deployment Attack

5.2 Experimental Results

Semi-supervised and Unsupervised Learning

DALLMi: Domain Adaption for LLM-Based Multi-label Classifier

2 Language Model and Domain Adaptation

3 DALLMi

References.

Contrastive Learning for Unsupervised Sentence Embedding with False Negative Calibration.

Notes:

Description based on publisher supplied metadata and other sources.

ISBN:

9789819722594

9819722594

OCLC:

1432236142