Machine Learning and Knowledge Discovery in Databases : Research Track / / edited by Danai Koutra [and four others]

• Record Nr.: UNISA996550555903316
• Pubbl/distr/stampa: Cham, Switzerland : , : Springer, Springer Nature Switzerland AG, , [2023]; ©2023
• ISBN: 3-031-43421-8
• Edizione: [First edition.]
• Descrizione fisica: 1 online resource (789 pages)
• Collana: Lecture Notes in Computer Science Series ; ; Volume 14172
• 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 IV -- Natural Language Processing -- Unsupervised Deep Cross-Language Entity Alignment -- 1 Introduction -- 2 Related Work -- 2.1 Supervised Entity Alignment -- 2.2 Semi-supervised Entity Alignment -- 2.3 Unsupervised Entity Alignment -- 3 Proposed Method -- 3.1 Base Symbol Definition -- 3.2 Feature Embedding Module -- 3.3 Alignment Module -- 4 Experiments -- 4.1 Cross-Lingual Dataset -- 4.2 Comparative -- 4.3 Evaluation Indicate and Experiment Setting -- 4.4 Experimental Results -- 5 Ablation Study -- 5.1 Translator and Encoder Analysis -- 5.2 Alignment Module Analysis -- 5.3 Additional Analysis -- 6 Error Mode Analysis -- 7 Conclusion and Future Research -- References -- Corpus-Based Relation Extraction by Identifying and Refining Relation Patterns -- 1 Introduction -- 2 Problem Formulation -- 3 Methodology -- 3.1 Representation of Relation Triple -- 3.2 Initial Weak Supervision Extraction -- 3.3 Weak Supervision Noise Reduction via Clustering -- 3.4 Generalization via Prompt-Tuning -- 4 Experiments -- 4.1 Relation Extraction -- 4.2 Positive and Negative Samples -- 4.3 Cluster Visualization -- 4.4 Ablation Study -- 5 Related Work -- 6 Conclusions ; -- References -- Learning to Play Text-Based Adventure Games with Maximum Entropy Reinforcement Learning -- 1 Introduction -- 2 Related Work -- 3 Problem Setting and Background -- 3.1 SAC for Discrete Action Spaces -- 4 Reward Shaping Method -- 5 Experimental Results -- 5.1 Datasets -- 5.2 Experimental Settings -- 5.3 Results -- 6 Limitations and Future Work -- 7 Conclusion -- References -- SALAS: Supervised Aspect Learning Improves Abstractive Multi-document Summarization Through Aspect Information Loss -- 1 Introduction.; 2 Related Work -- 2.1 Multi-document Summarization -- 2.2 Aspect-Related Text Generation -- 3 Methodology -- 3.1 Task and Framework Formulation -- 3.2 Encoder Probe -- 3.3 Decoder Probe -- 3.4 Aspect-Guided Generator -- 3.5 Training Objective -- 4 Experiments -- 4.1 Datasets -- 4.2 Baselines -- 4.3 Implementation Details -- 4.4 Main Results -- 4.5 Results of Ablation Study -- 4.6 Analysis and Discussion -- 5 Conclusion -- References -- KL Regularized Normalization Framework for Low Resource Tasks -- 1 Introduction -- 2 Related Work -- 3 Theoretical Foundation of KL Regularized Normalization -- 3.1 Preliminaries: Batch Normalization -- 3.2 KL Regularized Batch Normalization -- 4 Experiments -- 4.1 Model Architecture -- 4.2 Comparison Methods -- 4.3 Training Details -- 4.4 Analysis of Increased Expressive Power -- 4.5 Analysis of out of Domain Generalization -- 4.6 Impact of KL-Norm on Overfitting -- 4.7 Analysis of Removal of Irrelevant Features -- 4.8 Analysis of Model Parameters -- 4.9 Ablation Study -- 5 Conclusion -- References -- Improving Autoregressive NLP Tasks via Modular Linearized Attention -- 1 Introduction -- 2 Background and Motivation -- 2.1 Fundamentals of Transformers and Attention -- 2.2 Linear Transformers -- 2.3 cosFormer and Re-weighting Mechanisms -- 2.4 Motivation for Investigation -- 3 Proposed Approach -- 3.1 Modular Linearized Attention -- 3.2 Augmenting CosFormer for Decoder Cross-Attention -- 3.3 Closing the Gap: Target Sequence Length Prediction -- 4 Training and Evaluation Details -- 4.1 Model Configurations and Training Hyperparameters -- 4.2 Evaluation Setup and Metrics -- 5 Results -- 5.1 TTS Training Results for Targeted Ablations -- 5.2 Training and Evaluation Results for Finalized TTS Configurations -- 5.3 en-de NMT and SimulST Training and Evaluation Results -- 6 Conclusion -- References.; Enhancing Table Retrieval with Dual Graph Representations -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Task Definition -- 3.2 Graph Construction -- 3.3 Dual-Graph Representation Learning -- 3.4 Prediction -- 4 Experiments -- 4.1 Datasets -- 4.2 Baselines -- 4.3 Implementation Details -- 4.4 Results -- 4.5 Analyses -- 5 Conclusion -- References -- A Few Good Sentences: Content Selection for Abstractive Text Summarization -- 1 Introduction -- 2 SWORTS Selection -- 2.1 Content Selection Metrics -- 2.2 SWORTS Selection Pipeline -- 2.3 COMET Variants -- 3 Experimental Setup -- 3.1 Self-training -- 3.2 Cross-Training -- 3.3 Zero-Shot Adaptation -- 3.4 Experimental Settings -- 4 Results and Analysis -- 4.1 Self-training -- 4.2 Cross-training -- 4.3 Zero-Shot Adaptation -- 4.4 Limitations and Opportunities with SWORTS Selection -- 5 Conclusion -- References -- Encouraging Sparsity in Neural Topic Modeling with Non-Mean-Field Inference -- 1 Introduction -- 2 Related Work -- 3 Approach -- 3.1 VAE-LDA -- 3.2 SpareNTM -- 3.3 Objective Function -- 3.4 Neural Network Architecture -- 4 Experiment -- 4.1 Datasets -- 4.2 Baseline Methods and Parameter Settings -- 4.3 Evaluation Metric -- 4.4 Experimental Results -- 5 Conclusion -- References -- Neuro/Symbolic Learning -- The Metric is the Message: Benchmarking Challenges for Neural Symbolic Regression -- 1 Introduction -- 2 ; Methods -- 2.1 Post Equation Generation Coefficient Fitting -- 2.2 Benchmarks -- 2.3 Metrics -- 2.4 NSR Methods -- 2.5 Control Equations -- 3 Results -- 3.1 Numeric Metrics -- 3.2 Symbolic Metrics -- 4 Discussion -- 5 Conclusion -- A Appendix -- B Ethics -- References -- Symbolic Regression via Control Variable Genetic Programming -- 1 Introduction -- 2 Preliminaries -- 3 Control Variable Genetic Programming -- 3.1 Control Variable Experiment.; 3.2 Control Variable Genetic Programming -- 3.3 Theoretical Analysis -- 4 Related Work -- 5 Experiments -- 5.1 Experimental Settings -- 5.2 Experimental Analysis -- 6 Conclusion -- References -- Neural Class Expression Synthesis in ALCHIQ(D) -- 1 Introduction -- 2 Related Work -- 3 Background -- 3.1 Description Logics -- 3.2 Refinement Operators -- 3.3 Class Expression Learning -- 3.4 Knowledge Graph Embedding -- 3.5 The Set Transformer Architecture -- 4 Proposed Approach (NCES2) -- 4.1 Preliminaries -- 4.2 Learning Problem -- 4.3 Encoding Positive and Negative Examples -- 4.4 Loss Function -- 4.5 Measuring Performance During Training -- 4.6 Class Expression Synthesis -- 4.7 Model Ensembling -- 5 Evaluation -- 5.1 Experimental Setup -- 5.2 Results and Discussion -- 6 Conclusion and Future Work -- References -- Deep Explainable Relational Reinforcement Learning: A Neuro-Symbolic Approach -- 1 Introduction -- 2 Related Works -- 3 Preliminaries -- 3.1 Logic Programming -- 3.2 Relational Markov Decision Process -- 3.3 Problem Statement -- 4 Proposed Approach -- 4.1 Rule Generation -- 4.2 Inference -- 4.3 Semantic Constraints -- 5 Experiments -- 5.1 Experimental Setup -- 5.2 Results -- 6 Conclusion -- References -- ReOnto: A Neuro-Symbolic Approach for Biomedical Relation Extraction -- 1 Introduction -- 2 Related Work -- 3 Problem Formulation and Approach -- 3.1 Symbolic Module -- 3.2 Encoding Module -- 3.3 Graph Neural Network -- 4 Experimental Setup -- 5 Results -- 6 Ablation Study -- 6.1 Effectiveness of Number of Ontologies -- 6.2 Effectiveness of Number of Hops -- 7 Case Study -- 8 Conclusion and Future Work -- References -- Optimization -- NKFAC: A Fast and Stable KFAC Optimizer for Deep Neural Networks -- 1 Introduction -- 2 Backgrounds and Preliminaries -- 2.1 Matrix Inverse -- 2.2 KFAC Algorithm -- 2.3 About Other Second-Order Optimizers.; 3 Methodology -- 3.1 Motivation -- 3.2 Newton's Iteration of Matrix Inverse -- 3.3 NKFAC -- 3.4 Implementations and AdaNKFAC -- 4 Experiments -- 4.1 Experimental Setup -- 4.2 Results on CIFAR100/10 -- 4.3 Results on ImageNet -- 4.4 Results on COCO -- 5 Conclusion -- References -- Exact Combinatorial Optimization with Temporo-Attentional Graph Neural Networks -- 1 Introduction -- 2 Related Work -- 3 Background -- 4 Methodology -- 5 Experiments -- 6 Conclusion -- 7 Statement of Ethics -- References -- Improved Multi-label Propagation for Small Data with Multi-objective Optimization -- 1 Introduction -- 2 Theoretical Background -- 3 Label Propagation on Absorbing Markov Chain -- 3.1 Computing Transition Probabilities -- 4 Leveraging Label Dependence: Interpolated Label Propagation as a Stacking Method -- 5 Evaluation Measures and Thresholding Strategies -- 6 Finding Compromise Solution for Multiple Evaluation Measures -- 7 Experiments -- 7.1 Experimental Setup -- 7.2 Results -- 8 Conclusions -- References -- Fast Convergence of Random Reshuffling Under Over-Parameterization and the Polyak-Łojasiewicz Condition -- 1 Introduction -- 2 Related Work -- 3 Assumptions -- 4 Contributions -- 5 Convergence Results -- 5.1 Proof Sketch -- 6 Experimental Results -- 6.1 Synthetic Experiments -- 6.2 Binary Classification Using RBF Kernels -- 6.3 Multi-class Classification Using Deep Networks -- 7 Conclusion -- References -- A Scalable Solution for the Extended ; Multi-channel Facility Location Problem -- 1 Introduction -- 1.1 Contributions -- 1.2 Problem Formulation -- 2 Background -- 2.1 Submodularity -- 2.2 Optimal Transport (OT) -- 3 A Scalable Approximate Solution -- 3.1 Fast Value Oracle -- 4 Related Work -- 5 Experiments -- 6 Conclusion -- References -- Online State Exploration: Competitive Worst Case and Learning-Augmented Algorithms -- 1 Introduction.; 1.1 The Online State Exploration Problem.