Analysis of images, social networks and texts : 10th international conference, AIST 2021, Tbilisi, Georgia, December 16-18, 2021, revised selected papers / / Evgeny Burnaev [fourteen others], (editors)

• Pubblicazione: Cham, Switzerland : , : Springer, , [2022]; ©2022
• Descrizione fisica: 1 online resource (358 pages)
• Disciplina: 006.312
• Soggetto topico: Data mining; Image analysis; Natural language processing (Computer science)
• Persona (resp. second.): BurnaevEvgeny
• Nota di bibliografia: Includes bibliographical references and index.
• Nota di contenuto: Intro -- Preface -- Organization -- Abstracts of Invited Talks -- Is It Time to Move Beyond Sentence Classification? -- Drug and Disease Interpretation Learning with Biomedical Entity Representation Transformer -- Contents -- Invited Papers -- On Georgian Text Processing Toolkit Development -- 1 Introduction -- 2 Problems -- 2.1 Too Many Synonyms -- 2.2 Polysemy -- 3 Main Tasks for Toolkit Development -- 4 Conclusion -- References -- Taxonomy Enrichment with Text and Graph Vector Representation -- 1 Introduction -- 2 Related Work -- 3 Base Methods -- 3.1 Baseline -- 3.2 DWRank -- 3.3 Word Representations for DWRank -- 4 DWRank-Meta -- 5 Evaluation -- 5.1 Dataset -- 5.2 Evaluation Metrics -- 5.3 Results -- 6 Conclusions -- References -- Natural Language Processing -- Near-Zero-Shot Suggestion Mining with a Little Help from WordNet -- 1 Introduction -- 2 Related Work -- 3 Data -- 4 Approach -- 5 Results -- 6 Conclusion -- References -- Selection of Pseudo-Annotated Data for Adverse Drug Reaction Classification Across Drug Groups -- 1 Introduction -- 2 Related Work -- 3 Datasets and Models -- 4 Experiments -- 5 Conclusion -- References -- Building a Combined Morphological Model for Russian Word Forms -- 1 Introduction -- 2 Related Work -- 3 Architecture of Combined Model -- 4 Training and Evaluating the Model -- 5 Conclusions -- References -- SocialBERT - Transformers for Online Social Network Language Modelling -- 1 Introduction -- 2 Related Work -- 2.1 Author as Network Node -- 2.2 Language Modelling -- 2.3 Knowledge Injection -- 3 Proposed Approach -- 3.1 Adjacency Matrix Generation -- 3.2 Correlation Coefficient -- 3.3 Cosine Coefficient -- 3.4 Matrix Factorization -- 3.5 Random Walk -- 3.6 BERT Training -- 4 Experimental Setup -- 4.1 Data Collection -- 4.2 Network Modelling -- 4.3 Language Model Training -- 5 Results -- 6 Conclusion. ; References -- Lexicon-Based Methods vs. BERT for Text Sentiment Analysis -- 1 Introduction -- 2 Lexicon-Based Methods and Tools -- 3 Materials and Methods -- 3.1 Lexicon-Based Methods Adaptation -- 3.2 Sentiment Lexicons -- 3.3 Text Corpora -- 4 Results -- 4.1 Experimental Setup -- 4.2 Results of Experiments -- 5 Discussion -- 6 Conclusion -- References -- Multilingual Embeddings for Clustering Cultural Events -- 1 Introduction -- 2 Related Work -- 3 Data Description -- 4 Experimental Setup and Results -- 5 Human Interpretation -- 6 Conclusion and Future Work -- References -- Jokingbird: Funny Headline Generation for News -- 1 Introduction -- 2 Related Work -- 3 Data -- 4 Methods -- 4.1 Pipeline -- 4.2 Word Selection -- 4.3 Word Substitution -- 4.4 Threshold Selection -- 4.5 Humor Classifier -- 5 Results and Evaluation -- 6 Conclusion -- References -- Learning to Rank with Capsule Neural Networks -- 1 Introduction -- 2 Related Work -- 3 Ranking Model -- 4 Experiments -- 4.1 Datasets -- 4.2 Data Preprocessing -- 4.3 Evaluation -- 4.4 Training Details -- 4.5 Results: ArXiv Triplets Dataset -- 4.6 Results: MIND Dataset -- 4.7 Comparison to Alternative Approaches -- 5 Conclusion -- References -- Building a Bilingual QA-system with ruGPT-3 -- 1 Introduction -- 2 Previous Work -- 3 Data -- 3.1 Data Translation -- 3.2 Data Formatting -- 4 Modeling Bilingual QA -- 4.1 Model Training -- 4.2 Baseline -- 5 Metrics and Evaluation -- 6 Discussion and Impact -- 7 Conclusion -- References -- Sculpting Enhanced Dependencies for Belarusian -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Rules -- 3.2 Cross-Lingual Transfer -- 3.3 Training Belarusian Model -- 4 Data -- 5 Results -- 6 Analysis -- 6.1 Common Mistakes for Russian-Belarusian and Ukrainian-Belarusian Transfer -- 6.2 The Mistakes of Russian-Belarusian Transfer. ; 6.3 The Mistakes of Ukrainian-Belarusian Transfer -- 6.4 The Mistakes of Alignment Method -- 6.5 The Mistakes of the Model -- 6.6 The Mistakes of the Model Trained on Belarusian Data -- 7 Discussion and Concluding Remarks -- References -- Improving Morpheme Segmentation Using BERT Embeddings -- 1 Introduction -- 2 Task Description -- 3 Model Description -- 3.1 Model Architecture -- 3.2 Datasets -- 3.3 Implementation and Training -- 4 Experiments and Results -- 4.1 The Effect of BERT -- 5 Ablation Studies -- 5.1 Embeddings or Tokenization? -- 5.2 BERT or Other Embeddings? -- 6 Discussion -- 7 Related Work -- 8 Conclusion -- 8.1 Preprocessing Details -- 8.2 BERT Models -- 8.3 Convolutional Network Parameters -- 8.4 Ablation Studies Details -- References -- Training Dataset and Dictionary Sizes Matter in BERT Models: The Case of Baltic Languages -- 1 Introduction -- 2 Existing Models for Estonian, Latvian, and Lithuanian -- 3 LitLat BERT and Est-RoBERTa -- 3.1 Datasets and Preprocessing -- 3.2 Architecture and Training -- 4 Evaluation -- 4.1 Evaluation Settings -- 4.2 Results -- 5 Discussion -- 6 Conclusions -- References -- Computer Vision -- Development of a Method for Iris-Based Person Recognition Using Convolutional Neural Networks -- 1 Introduction -- 2 Method -- 2.1 Standard Identification Scheme -- 2.2 Segmentation of Iris Image -- 2.3 Iris Normalization Step -- 2.4 Feature Extraction -- 2.5 Dimension Reduction -- 2.6 Classification Step -- 3 Experiments -- 3.1 Description of the Dataset -- 3.2 Experimental Research Procedure -- 3.3 Results of Experimental Research -- 4 Conclusion -- References -- Data Dimensionality Reduction Technique Based on Preservation of Hellinger Divergence -- 1 Introduction -- 2 Methods -- 2.1 Approximation of Hellinger Divergence by Euclidean Distances in Lower-Dimension Space. ; 2.2 Preserving Hellinger Divergence in Lower-Dimensional Space -- 3 Experiments -- 4 Conclusion -- References -- Group-Level Affect Recognition in Video Using Deviation of Frame Features -- 1 Introduction -- 2 Proposed Approach -- 3 Experimental Results -- 4 Conclusion -- References -- Outfit Recommendation using Graph Neural Networks via Visual Similarity -- 1 Introduction -- 2 Literature Review -- 2.1 Outfit Compatibility Prediction -- 2.2 Graph Embeddings -- 3 Dataset -- 4 Proposed Method -- 4.1 Problem Formulation -- 4.2 Features -- 4.3 Graph Construction -- 4.4 Node Embeddings -- 4.5 Classifier -- 5 Experiments -- 5.1 Outfit Compatibility Prediction -- 5.2 Fill-in-the-Blank Fashion Recommendation -- 6 Discussion -- 7 Conclusion -- References -- Data Analysis and Machine Learning -- Scalable Computation of Prediction Intervals for Neural Networks via Matrix Sketching -- 1 Introduction and Related Work -- 2 Approximate Prediction Intervals for Neural Networks -- 2.1 Prediction Intervals for Neural Networks -- 2.2 Simplification Using SVD -- 2.3 Proposed Approximation -- 2.4 Motivation Behind the Approximation -- 2.5 Hyperparameters -- 3 Experiments -- 3.1 Methods -- 3.2 Metrics and Results -- 3.3 Implementation -- 4 Discussion and Future Work -- 5 Conclusions -- References -- Application of Data Analysis Methods for Optimizing the Multifunctional Service Center Operation -- 1 Introduction -- 2 Approaches of Analyzing, Forecasting and Optimizing the MFC Operation -- 2.1 FMR-XYZ Analysis -- 2.2 Forecasting Methods -- 2.3 Optimization Task -- 3 Product Analytics of the MFC Services -- 4 The MFC's Operation Optimization -- 5 Conclusions -- References -- Depression Detection by Person's Voice -- 1 Introduction -- 2 Related Work -- 2.1 Acoustic Features -- 2.2 Classifiers in Speech Processing -- 3 Experiment Setup -- 4 Results -- 5 Discussion. ; 6 Conclusion -- 7 Future Work -- References -- Social Network Analysis -- Research Papers Recommendation -- 1 Introduction -- 2 Related Work -- 2.1 Text-Based Methods -- 2.2 Structural Graph Embeddings -- 2.3 Text-Attributed Network Embedding -- 2.4 Recommender Systems -- 3 Link Prediction on Citation Network -- 4 Dataset -- 5 Experiments Settings -- 5.1 Negative Edges Generation -- 5.2 Train and Test Splitting -- 5.3 Models Parameters -- 6 Results -- 6.1 Link Prediction Based on Network Embedding -- 6.2 Link Prediction Based on Text-Based Methods -- 6.3 Fusion Approach -- 7 Conclusion and Discussion -- References -- Multimodal Space of Users' Interests and Preferences in Social Networks -- 1 Introduction -- 2 Related Work -- 3 Methods -- 4 Experiments -- 4.1 Group Subscriptions -- 4.2 Music Subscriptions -- 4.3 Multimodal Analysis -- 4.4 Model Validation -- 5 Conclusion and Future Work -- References -- Citation Network Applications in a Scientific Co-authorship Recommender System -- 1 Introduction -- 2 Related Work -- 3 Data -- 4 Methods -- 5 Results -- 6 Conclusion and Outlook -- References -- Theoretical Machine Learning and Optimization -- How Fast Can the Uniform Capacitated Facility Location Problem Be Solved on Path Graphs -- 1 Introduction -- 2 Problem Statement and Basic Notation -- 3 Properties of UCFLP-Path Solutions -- 4 Polynomial-Time Algorithm for the UCFLP-Path -- 5 Comparison Analysis -- 6 Conclusion and Future Work -- References -- On a Weakly Supervised Classification Problem -- 1 Introduction -- 1.1 Related Works and Preliminary Results -- 2 Problem Settings and Notation -- 3 Proposed Weakly Supervised Classification Scheme -- 3.1 Optimization Problem -- 3.2 Low-Rank Matrix Representation -- 3.3 Algorithm -- 4 Practical Application: Recognition of Acute Stroke from CT Scans -- 4.1 Data Preprocessing and Feature Extraction. ; 4.2 Classification with Fully Supervised ML Algorithms.
• Titolo autorizzato: Analysis of images, social networks and texts
• ISBN: 3-031-16500-4
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 996500067203316
• Lo trovi qui: Univ. di Salerno
• Serie: Lecture notes in computer science ; ; Volume 13217.