11343nam 22005533 450 991087806970332120240802080341.09783031670084(electronic bk.)9783031670077(MiAaPQ)EBC31572345(Au-PeEL)EBL31572345(CKB)33566329700041(EXLCZ)993356632970004120240802d2024 uy 0engurcnu||||||||txtrdacontentcrdamediacrrdacarrierRecent Trends in Analysis of Images, Social Networks and Texts 11th International Conference, AIST 2023, Yerevan, Armenia, September 28-30, Revised Selected Papers1st ed.Cham :Springer International Publishing AG,2024.©2024.1 online resource (332 pages)Communications in Computer and Information Science Series ;v.1905Print version: Ignatov, Dmitry I. Recent Trends in Analysis of Images, Social Networks and Texts Cham : Springer International Publishing AG,c2024 9783031670077 Intro -- Preface -- Organisation -- Abstracts of Posters -- Time Series Analysis and Complexity Theory -- Theoretical Guarantees for Neural Control Variates in MCMC -- Guaranteed Optimal Generative Modeling with Maximum Deviation from the Empirical Distribution -- Analyzing Local Representative Power of Vision Transformers -- Towards Robust Entity Matching: Cases Studies for the Pharmaceutical Databases -- Evaluating Naturalness of the Text Using Quantitative Features -- Smaller3D: Smaller Models for 3D Semantic Segmentation Using Minkowski Engine and Knowledge Distillation Methods -- Machine Learning Pipelines Synthesis with Large Language Models -- Assessing Proficiency Levels of Russian Language Learners: Evaluation of Lexical Skills -- Genetic Variability and Disease Resistance Analysis of Grapevine from Armenia -- Raw Operational Data Visualization -- Using Physics-Informed Neural Networks to Predict Spatiotemporal Dynamical Processes Driven by Complex Ginzburg-Landau Equation -- Contents -- Natural Language Processing -- Determination of the Number of Topics Intrinsically: Is It Possible? -- 1 Introduction -- 2 Related Work -- 3 Intrinsic Quality Metrics -- 4 Methodology -- 4.1 Topic Models Studied -- 4.2 Corpora Used -- 5 Results and Discussion -- 5.1 Common Issues Encountered -- 5.2 Properties of the Studied Metrics -- 6 Conclusion -- References -- Sentence Difficulty in Three Languages: Russian Dataset Compared to Italian and English -- 1 Introduction -- 2 Related Work -- 2.1 Datasets and Models -- 3 Datasets -- 3.1 Sentence Complexity Dataset for Russian -- 3.2 Comparing Russian Data to English, Italian and Arabic -- 4 Modeling Sentence Complexity -- 4.1 Decision Trees, SVMs, and Linear Regression -- 4.2 Graph Neural Networks -- 4.3 Pre-trained Language Models -- 5 Results and Discussion -- 6 Conclusion -- References.Machine Translation Models Stand Strong in the Face of Adversarial Attacks -- 1 Introduction -- 2 Related Work -- 3 Methods -- 3.1 General Description of a Machine Translation Model -- 3.2 Gradient Machine Translation Attack -- 3.3 BLEUER Attack -- 3.4 MBART Attack -- 3.5 Synthetic Attacks -- 4 Experiments -- 4.1 Metrics -- 4.2 Baselines -- 4.3 Attack Examples -- 4.4 Experiment Setup -- 4.5 Main Results -- 4.6 Performance with Respect to Different Metrics -- 5 Conclusion -- References -- Whether Large Language Models Learn at the Inference Stage? Effects of Active Learning and Labelling with LLMs on their Reasoning -- 1 Introduction -- 2 Related Works -- 3 Methodology -- 4 Experiment Results -- 5 Conclusion -- 6 Future Directions -- References -- Machine Translation for Russian-Khakas Language Pair: Translation Results in Low-Resource Setting -- 1 Introduction -- 2 Related Work -- 3 Data -- 3.1 Choosing the Language for Parent Model -- 3.2 Data for Child Model -- 4 Methods -- 5 Experiments -- 6 Results -- 6.1 Metrics -- 6.2 Case and Error Analysis -- 7 Conclusion and Future Work -- References -- Automatic Aspect Extraction from Scientific Texts -- 1 Introduction -- 2 Related Work -- 2.1 Background -- 2.2 Datasets -- 2.3 Methods -- 3 Dataset Creation -- 4 Automatic Aspect Extraction -- 4.1 Model -- 5 Experiments and Results -- 5.1 Evaluation Method -- 5.2 Experiments with Models -- 5.3 Cross Domain Experiments -- 5.4 Analysis of Automatic Annotation -- 6 Discussion -- 7 Conclusion -- References -- User Review Summarization in Russian -- 1 Introduction -- 2 Related Works -- 3 Data -- 4 Methodology -- 4.1 Abstractive Methods -- 4.2 Extractive Methods -- 5 Experiments -- 5.1 Metrics -- 5.2 Data Preprocessing -- 5.3 Seed Words Extraction -- 5.4 Review Summarization -- 5.5 Manual Analysis -- 6 Conclusion -- References.Tuning-Free Discriminative Nearest Neighbor Few-Shot Intent Detection via Consecutive Knowledge Transfer -- 1 Introduction -- 2 Background -- 2.1 OOS Detection -- 3 Proposed Method -- 3.1 Pre-training -- 3.2 Paraphrases Augmentation -- 4 Experimental Settings -- 4.1 Evaluation Datasets -- 4.2 Evaluation Metrics -- 4.3 Model Configurations -- 5 Experimental Results -- 5.1 Pre-training Results -- 5.2 Comparison with Existing Solutions -- 6 Conclusion -- A Appendix -- References -- Prompt-Tuning for Targeted Sentiment Analysis in Russian -- 1 Introduction -- 2 Related Work -- 3 Methods -- 4 Dataset -- 4.1 Dataset for the RuSentNE-2023 Competition -- 4.2 Evaluation Metrics -- 5 Experiments -- 5.1 Implementation Details -- 5.2 Manual Prompts -- 5.3 Prompt-Tuning with Mixed Template -- 5.4 Prompt-Tuning with Rules -- 5.5 Knowledgeable Verbalizer -- 6 Results and Discussion -- 6.1 Cross-validation -- 6.2 RuSentNE-2023 Post-evaluation -- 7 Conclusion -- References -- On the Way to Controllable Text Summarization in Russian -- 1 Introduction -- 2 Related Work -- 2.1 Extractive Summarization -- 2.2 Abstractive Summarization -- 2.3 Summarization in Russian -- 3 HydraSum Method -- 4 Experimental Setup -- 4.1 Models -- 4.2 Data -- 4.3 Metrics -- 5 Results -- 6 Conclusion -- References -- Document-Level Relation Extraction in Russian -- 1 Introduction -- 2 Related Work -- 3 Dataset -- 4 Approaches -- 4.1 BiLSTM and Context-Aware LSTM -- 4.2 DocuNET -- 4.3 DREEAM -- 5 Experiments -- 5.1 Data Processing -- 5.2 DREEAM and DocuNET Model -- 6 Conclusion -- References -- The Battle of Information Representations: Comparing Sentiment and Semantic Features for Forecasting Market Trends -- 1 Introduction -- 2 Contribution -- 3 Related Work -- 4 Methods -- 4.1 Data Preparation -- 4.2 Predictive Model Parameters -- 4.3 Loss Function -- 5 Results -- 5.1 Daily Return Analysis.5.2 Relation Between Tweet Sentiments and Stock Market -- 5.3 Connection of Sentiments and Embeddings -- 5.4 Price Prediction Results and Sensitivity Analysis -- 6 Conclusions -- 7 Statement on Computational Resources and Environmental Impact -- References -- Computer Vision -- Interactive Image Segmentation with Superpixel Propagation -- 1 Introduction -- 2 Interactive Segmentation Method -- 2.1 Overview of the Method -- 2.2 Fast Dashing Method and Arrival Time Redistribution -- 3 Experiments and Results -- 3.1 Datasets -- 3.2 Quantitative and Qualitative Comparison -- 3.3 Comparison with Naive Approach -- 4 Conclusions -- 4.1 Limitations -- References -- Gesture Recognition on Video Data -- 1 Introduction -- 2 Related Work -- 2.1 Datasets -- 2.2 Current SOTA Methods -- 3 Experiments and Results -- 3.1 Hand Keypoints Approaches -- 3.2 Image-Based Approaches -- 4 Proposed Approach -- 4.1 Training -- 4.2 Inference -- 5 Conclusion -- 5.1 Future Work -- References -- Semantic-Aware GAN Manipulations for Human Face Editing -- 1 Introduction -- 2 Related Work -- 3 Methods -- 3.1 Comparable Methods -- 3.2 Latent Space Manipulation Improvements -- 3.3 Experiment Settings -- 4 Results -- 4.1 Manipulations in GAN Latent Space -- 4.2 Manipulations with Real Images -- 4.3 Improving Quality Manipulation with Translation in Extended Latent Space -- 5 Conclusion -- References -- Learning Facial Expression Recognition In-the-Wild from Synthetic Data Based on an Ensemble of Lightweight Neural Networks -- 1 Introduction -- 2 Research Methods -- 2.1 Dataset -- 2.2 Data Preprocessing -- 2.3 Ensemble Approach for Facial Expressions Recognition -- 2.4 End-to-End Video Analytics Application -- 3 Experimental Results -- 4 Conclusion -- References -- Acne Recognition: Training Models with Experts -- 1 Introduction -- 2 Related Work -- 3 Data and Problem Statement.4 Experiments and Results -- 5 Conclusion and Future Work -- References -- Data Analysis and Machine Learning -- Application of Multimodal Machine Learning for Image Recommendation Systems -- 1 Introduction -- 1.1 Recommender System -- 1.2 Multimodality -- 2 Related Work -- 3 Model -- 3.1 Dataset Preparation -- 3.2 Model -- 4 Experiment Settings -- 4.1 Experimental Datasets -- 4.2 Experiments with Yandex Dataset -- 4.3 Experiments with Flickr Image Dataset -- 5 Results -- 6 Conclusion -- References -- Date-Driven Approach for Identifying State of Hemodialysis Fistulas: Entropy-Complexity and Formal Concept Analysis -- 1 Introduction -- 2 Related Works -- 3 Materials and Methods -- 4 Results -- 5 Conclusions and Future Directions -- References -- Network Analysis -- Approximate Density Computation for OA-Biclustering -- 1 Introduction -- 2 Basic Notions for Two-Mode Network Analysis -- 2.1 Basic Definition of FCA -- 2.2 Basic Definition of Bicluster and Its Properties -- 3 Search Algorithm for Biclusters, Complexity Assessment of Algorithm -- 4 Approximate Density Estimation for the Large Biclusters -- 5 Search for the Optimal Set of Biclusters in the Context Coverage Problem -- 5.1 A Greedy Approach to Cover a Context -- 6 Experiments -- 6.1 Description of Data -- 6.2 Approximation of Formal Concepts by Biclusters -- 6.3 Approximate Density Estimation -- 6.4 Genome Association Study -- 6.5 Community Detection Cases -- 7 Conclusion -- References -- Theoretical Machine Learning and Optimization -- Distributed Bayesian Coresets -- 1 Introduction -- 2 Background -- 2.1 Bayesian Coresets -- 2.2 Sensitivity -- 2.3 Related Work -- 3 Distributed Setting -- 4 Experiments -- 4.1 Multivariate Gaussian -- 4.2 Gaussian Mixture -- 4.3 Classification -- 4.4 Number of Processors -- 5 Conclusion -- References -- Demo Paper.Development of a Visualization Tool for the Occurrence of Life Events on the Demographic Lexis Grid.Communications in Computer and Information Science SeriesIgnatov Dmitry I1733492Khachay Michael1369710Kutuzov Andrey1733493Madoyan Habet1733494Makarov Ilya1733495Nikishina Irina1733496Panchenko Alexander1733497Panov Maxim1733498M. Pardalos Panos1751098Savchenko Andrey V756076MiAaPQMiAaPQMiAaPQ9910878069703321Recent Trends in Analysis of Images, Social Networks and Texts4185939UNINA