Doklady Mathematics最新文献_第6页

Towards Foundation Time Series Model: To Synthesize or Not to Synthesize? 面向基础时间序列模型：综合还是不综合？

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700504

A. Temirkhanov, A. Kostromina, O. Tsymboi, K. Kuvshinova, E. Kovtun, D. Simakov

引用次数: 0

Interpreting Transformer-Based Classifiers via Clustering 通过聚类解释基于转换器的分类器

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700401

G. M. Gritsai, A. V. Grabovoy

{"title":"Interpreting Transformer-Based Classifiers via Clustering","authors":"G. M. Gritsai, A. V. Grabovoy","doi":"10.1134/S1064562425700401","DOIUrl":"10.1134/S1064562425700401","url":null,"abstract":"Transformer-based models, especially BERT-like architectures, have become the standard for solving Natural Language Processing (NLP) tasks: text classification, summarisation, question answering. Their high performance is undoubted, however, the key challenge remains interpretability. Understanding the reasons for the decisions of models is critical for increasing confidence in them, detecting bias, and complying with ethical and legal standards. Existing explanation methods focus on identifying individual important tokens or interactions only between adjacent tokens or their pairs, ignoring the global context. This limits their usefulness because such explanations often do not capture the logic of decision making at a level that humans can understand. To bridge this gap, we introduce an approach that translates model predictions into natural language explanations. This algorithm is fitted through clustering of transformer layers, cluster labels are extracted and indices are formed to select close examples. Given examples are fed into the Large Language Models (LLM) to identify key common features in natural language. Frequency analysis of the features in the examples forms the basis of evidence with certain probability. In the case study of detecting machine-generated texts, our approach reveals how classifiers might rely on stylistic cues or structural anomalies.","PeriodicalId":531,"journal":{"name":"Doklady Mathematics","volume":"112 2","pages":"443 - 454"},"PeriodicalIF":0.6,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147560736","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Multi-Aspect Evaluation of Tokenizer Adaptation Methods for Large Language Models on Russian 面向俄语大型语言模型的标记器自适应方法的多角度评价

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700395

G. D. Andriushchenko, M. E. Godunova, V. V. Ivanov, D. S. Kuzmin, A. A. Parinov, A. Yu. Schenikova, E. V. Zhemchuzhina

引用次数: 0

Onboard Instance Segmentation of Railway Infrastructure and Obstacles for Autonomous Train Systems 自主列车系统中铁路基础设施和障碍的车载实例分割

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700413

V. A. Fedorov

{"title":"Onboard Instance Segmentation of Railway Infrastructure and Obstacles for Autonomous Train Systems","authors":"V. A. Fedorov","doi":"10.1134/S1064562425700413","DOIUrl":"10.1134/S1064562425700413","url":null,"abstract":"This paper proposes an enhanced YOLOv11 (You Only Look Once) architecture for onboard instance segmentation of railway infrastructure in autonomous train systems. The solution addresses the critical challenge of maintaining both real-time performance and segmentation accuracy simultaneously. The proposed method incorporates three key innovations: a computationally efficient SimSPPF (Simplified Spatial Pyramid Pooling) component, integrated CBAM (Convolutional Block Attention Module) attention mechanisms, and optimized scaling parameters. A comprehensive evaluation was conducted across 30 distinct model configurations, formed by combining six YOLOv11 variants (nano, small, medium, large, extra-large, and our improved YOLOv11sim) with five input resolutions (640 × 640 to 1920 × 1920). All models were trained and validated on our novel Russian railway infrastructure dataset, containing 20,000 annotated images capturing diverse infrastructure elements and operational scenarios specific to Russian rail networks. According to experimental results, our YOLOv11sim achieves better accuracy-speed compromises, especially at high resolutions, with inference times of 15–30% faster than comparable baselines while retaining 92–96% large-model accuracy. Even in cluttered environments, the enhanced architecture effectively divides a variety of elements, from entire track components to tiny, crucial items like pickets and signal lights, with exact mask boundaries. By combining the performance of embedded hardware with the accuracy required by safety-critical railway operations, these developments directly address the fundamental needs for deployable perception systems in autonomous trains.","PeriodicalId":531,"journal":{"name":"Doklady Mathematics","volume":"112 2","pages":"433 - 442"},"PeriodicalIF":0.6,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147560728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing fMRI Data Decoding with Spatiotemporal Characteristics in Limited Dataset 利用有限数据集的时空特征增强fMRI数据解码

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700383

D. Dorin, A. Grabovoy, V. Strijov

{"title":"Enhancing fMRI Data Decoding with Spatiotemporal Characteristics in Limited Dataset","authors":"D. Dorin, A. Grabovoy, V. Strijov","doi":"10.1134/S1064562425700383","DOIUrl":"10.1134/S1064562425700383","url":null,"abstract":"This study investigates the impact of spatiotemporal characteristics on the quality of the decoding of functional Magnetic Resonance Imaging (fMRI) data. Neural network architectures are limited in handling fMRI data due to the small sample size, high sample variability, and significant computational resources required. We propose a methodology for fMRI decoding with an insufficient dataset. We examine the unique structural features of each subject’s brain. To build a decoding methodology, we propose an algorithm for extracting a unique activity mask from the brain for each subject. This algorithm reduces the spatial dimensionality of fMRI time series through weighting stimulated brain regions using cross-correlation. We developed a classification model for the fMRI time series data from a single subject. It combines two parts: the first part is the brain activity masks extracted for each activity class using the extraction algorithm. The second part is an encoder that employs Riemannian geometry to extract spatiotemporal characteristics. The computational experiment analyzes the proposed methodology on a sample obtained from tomographic examinations of six subjects. An ablation analysis of the proposed classification model shows a significant decrease in quality when any component of the model is missing. Comparisons with neural network-based approaches demonstrate the superior performance of the proposed methodology, especially in scenarios with limited data.","PeriodicalId":531,"journal":{"name":"Doklady Mathematics","volume":"112 2","pages":"465 - 480"},"PeriodicalIF":0.6,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1134/S1064562425700383.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147560743","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Adapting the AI Scientist for Enterprise: Solving Real Business Problems with Autonomous Text-To-SQL Research 使人工智能科学家适应企业：通过自主文本到sql研究解决实际业务问题

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700474

V. Fedorov, D. Lavitskaya, D. Ibragimov, D. Safronov, A. Balles, A. Gribanova, M. Radionov

{"title":"Adapting the AI Scientist for Enterprise: Solving Real Business Problems with Autonomous Text-To-SQL Research","authors":"V. Fedorov, D. Lavitskaya, D. Ibragimov, D. Safronov, A. Balles, A. Gribanova, M. Radionov","doi":"10.1134/S1064562425700474","DOIUrl":"10.1134/S1064562425700474","url":null,"abstract":"One of the grand challenges in artificial intelligence is automating complex, multi-step reasoning tasks that require deep understanding of both natural language and structured data. While large language models have shown promise in code generation and natural language understanding, their ability to autonomously conduct end-to-end research in specialized domains remains limited. We ask a key question: can AI-driven research reliably solve real enterprise problems? This paper presents a fully automated framework for AI-driven research applied to the text-to-SQL problem, enabling frontier language models to independently generate novel ideas, design experiments, implement solutions, and communicate findings. We adapt the AI Scientist [1] – a comprehensive AI agent for autonomous scientific discovery – to the domain of semantic parsing, where it formulates hypotheses about improving text-to-SQL accuracy, writes executable code, runs experiments on benchmark datasets, visualizes performance gains, and produces complete scientific papers summarizing its results. Each full research cycle costs less than $5, making it a scalable and cost-effective approach for rapid innovation. This work moves toward self-improving AI systems in natural language processing, demonstrating robust performance across both proprietary and public benchmarks. Where AI agents not only solve tasks but also advance the state of the art by conducting independent research. Our code and generated papers are open-sourced at https://gitverse.ru/tr1ggers/AIScientist_Text2SQL.git.","PeriodicalId":531,"journal":{"name":"Doklady Mathematics","volume":"112 2","pages":"382 - 391"},"PeriodicalIF":0.6,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1134/S1064562425700474.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147560079","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Mind and Motion Aligned: A Joint Evaluation IsaacSim Benchmark for Task Planning and Low-Level Policies in Mobile Manipulation 思维和运动对齐：移动操作中任务规划和低级策略的联合评估IsaacSim基准

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700450

N. E. Kachaev, A. N. Spiridonov, A. S. Gorodetsk, K. F. Muravyev, N. S. Oskolkov, A. Narendra, V. I. Shakhuro, D. A. Makarov, A. I. Panov, P. D. Fedotova, A. K. Kovalev

{"title":"Mind and Motion Aligned: A Joint Evaluation IsaacSim Benchmark for Task Planning and Low-Level Policies in Mobile Manipulation","authors":"N. E. Kachaev, A. N. Spiridonov, A. S. Gorodetsk, K. F. Muravyev, N. S. Oskolkov, A. Narendra, V. I. Shakhuro, D. A. Makarov, A. I. Panov, P. D. Fedotova, A. K. Kovalev","doi":"10.1134/S1064562425700450","DOIUrl":"10.1134/S1064562425700450","url":null,"abstract":"Benchmarks are crucial for evaluating progress in robotics and embodied AI. However, a significant gap exists between benchmarks designed for high-level language instruction following, which often assume perfect low-level execution, and those for low-level robot control, which rely on simple, one-step commands. This disconnect prevents a comprehensive evaluation of integrated systems where both task planning and physical execution are critical. To address this, we propose Kitchen-R, a novel benchmark that unifies the evaluation of task planning and low-level control within a simulated kitchen environment. Built as a digital twin using the Isaac Sim simulator and featuring more than 500 complex language instructions, Kitchen-R supports a mobile manipulator robot. We provide baseline methods for our benchmark, including a task-planning strategy based on a vision-language model and a low-level control policy based on diffusion policy. We also provide a trajectory collection system. Our benchmark offers a flexible framework for three evaluation modes: independent assessment of the planning module, independent assessment of the control policy, and, crucially, an integrated evaluation of the whole system. Kitchen-R bridges a key gap in embodied AI research, enabling more holistic and realistic benchmarking of language-guided robotic agents.","PeriodicalId":531,"journal":{"name":"Doklady Mathematics","volume":"112 2","pages":"400 - 408"},"PeriodicalIF":0.6,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1134/S1064562425700450.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147560729","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Pairwise Image Matching for Plagiarism Detection 用于抄袭检测的成对图像匹配

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700486

D. D. Dorin, K. D. Varlamova, A. V. Grabovoy

{"title":"Pairwise Image Matching for Plagiarism Detection","authors":"D. D. Dorin, K. D. Varlamova, A. V. Grabovoy","doi":"10.1134/S1064562425700486","DOIUrl":"10.1134/S1064562425700486","url":null,"abstract":"Plagiarism detection represents a critical task across various fields, including academic publishing, journalism, e-commerce, and media verification. While substantial attention focuses on identifying textual plagiarism, image plagiarism, particularly in biology and medicine, remains a significant concern. Automated retrieval systems often surface numerous potential candidates, but a high rate of false positives – pairs incorrectly flagged as plagiarism – necessitates highly accurate pairwise matching for verification. Manual alterations to images, such as rotations, mirroring, conversion to grayscale, and color distortion constitute forms of plagiarism. This work addresses the critical need for false positive rate (FPR) minimization in pairwise image plagiarism detection through rigorous analysis of similarity scoring models. The proposed approach employs a siamese network with three key components: a weight-shared encoder, a symmetric fusion module with order-invariant embedding combination, and a similarity classification head. Training employs a hybrid self-supervised strategy with plagiarism-mimicking augmentations, combining cross-entropy loss and contrastive regularization. Ablation studies evaluate encoder architectures and fusion strategies. For comparison, identical siamese architectures utilize frozen state-of-the-art self-supervised representations Barlow Twins and CLIP, with fusion modules and classification heads trained identically. Experimental validation across multi-domain images demonstrates that end-to-end trained models consistently outperform approaches using frozen state-of-the-art representations.","PeriodicalId":531,"journal":{"name":"Doklady Mathematics","volume":"112 2","pages":"370 - 381"},"PeriodicalIF":0.6,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1134/S1064562425700486.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147560740","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Monitoring Horses in Stalls: From Object to Event Detection 监测马厩中的马：从对象到事件检测

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700437

D. Galimzianov, V. Vyshegorodtsev, I. Nezhivykh

引用次数: 0

RES-LT: Extracting Higher-Order Topological Features from Protein Language Models for Enhanced Per-Residue Classification 从蛋白质语言模型中提取高阶拓扑特征用于增强每残基分类

IF 0.6 4区数学

Doklady Mathematics Pub Date : 2026-03-18 DOI: 10.1134/S1064562425700498

M. P. Ivanova, I. E. Trofimov, A. V. Mironenko, P. V. Strashnov, M. K. Kravchenko, S. A. Barannikov, E. V. Burnaev

引用次数: 0