检索结果-内蒙古大学图书馆

8th International Conference on Computer Science and Artificial Intelligence, CSAI 2024

作者： Liu, Shuang Kabanikhin, Sergey Igorevich Strijhak, Sergei Vladimirovich Novosibirsk State University Akademgorodok Novosibirsk Russia Sobolev Institute of Mathematics Siberian Branch of the Russian Academy of Sciences Novosibirsk Russia Ivannikov Institute of System Programming of the Russian Academy of Sciences Moscow Russia

ISBN: (纸本)9798400718182

This paper explores the connection between systems of linear algebraic equations (SLAE) and machine learning methods, including regularization techniques, to establish a more novel neural network model based on linear neural networks. The goal is to construct a weight matrix for the neural network, which, by simulating the process of finding pseudo-solutions to SLAE, can generate the optimal answer for any input data. In this new neural network model, linear operations are performed first, followed by nonlinear operations, ultimately yielding an optimized weight matrix that serves as the pseudo-solution to the SLAE. The paper demonstrates how linear neural networks can be simplified to SLAE, how adding nonlinear layers to the linear neural network model can improve accuracy, and how machine learning methods can be used to find pseudo-solutions to SLAE. © 2024 Copyright held by the owner/author(s).

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Multi-Record Web Page Information Extraction From News Websites

arXiv

引用

arXiv 2025年

作者： Kustenkov, Alexander Varlamov, Maksim Yatskov, Alexander Ivannikov Institute for System Programming Russian Academy of Sciences Moscow Russia Lomonosov Moscow State University Moscow Russia

In this paper, we focused on the problem of extracting information from web pages containing many records, a task of growing importance in the era of massive web data. Recently, the development of neural network methods has improved the quality of information extraction from web pages. Nevertheless, most of the research and datasets are aimed at studying detailed pages. This has left multi-record"list pages" relatively understudied, despite their widespread presence and practical significance. To address this gap, we created a large-scale, open-access dataset specifically designed for list pages. This is the first dataset for this task in the Russian language. Our dataset contains 13,120 web pages with news lists, significantly exceeding existing datasets in both scale and complexity. Our dataset contains attributes of various types, including optional and multi-valued, providing a realistic representation of real-world list pages. These features make our dataset a valuable resource for studying information extraction from pages containing many records. Furthermore, we proposed our own multi-stage information extraction methods. In this work, we explore and demonstrate several strategies for applying MarkupLM to the specific challenges of multi-record web pages. Our experiments validate the advantages of our methods. By releasing our dataset to the public1, we aim to advance the field of information extraction from multi-record pages. © 2025, CC BY-NC-ND.

关键词： Data collection

来源：评论

学校读者我要写书评

暂无评论

Complexity of the Uniform Membership Problem for Hyperedge Replacement Grammars

arXiv

引用

arXiv 2025年

作者： Pshenitsyn, Tikhon Steklov Mathematical Institute 8 Gubkina st. Moscow119991 Russia Ivannikov Institute for System Programming 25 Alexander Solzhenitsyn st. Moscow109004 Russia

We investigate complexity of the uniform membership problem for hyperedge replacement grammars in comparison with other mildly context-sensitive grammar formalisms. It turns out that the complexity of the problem considered depends heavily on how one defines a hypergraph. There are two commonly used definitions in the field which differ in whether repetitions of attachment nodes of a hyperedge are allowed in a hypergraph or not. We show that, if repetitions are allowed, then the problem under consideration is EXPTIME-complete even for string-generating hyperedge replacement grammars while it is NP-complete if repetitions are disallowed. We also prove that checking whether a hyperedge replacement grammar is string-generating is EXPTIME-complete. Copyright © 2025, The Authors. All rights reserved.

关键词： Context sensitive grammars

来源：评论

学校读者我要写书评

暂无评论

Decentralised convex optimisation with probability-proportional-to-size quantization

arXiv

引用

arXiv 2025年

作者： Pasechnyuk, D.A. Dvurechensky, P. Uribe, C.A. Gasnikov, A.V. Mohamed bin Zayed University of Artificial Intelligence United Arab Emirates Ivannikov Institute for System Programming the Russian Academy of Sciences Russia Weierstrass Institute for Applied Analysis and Stochastics Germany Rice University United States Moscow Institute of Physics and Technology Russia

Communication is one of the bottlenecks of distributed optimisation and learning. To overcome this bottleneck, we propose a novel quantization method that transforms a vector into a sample of components’ indices drawn from a categorical distribution with probabilities proportional to values at those components. Then, we propose a primal and a primal-dual accelerated stochastic gradient methods that use our proposed quantization, and derive their convergence rates in terms of probabilities of large deviations. We focus on affine-constrained convex optimisation and its application to decentralised distributed optimisation problems. To illustrate the work of our algorithm, we apply it to the decentralised computation of semi-discrete entropy regularized Wasserstein *** Codes 65K10 © 2025, CC BY.

关键词： Optimization algorithms

来源：评论

学校读者我要写书评

暂无评论

Methods of Visualisation for Flows with Internal Waves Attractors

Scientific Visualization

引用

Scientific Visualization 2021年第5期13卷 113-121页

作者： Ryazanov, D.A. Elistratov, S.A. Kraposhin, M.V. Ivannikov Institute for System Programming The RAS Russia Lomonosov Moscow State University Russia

Application of different approaches for visualization of hydrodynamic fields of internal waves is closely connected with the possibility of extracting important numerical characteristics of the flows. In this paper we describe methods of visualisation, which were shown to be very effective for the illustration of the internal wave attractors both for laboratory experiments and numerical simulations. On the other hand, some novel approaches for description of the wave flows with accumulation of energy are discussed. Methods of vortices identification show interesting properties of the wave flows, and may give an alternative for the estimation of the width of wave beams. At the same time, the limitations of the applicability of the described methods strongly depend on the space and time resolution of the available data, which is especially important in the areas of the internal wave beam reflection. © 2021 National Research Nuclear University. All rights reserved.

关键词： Open source software

来源：评论

学校读者我要写书评

暂无评论

POD-based Hydrodynamical Structures Visualization in Flows with an Internal Wave Attractor

Scientific Visualization

引用

Scientific Visualization 2023年第2期15卷 125-133页

作者： Elistratov, S.A. Shirshov Institute of Oceanology of RAS Moscow Russia Ivannikov Institute for System Programming of RAS Moscow Russia

Hydrodynamical structure attending a flow can be hid and hardly to reveal. One of the methods to find them is to use mode decomposition (such as Proper orthogonal decomposition, POD). The method represents the field given as a series of spatial modes multiplied by corresponding temporal coefficients. In the article the method is discussed applyingly to a complex flow with a wave attractor structure. Attractor modes present structured vortex-like figure which cannot be claimed to be aleatory. As it turns out POD modes are not just a formal decomposition but have a physical origin: they are connected with instability cascade minor frequencies, as spectral investigation shows. Another consequence of that is that one of the collateral structure maximum can be visible. This proposition is proven as the structure is found to be visible in the flow itself. © 2023 National Research Nuclear University. All rights reserved.

关键词： Visualization

来源：评论

学校读者我要写书评

暂无评论

Neuron Coverage Maximization for Effective Test Set Construction with Respect to the Model

引用

Journal of Mathematical Sciences (United States) 2024年第2期285卷 189-200页

作者： Kushchuk, D. Ryndin, M. Ivannikov Institute for System Programming of the RAS Moscow Russian Federation

Real world data is not stationary and thus models must be monitored in production. One way to be sure in a model’s performance is regular testing. If the labels are not available, the task of minimizing the labeling cost can be formulated. In this work, we investigate and develop various ways to construct a minimum test set for a given trained model, in a fashion where the accuracy of the model calculated on the chosen subset is as close to the real one as possible. We focus on the white box scenario and propose a novel approach that uses neuron coverage as a observable functional to maximize in order to minimize the number of samples. We evaluate the proposed approach and compare it to Bayesian methods and stratification algorithms that are the main approaches to solve this task in literature. The developed method shows approximately the same level of performance but has a number of advantages over its competitors. It is deterministic, thus eliminating the dispersion of the results. Also, this method can give one a hint on the optimal budget. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Realistic Adversarial Attacks on Object Detectors Using Generative Models

引用

Journal of Mathematical Sciences (United States) 2024年第2期285卷 245-254页

作者： Shelepneva, D. Arkhipenko, K. Ivannikov Institute for System Programming of the RAS Moscow Russian Federation

An important limitation of existing adversarial attacks on real-world object detectors lies in their threat model: adversarial patch-based methods often produce suspicious images while image generation approaches do not restrict the attacker’s capabilities of modifying the original scene. We design a threat model where the attacker modifies individual image segments and is required to produce realistic images. We also develop and evaluate a white-box attack that utilizes generative adversarial nets and diffusion models as a generator of malicious images. Our attack is able to produce high-fidelity images as measured by the Fréchet inception distance (FID) and reduces the mAP of Faster R-CNN model by over 0.2 on Cityscapes and COCO-Stuff datasets. A PyTorch implementation of our attack is available at https://***/DariaShel/gan-attack. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.

关键词： adversarial examples diffusion models generative adversarial networks object detectors

来源：评论

学校读者我要写书评

暂无评论

Automatic verification of the text layer correctness in PDF documents

Automatic verification of the text layer correctness in PDF ...

引用

2024 ivannikov Memorial Workshop, IVMEM 2024

作者： Belyaeva, Oksana V. Golodkov, Aleksandr Bukhatov, Bekzat Programming of the RAS Ivannikov Institute for System Moscow Russia

ISBN: (纸本)9798331504663

PDF documents can contain incorrect textual layers due to low scanning quality, font embedding errors or other reasons. An incorrect text layer can significantly hamper automatic document processing and limit document parsing. To help tackle the problem, this work analyzed the causes of incorrect text layers (gibberish characters) and described a method to automatically extract text information by detecting the correct text layer of PDF files, using OCR and a PDF extractor. To determine the correctness/incorrectness of the text layer obtained from PDF extractors in the method, we trained a binary classifier on the generated *** carried out experiments and assessed the quality of various classification models for text correctness, tested the method on real data, thereby confirming the effectiveness of the developed approach. The obtained results can be useful for further development of methods for automatically checking textual layers and improving the quality of automatic processing PDF documents. © 2024 IEEE.

关键词： Automatic PDF text extraction Corrupted PDF Gibberish text detection Machine learning Reasons for text incorrectness Text classification

来源：评论

学校读者我要写书评

暂无评论

Towards Symbolic Pointers Reasoning in Dynamic Symbolic Execution

Towards Symbolic Pointers Reasoning in Dynamic Symbolic Exec...

引用

2021 ivannikov Memorial Workshop, IVMEM 2021

作者： Kuts, Daniil Ivannikov Institute for System Programming RAS Moscow Russia

ISBN: (纸本)9781665423274

Dynamic symbolic execution is a widely used technique for automated software testing, designed for execution paths exploration and program errors detection. A hybrid approach has recently become widespread, when the main goal of symbolic execution is helping fuzzer increase program coverage. The more branches symbolic executor can invert, the more useful it is for fuzzer. A program control flow often depends on memory values, which are obtained by computing address indexes from user input. However, most DSE tools don't support such dependencies, so they miss some desired program branches. We implement symbolic addresses reasoning on memory reads in our dynamic symbolic execution tool Sydr. Possible memory access regions are determined by either analyzing memory address symbolic expressions, or binary searching with SMT-solver. We propose an enhanced linearization technique to model memory accesses. Different memory modeling methods are compared on the set of programs. Our evaluation shows that symbolic addresses handling allows to discover new symbolic branches and increase the program coverage. © 2021 IEEE.

关键词： Software testing

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：