检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Lukyanov, Kirill Drobyshevskiy, Mikhail Shaikhelislamov, Danil Turdakov, Denis Ivannikov Institute for System Programming The Russian Academy of Sciences Moscow Russia Moscow Institute of Physics and Technology National Research University Moscow Russia

Social networks crawling is in the focus of active research the last years. One of the challenging task is to collect target nodes in an initially unknown graph given a budget of crawling steps. Predicting a node property based on its partially known neighbourhood is at the heart of a successful crawler. In this paper we adopt graph neural networks for this purpose and show they are competitive to traditional classifiers and are better for individual cases. Additionally we suggest a training sample boosting technique, which helps to diversify the training set at early stages of crawling and thus improves the predictor quality. The experimental study on three types of target set topology indicates GNN based approach has a potential in crawling task, especially in the case of distributed target nodes. © 2024, CC BY-NC-ND.

关键词： Budget control

来源：评论

学校读者我要写书评

暂无评论

Accelerated Methods with Compressed Communications for Distributed Optimization Problems under Data Similarity

arXiv

引用

arXiv 2024年

作者： Bylinkin, Dmitry Beznosikov, Aleksandr Moscow Institute of Physics and Technology Russia Ivannikov Institute for System Programming of the Russian Academy of Sciences Russia Sber AI Lab Skoltech Russia

In recent years, as data and problem sizes have increased, distributed learning has become an essential tool for training highperformance models. However, the communication bottleneck, especially for high-dimensional data, is a challenge. Several techniques have been developed to overcome this problem. These include communication compression and implementation of local steps, which work particularly well when there is similarity of local data samples. In this paper, we study the synergy of these approaches for efficient distributed optimization. We propose the first theoretically grounded accelerated algorithms utilizing unbiased and biased compression under data similarity, leveraging variance reduction and error feedback frameworks. In terms of communication time our theory gives O ( 1 + h M-1/4 + ω-1/2 ip δ/μ ) complexity for unbiased compressors and O ( 1 + β1/4 √ δ/μ ) for biased ones, where M is the number of computational nodes, β is the compression power, δ is the similarity measure and μ is the parameter of strong convexity of the objective. Our theoretical results are of record and confirmed by experiments on different average losses and datasets. Copyright © 2024, The Authors. All rights reserved.

关键词： Data compression

来源：评论

学校读者我要写书评

暂无评论

Improving Scalability of Inter-module Source Code Static Taint Analysis

Improving Scalability of Inter-module Source Code Static Tai...

引用

Ivannikov ISPRAS Open Conference (ISPRAS)

作者： D.A. Chibisov N.V. Shimchik V.N. Ignatyev Ivannikov Institute for System Programming of the Russian Academy of Sciences (ISP RAS) Lomonosov Moscow State University Moscow Russian Federation

Taint analysis mainly detects data leaks and unsafe use of untrusted data obtained from external sources. It reports dataflow paths connecting taint sources and taint sinks in the analyzed program. This paper focuses on inter-modular aspect of project analysis: large projects may contain dozens or even hundreds of target executable files to be analyzed. If analyzed separately, scalability problem arises due to high percentage of reused source files. Here we discuss the idea of linking these bitcode files together to form a single project-wide bitcode file and some caveats of this method: e.g. it makes possible to have execution paths visiting functions' definitions which never appear in the same target file. We also propose a method of eliminating such paths for IFDS-based analyzers. The methods were implemented in static taint analyzer Irbis and evaluated on several real projects with hundreds of thousands lines of code.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Classification of Static Analyzer Warnings using Machine Learning Methods

Classification of Static Analyzer Warnings using Machine Lea...

引用

Ivannikov Memorial Workshop (IVMEM)

作者： U. V. Tsiazhkorob V. N. Ignatyev Ivannikov Institute for System Programming of the Russian Academy of Sciences (ISP RAS) Moscow Russian Federation Moscow Institute of Physics and Technology Moscow Russian Federation Lomonosov Moscow State University Moscow Russian Federation

ISBN: (数字)9798331504663

ISBN: (纸本)9798331504670

The paper is devoted to the approach for static analyzer’s warnings classification using machine learning methods. Static analysis plays a crucial role in software development by identifying potential errors and vulnerabilities in source code. However, the huge amount of warnings generated by static analyzers often overwhelms developers, making it challenging to prioritize and address them effectively. This paper proposes a novel approach to automatically classify static analyzer warnings using machine learning techniques. It showed the result up to 92% accuracy on real-world projects. Our method uses a set of code metrics collected during the analysis process to generate features for training the classifier. In the paper it is explored various strategies for feature selection and model training, considering different types of warnings and analyzers. The effectiveness of the proposed approach in warning classification was experimentally demonstrated. The developed method will be used to automatically review the results generated by the industrial static analyzer.

关键词： Training Measurement Reviews Source coding Conferences Computer bugs Static analysis

来源：评论

学校读者我要写书评

暂无评论

Improving the Accuracy of Library Function Modeling in the Static Analyzer

Improving the Accuracy of Library Function Modeling in the S...

引用

Ivannikov ISPRAS Open Conference (ISPRAS)

作者： M. G. Biktimirov V. N. Ignatyev M. V. Belyaev Ivannikov Institute for System Programming of the Russian Academy of Sciences (ISP RAS) Lomonosov Moscow State University Moscow Russian Federation

Program's source code usually contains many calls of library functions without available sources, therefore there is no way to analyze them like user code. To achieve accuracy required by industrial static analyzer, various library modeling methods are used. The paper presents one of the methods used by the SharpChecker static source code analyzer for modeling library functions using C# specifications in the symbolic execution engine. The approach allows to store and explore predicates of events, important side-effects and other information useful for error detectors with reasonable accuracy. It makes it possible to model side-effects of user-defined lambda-functions passed to external methods as parameters, which is especially important for frequently used C# LINQ methods. The effectiveness of these improvements was evaluated on the set of open source projects containing 6 million lines of code.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Some Aspects of Lexicalisation of Case Forms with Spatial Meaning in the Udmurt Language

Some Aspects of Lexicalisation of Case Forms with Spatial Me...

引用

Ivannikov ISPRAS Open Conference (ISPRAS)

作者： N. V. Kondratieva Udm FIC Ural Branch of the Russian Academy of Sciences Institute of History Language and Literature Izhevsk Russia Institute for System Programming of the Russian Academy of Sciences Moscow Russia

. The article is devoted to the study of the issues of lexicalisation of the so-called “somatic lexicon” in the paradigm of spatial case forms in the modern Udmurt language in a typological aspect. The relevance of the research lies in the fact that the study of the above questions of Udmurt linguistics opens new perspectives to some lexico-grammatical problems of Permic linguistics both from a diachronic and synchronistic aspect. As a result of the study it was revealed that the somatic lexicon is represented in different degrees in the case paradigm of the modern Udmurt language, the most common being the use of the illative case. Apparently, this also explains the fact that the process of lexicalisation is most often implemented in the specified grammatical form. This is also confirmed by fixed word combinations, where one element has now lost its independent distribution: tode vajyny ‘to remember’, bure vajyny ‘to remember’, sane ponyny ‘to take into account’ and others. The phenomenon of lexicalisation, that is the transition of combinations of separate elements into fixed word combinations, in Udmurt is most often affected by linguistic units containing the verb forms pyryny ‘to enter, come in and other words with the given meaning’,lyktyny ‘to come, come in, etc.’, voź yny ‘hold’, potyny 'go out, go out, etc.” koš kyny ‘go away, leave, etc. ‘ with examples like pele pyryny ‘remember’, jyre lyktyny ‘come to mind’, ś ulmaz voźyny ‘hold in the heart’, jyryś potyny ‘forget’, jyryś koš kyny ‘forget’, etc. The full illative - inessive - elative paradigm can be observed in the word-formation of a limited number of somatic lexical units such as jyr ‘head’, pel’ ‘ear’, or ś ulem ‘heart’. A typologically similar tendency can be observed in other languages, for example in russian. Evidence is found in written attestations from the turn of the 19th-20th centuries: vojti v golovu ‘to enter the head’ - derzhat’ v golove ‘to keep in the head’ - vyjti iz golovy 'to leave the

关键词：

来源：评论

学校读者我要写书评

暂无评论

Regularization of linear machine learning problems

arXiv

引用

arXiv 2024年

作者： Liu, S. Kabanikhin, S.I. Strijhak, S.V. Novosibirsk State University Novosibirsk Russia Sobolev Institute of Mathematics Siberian Branch of the Russian Academy of Sciences Novosibirsk Russia V.P. Ivannikov Institute of System Programming of the Russian Academy of Sciences Moscow Russia

In this paper, we consider the simplest version of a linear neural network(LNN). Assuming that for training (constructing an optimal weight matrix $Q$)we have a set of training pairs, i.e. we know the input data \\begin{equation}G=\\left\\{g^{\\left(1\ight)},g^{\\left(2\ight)},\\cdots,g^{\\left(K\ight)}\ight\\},\\end{equation} as well as the correct answers to these input data\\begin{equation}H=\\left\\{h^{\\left(1\ight)},h^{\\left(2\ight)},\\cdots,h^{\\left(K\ight)}\ight\\}.\\end{equation} We will study the possibilities of constructing a weight matrix$Q$ of a neural network that will give correct answers to arbitrary input databased on the connection of the specified problem with a system of linearalgebraic equations (SLAE). Consider a class of neural networks in which eachneuron has only one output signal and performs linear operations. We will showhow such LNEs are reduced to SLAEs. Since the questions $G$ and the correctanswers $H$ are known to us, the desired weight matrix $Q$ must satisfy theequations \\begin{equation} Qg^{\\left(k\ight)}=h^{\\left(k\ight)},k=1,2,\\cdots,K. \\end{equation} It is required to restore $Q$. In the generalcase, the matrix $Q$ is rectangular $Q=Q_{MN}=\\left\\{q_{mn}\ight\\}$, $m$ isthe row number, and $g^{\\left(k\ight)}\\in\\mathbb{R}^N$,$h^{\\left(k\ight)}\\in\\mathbb{R}^M$. Let $G_{NK}$ be a matrix composed ofcolumns $g^{\\left(1\ight)}, g^{\\left(2\ight)},\\cdots,g^{\\left(k\ight)}$, and$H_{MK}$ be a matrix composed of columns$h^{\\left(1\ight)},h^{\\left(2\ight)},\\cdots,h^{\\left(k\ight)}$. Then, withrespect to $Q_{MN}$, we obtain a matrix SLAE. \\begin{equation}Q_{MN}G_{NK}=H_{MK}. \\end{equation} This paper will present methods forregularizing the constructed system. © 2024, CC BY.

关键词： Input output programs

来源：评论

学校读者我要写书评

暂无评论

Revisiting Control Flow Recovery from Binaries: An Abstract Interpretation-based Approach

Revisiting Control Flow Recovery from Binaries: An Abstract ...

引用

Ivannikov ISPRAS Open Conference (ISPRAS)

作者： Dmitrii Ilichev Mikhail Solovev Ivannikov Institute for System Programming Russian Academy of Sciences Moscow Russia Faculty of Computational Mathematics and Cybernetics Moscow State University Moscow Russia

ISBN: (数字)9798331526023

ISBN: (纸本)9798331526030

Static program analysis requires control flow graphs to be reconstructed from binary code. However, instruction recovery is challenging due to statically unknown targets of indirect branches, leading to incomplete or impractically over-approximated control flow graphs. This paper presents a context-sensitive approach to interproce-dural control flow graph recovery from binary code that employs abstract interpretation to handle indirect jumps. The algorithm can be applied to arbitrary binary code. We introduce such a concept of function that allows us to identify subroutines with irregular call and return patterns. Experiments indicate that the proposed technique produces more precise control flow graphs compared to state-of-the-art tools like IDA Pro and Cutter.

关键词： Algorithms Instruction sets Binary codes Flow graphs

来源：评论

学校读者我要写书评

暂无评论

Numerical Modeling of a Melt Flow by the Czochralski Method in the OpenFOAM Package Using a Quasi-Hydrodynamic Algorithm

引用

Mathematical Models and Computer Simulations 2023年第1-Sup期15卷 S131-S142页

作者： Kirushina, M.A. Elizarova, T.G. Epikhin, A.S. Keldysh Institute of Applied Mathematics Russian Academy of Sciences Moscow Russian Federation Ivannikov Institute for System Programming Russian Academy of Sciences Moscow Russian Federation

Abstract: The formulation of the problem, the numerical solution method, and description of the test calculation case for modeling a melt flow in the problem of crystal growth by the Czochralski (Chochralsky) method are presented. The mathematical model is based on regularized or quasi-hydrodynamic (QHD) equations to describe a viscous, incompressible heat-conducting fluid. The numerical algorithm is implemented on the OpenFOAM package within the mulesQHDFoam solver. The proposed approach allows calculations of an unsteady three-dimensional flow in a crucible with the given geometry for high Reynolds and Grashof numbers. As an example, the results for the flow in a simplified configuration are given. An asymmetric transient flow is obtained. © 2023, Pleiades Publishing, Ltd.

关键词： Czochralski method OpenFOAM package quasi-hydrodynamic algorithm viscous incompressible fluid

来源：评论

学校读者我要写书评

暂无评论

Page Layout Analysis for Refining Table Extraction from PDF Documents

Page Layout Analysis for Refining Table Extraction from PDF ...

引用

2021 Ivannikov Ispras Open Conference, ISPRAS 2021

作者： Mikhailov, Andrey Shigarov, Alexey Matrosov Institute For System Dynamics And Control Theory Siberian Branch Of The Russian Academy Of Sciences Irkutsk664033 Russia Ivannikov Institute For System Programming Russian Academy Of Sciences Moscow109004 Russia

ISBN: (纸本)9781665423304

Perhaps, PDF is the most popular format to share non-editable documents. PDF documents are often untagged. In particular, this means that positions and the cell structure of tables are not designated explicitly. PDF table detection predicts bounding boxes of tables on document pages. Some of the predictions inevitably happen to be false. This negatively affects the accuracy of table structure recognition. We argue that the page layout analysis in pre- and post-processing can refine the table detection. We suggest pre-processing algorithms for the recognition of headings, running titles, paragraphs, and images in PDF pages. This allows selecting areas of interest inside pages where real tables can be placed. Then we use deep neural networks to predict tables only in these areas. We also propose post-processing algorithms to verify predictions and filter out false table candidates after table detection. Our empirical study shows that the proposed approach reduces errors in the table detection and improve the PDF table extraction overall. © 2021 IEEE.

关键词： Extraction

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：