检索结果-内蒙古大学图书馆

Highly Smooth Zeroth-Order Methods for Solving Optimization Problems under the PL Condition

COMPUTATIONAL MATHEMATICS AND MATHEMATICAL PHYSICS 2024年第4期64卷 739-770页

作者： Gasnikov, A. V. Lobanov, A. V. Stonyakin, F. S. Moscow Inst Phys & Technol Dolgoprudnyi 141700 Russia Innopolis Univ Innopolis 420500 Russia Russian Acad Sci Inst Syst Programming Moscow 125047 Russia VI Vernadsky Crimean Fed Univ Simferopol 295007 Russia

In this paper, we study the black box optimization problem under the Polyak-Lojasiewicz (PL) condition, assuming that the objective function is not just smooth, but has higher smoothness. By using "kernel-based" approximations instead of the exact gradient in the Stochastic Gradient Descent method, we improve the best-known results of convergence in the class of gradient-free algorithms solving problems under the PL condition. We generalize our results to the case where a zeroth-order oracle returns a function value at a point with some adversarial noise. We verify our theoretical results on the example of solving a system of nonlinear equations.

关键词： black-box optimization gradient-free methods kernel approximation maximum noise level

来源：评论

学校读者我要写书评

暂无评论

The Software system LingvoDoc and the Possibilities It Offers for Documentation and Analysis of Ob-Ugric Languages

引用

DOKLADY MATHEMATICS 2022年第3期105卷 187-206页

作者： Normanskaja, Yu, V Borisenko, O. D. Beloborodov, I. B. Avetisyan, A., I Russian Acad Sci Ivannikov Inst Syst Programming Moscow Russia Russian Acad Sci Inst Linguist Moscow Russia

The LingvoDoc system (http://*** ) provides a service for collaborative language documentation and computations on the collected data. This software system provides GraphQL HTTP API for all the system components and allows its users to build their own extensions for data analysis or even to integrate it with their own software. Thanks to a special database and application design pattern, it is possible to construct offline applications integrated with the LingvoDoc system: these applications would need to have an internet connection only once to synchronize basic data types and for authentification purposes. The system itself allows users to construct multilayer dictionaries, attach them to the geographical map, fill documents with metadata, share access to dictionaries with other users or with everyone. The LingvoDoc system provides fine-grained access control lists for sharing, which allows to separate users into groups of dictionary editors, proofreaders and read-only users. The system also provides some computational algorithms on the stored data: phonology computations, automatic and guided deduplication inside the dictionaries etc. The system allows users to choose the dictionary structure. The system supports the following data types: text, images, sounds (way, mp3, and flac), markups (ELAN and Praat formats), directed and undirected links between stored entities. A user can choose the most suitable format for their dictionary. Also, the system provides ELAN corpora storage, viewer and processing. In LingvoDoc there are 13 programs made for authors of the dictionary (only 4 of them are available for all users of the system). These programs analyze language data from phonetical, morphological and etymological point of view. This analysis previously was performed manually by linguists. Our programs allows do it tens and sometimes hundreds times faster. This paper presents the documentation and an analysis of Ob-Ugric languages using the LingvoDo

关键词： software documentation and analysis of languages data analysis

来源：评论

学校读者我要写书评

暂无评论

Protecting Locks Against Unbalanced Unlock() 23

Protecting Locks Against Unbalanced Unlock()

引用

35th ACM Symposium on Parallelism in Algorithms and Architectures (SPAA)

作者： Shahare, Vivek Chabbi, Milind Hegde, Nikhil Indian Inst Technol Dharwad Karnataka India Uber Technol Inc Programming Syst Grp Sunnyvale CA USA

ISBN: (纸本)9781450395458

The lock is a building-block synchronization primitive that enables mutually exclusive access to shared data in shared-memory parallel programs. Mutual exclusion is typically achieved by guarding the code that accesses the shared data with a pair of lock() and unlock() operations. Concurrency bugs arise when this ordering of operations is violated. In this paper, we study a particular pattern of misuse where an unlock() is issued without first issuing a lock(), which can happen in code with complex control flow. This misuse is surprisingly common in several important open-source repositories we study. We systematically study what happens due to this misuse in several popular locking algorithms. We study how misuse can be detected and how the locking protocols can be fixed to avoid the unwanted consequences of misuse. Most locks require simple changes to detect and prevent this misuse. We evaluate the performance traits of modified implementations, which show mild performance penalties in most scalable locks.

关键词： locks threads concurrency unbalanced-unlock synchronization

来源：评论

学校读者我要写书评

暂无评论

Precedence constrained generalized traveling salesman problem: Polyhedral study, formulations, and branch-and-cut algorithm

引用

EUROPEAN JOURNAL OF OPERATIONAL RESEARCH 2023年第2期309卷 488-505页

作者： Khachai, Daniil Sadykov, Ruslan Battaia, Olga Khachay, Michael Kedge Business Sch 680 Cours Liberat Talence France Univ Bordeaux Inria Ctr 200 Ave Vieille Tour Talence France RAS Ivannikov Inst Syst Programming Moscow Russia

The Precedence Constrained Generalized Traveling Salesman Problem (PCGTSP) is an extension of two well-known combinatorial optimization problems - the Generalized Traveling Salesman Problem (GTSP) and the Precedence Constrained Asymmetric Traveling Salesman Problem (PCATSP), whose path version is known as the Sequential Ordering Problem (SOP). Similarly to the classic GTSP, the goal of the PCGTSP, for a given input digraph and partition of its node set into clusters, is to find a minimum cost cyclic route (tour) visiting each cluster in a single node. In addition, as in the PCATSP, feasible tours are restricted to visit the clusters with respect to the given partial order. Unlike the GTSP and SOP, to the best of our knowledge, the PCGTSP still remain to be weakly studied both in terms of polyhedral theory and algo-rithms. In this paper, for the first time for the PCGTSP, we propose several families of valid inequalities, establish dimension of the PCGTS polytope and prove sufficient conditions ensuring that the extended Balas' pi- and sigma-inequalities become facet-inducing. Relying on these theoretical results and evolving the state-of-the-art algorithmic approaches for the PCATSP and SOP, we introduce a family of MILP-models (formulations) and several variants of the branch-and-cut algorithm for the PCGTSP. We prove their high performance in a competitive numerical evaluation against the public benchmark library PCGTSPLIB, a known adaptation of the classic SOPLIB to the problem in question. (c) 2023 Elsevier B.V. All rights reserved.

关键词： Travelling salesman Integer programming Branch-and-cut algorithm Facet-inducing inequalities Polyhedral structure

来源：评论

学校读者我要写书评

暂无评论

Typed Unknown Values: A Step towards Solving the Problem of Missing Data Representation in Relational Databases

引用

programming AND COMPUTER SOFTWARE 2024年第7期50卷 532-548页

作者： Kuznetsov, S. D. Russian Acad Sci Ivannikov Inst Syst Programming Moscow 109004 Russia Lomonosov Moscow State Univ Moscow Russia Moscow Inst Phys & Technol Dolgoprudnyi 141700 Moscow Oblast Russia Natl Res Univ Higher Sch Econ Moscow 101978 Russia

The state of affairs in the field of missing data management in relational databases leaves much to be desired. The SQL standard uses the universal null value to represent missing data, and the management is based on three-valued logic, in which the null value is identified with the third Boolean value. This solution is conceptually inconsistent and often results in DBMS behavior that is not intuitive. An alternative approach based on typed special values leaves all handling of missing data to users. In this paper, we analyze the long history of research and development that led to this situation. We come to the conclusion that no other solution could have appeared in the SQL standard because of the choice of the mechanism of universal null value more than 50 years ago, whereas the alternative mechanism cannot provide system support for special values due to the use of two-valued logic. We propose a combined approach using typed special values based on three-valued logic. This approach allows one to use the semantics of data types when processing queries with conditions that involve unknown data. In addition, our approach makes it possible to define a full-fledged three-valued logic in which a special value of the Boolean type is the third Boolean value.

关键词： relational model missing data null value three-valued logic typed special values

来源：评论

学校读者我要写书评

暂无评论

Stochastic Gradient Methods with Preconditioned Updates

引用

JOURNAL OF OPTIMIZATION THEORY AND APPLICATIONS 2024年第2期201卷 471-489页

作者： Sadiev, Abdurakhmon Beznosikov, Aleksandr Almansoori, Abdulla Jasem Kamzolov, Dmitry Tappenden, Rachael Takac, Martin Russian Acad Sci ISP RAS Ivannikov Inst Syst Programming Moscow Russia Moscow Inst Phys & Technol MIPT Moscow Russia Mohamed Bin Zayed Univ Artificial Intelligence MBZ Abu Dhabi U Arab Emirates Univ Canterbury Christchurch New Zealand

This work considers the non-convex finite-sum minimization problem. There are several algorithms for such problems, but existing methods often work poorly when the problem is badly scaled and/or ill-conditioned, and a primary goal of this work is to introduce methods that alleviate this issue. Thus, here we include a preconditioner based on Hutchinson's approach to approximating the diagonal of the Hessian and couple it with several gradient-based methods to give new 'scaled' algorithms: Scaled SARAH and Scaled L-SVRG. Theoretical complexity guarantees under smoothness assumptions are presented. We prove linear convergence when both smoothness and the PL-condition are assumed. Our adaptively scaled methods use approximate partial second-order curvature information and, therefore, can better mitigate the impact of badly scaled problems. This improved practical performance is demonstrated in the numerical experiments also presented in this work.

关键词： Optimization Non-convex optimization Stochastic optimization Scaled methods Variance reduction

来源：评论

学校读者我要写书评

暂无评论

Artificial Intelligence (AI) Solution for Plasma Cells Detection

引用

programming AND COMPUTER SOFTWARE 2023年第8期49卷 873-880页

作者： Makarchuk, A. Asaturova, A. Ushakov, E. Tregubova, A. Badlaeva, A. Tabeeva, G. Karpulevich, E. Markin, Yu. Russian Acad Sci Ivannikov Inst Syst Programming Moscow 109004 Russia Minist Healthcare Russian Federat FSBI Natl Med Res Ctr Obstet Gynecol & Perinatol Moscow 117997 Russia

The article investigates the application of a neural network diagnosis model to histological images in order to detect plasma cells for chronic endometritis detection. A two-stage algorithm was developed for plasma cell detection. At the first stage, a CenterNet model was used to detect stromal and epithelial cells. The neural network was trained on an open dataset with histological images and further fine-tuned using an additional labeled dataset. A labeling protocol was used, and the coefficient of agreement between two experts was calculated, which turned out to be 0.81. At the second stage, using the developed algorithm based on computer vision methods, plasma cells were identified and their HSV color boundaries were calculated. For the two-stage algorithm the following quality metrics were obtained: precision = 0.70, recall = 0.43, f1-score = 0.53. The model then was modified to detect only plasma cells and trained on a dataset with histological images containing labeled plasma cells. The quality metrics of the modified detection model were obtained: precision = 0.73, recall = 0.89, f1-score = 0.8. As a result of the comparison, the modified detection model approach showed the best quality metrics. Automating the work of counting plasma cells will allow doctors to spend less time on routine activities.

关键词： Cytology

来源：评论

学校读者我要写书评

暂无评论

A Model and Declarative Language for Specifying Binary Data Formats

引用

programming AND COMPUTER SOFTWARE 2022年第7期48卷 469-483页

作者： Evgin, A. A. Solovev, M. A. Padaryan, V. A. Russian Acad Sci Ivannikov Inst Syst Programming Moscow 109004 Russia Moscow MV Lomonosov State Univ Moscow 1199919 Russia

Tasks related to binary data formats include parsing, generating, and conjoint code and data analysis. A key element for all of these tasks is a universal data format model. An approach to modeling binary data formats is proposed. The described model has sufficient expressive power for specifying the majority of widespread data formats. A distinctive feature of this model is its flexibility in specifying field locations and the ability to describe external fields the structure of which cannot be determined by parsing. The implemented infrastructure makes it possible to create and modify the representation using application programming interfaces. An algorithm is proposed for parsing binary data using the specified model based on the concept of computability of fields. A domain-specific language for data format specification is also described. The specified formats and potential practical applications of the model for programmatic analysis of formatted data are discussed.

关键词： Application programming interfaces (API)

来源：评论

学校读者我要写书评

暂无评论

Performance Model of Graphics Pipeline for a One-Pass Rendering of 3D Dynamic Scenes

引用

programming AND COMPUTER SOFTWARE 2021年第7期47卷 522-533页

作者： Gonakhchyan, V., I Russian Acad Sci Ivannikov Inst Syst Programming Moscow 109004 Russia

A one-pass rendering scheme of 3D dynamic scenes using modern graphics processing units (GPUs) and graphical interfaces is considered. This scheme uses the following methods and techniques: frustum culling, hardware occlusion queries, fragmentation, and caching of command buffers. These methods require significant computational resources, and the amount of work on the stages of graphics pipeline depends on their results. For efficient rendering, the balanced use of resources during the pipelined processing and transferring graphics data is important. A performance model of the graphics pipeline as applied to rendering 3D dynamic scenes that makes it possible to estimate the required resources, depending on the selected base methods and characteristics of the scene being rendered is proposed. In distinction from the existing methods and models, the proposed model allows one to calculate the cost of composing command buffers using various recording techniques, calculate the cost of transferring, executing, and loading the results of hardware occlusion queries. Formulas for calculating the frame rendering time depending on the number of occlusion queries are derived. A method for estimating the number of hardware occlusion queries needed for efficient rendering of dynamic scenes is proposed. Computational experiments demonstrating the relevance of the proposed model and the efficiency of the developed method for rendering large dynamic scenes are carried out.

关键词： Computer Science general Software Engineering/programming and Operating systems Operating systems Software Engineering Artificial Intelligence

来源：评论

学校读者我要写书评

暂无评论

Lip-Reading Classification of Turkish Digits Using Ensemble Learning Architecture Based on 3DCNN

引用

APPLIED SCIENCES-BASEL 2025年第2期15卷 563-563页

作者： Erbey, Ali Barisci, Necaattin Usak Univ Distance Educ Vocat Sch Dept Comp Programming TR-64200 Usak Turkiye Gazi Univ Informat Inst Informat Syst TR-06560 Ankara Turkiye Gazi Univ Fac Technol Dept Comp Engn TR-06560 Ankara Turkiye

Understanding others correctly is of great importance for maintaining effective communication. Factors such as hearing difficulties or environmental noise can disrupt this process. Lip reading offers an effective solution to these challenges. With the growing success of deep learning architectures, research on lip reading has gained momentum. The aim of this study is to create a lip reading dataset for Turkish digit recognition and to conduct predictive analyses. The dataset has divided into two subsets: the face region and the lip region. CNN, LSTM, and 3DCNN-based models, including C3D, I3D, and 3DCNN+BiLSTM, were used. While LSTM models are effective in processing temporal data, 3DCNN-based models, which can process both spatial and temporal information, achieved higher accuracy in this study. Experimental results showed that the dataset containing only the lip region performed better;accuracy rates for CNN, LSTM, C3D, and I3D on the lip region were 67.12%, 75.53%, 86.32%, and 93.24%, respectively. The 3DCNN-based models achieved higher accuracy due to their ability to process spatio-temporal data. Furthermore, an additional 1.23% improvement was achieved through ensemble learning, with the best result reaching 94.53% accuracy. Ensemble learning, by combining the strengths of different models, provided a meaningful improvement in overall performance. These results demonstrate that 3DCNN architectures and ensemble learning methods yield high success in addressing the problem of lip reading in the Turkish language. While our study focuses on Turkish digit recognition, the proposed methods have the potential to be successful in other languages or broader lip reading applications.

关键词： lip-reading ensemble learning 3DCNN

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：