检索结果-内蒙古大学图书馆

Prototype Memory for Large-scale Face Representation Learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Smirnov, Evgeny Garaev, Nikita Galyuk, Vasiliy Lukyanets, Evgeny Speech Technology Center Saint Petersburg194044 Russia Department of Information Technologies and Programming ITMO University Saint Petersburg197101 Russia

Face representation learning using datasets with a massive number of identities requires appropriate training methods Softmax-based approach, currently the state-of-the-art in face recognition, in its usual "full softmax" form is not suitable for datasets with millions of persons Several methods, based on the "sampled softmax" approach, were proposed to remove this limitation These methods, however, have a set of disadvantages One of them is a problem of "prototype obsolescence": classifier weights (prototypes) of the rarely sampled classes receive too scarce gradients and become outdated and detached from the current encoder state, resulting in incorrect training signals This problem is especially serious in ultra-large-scale datasets In this paper, we propose a novel face representation learning model called Prototype Memory, which alleviates this problem and allows training on a dataset of any size Prototype Memory consists of the limited-size memory module for storing recent class prototypes and employs a set of algorithms to update it in appropriate way New class prototypes are generated on the fly using exemplar embeddings in the current mini-batch These prototypes are enqueued to the memory and used in a role of classifier weights for softmax classification-based training To prevent obsolescence and keep the memory in close connection with the encoder, prototypes are regularly refreshed, and oldest ones are dequeued and disposed of Prototype Memory is computationally efficient and independent of dataset size It can be used with various loss functions, hard example mining algorithms and encoder architectures We prove the effectiveness of the proposed model by extensive experiments on popular face recognition benchmarks. © 2021, CC BY.

关键词： Face recognition

Symbolic Security Predicates: Hunt Program Weaknesses

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Vishnyakov, Alexey Logunova, Vlada Kobrin, Eli Kuts, Daniil Parygina, Darya Fedotov, Andrey Ivannikov Institute for System Programming of the RAS Russia Lomonosov Moscow State University Russia Moscow Institute of Physics and Technology Moscow Russia

Dynamic symbolic execution (DSE) is a powerful method for path exploration during hybrid fuzzing and automatic bug detection. We propose security predicates to effectively detect undefined behavior and memory access violation errors. Initially, we symbolically execute program on paths that don’t trigger any errors (hybrid fuzzing may explore these paths). Then we construct a symbolic security predicate to verify some error condition. Thus, we may change the program data flow to entail null pointer dereference, division by zero, out-of-bounds access, or integer overflow weaknesses. Unlike static analysis, dynamic symbolic execution does not only report errors but also generates new input data to reproduce them. Furthermore, we introduce function semantics modeling for common C/C++ standard library functions. We aim to model the control flow inside a function with a single symbolic formula. This assists bug detection, speeds up path exploration, and overcomes overconstraints in path predicate. We implement the proposed techniques in our dynamic symbolic execution tool Sydr. Thus, we utilize powerful methods from Sydr such as path predicate slicing that eliminates irrelevant constraints. We present Juliet Dynamic to measure dynamic bug detection tools accuracy. The testing system also verifies that generated inputs trigger sanitizers. We evaluate Sydr accuracy for 11 CWEs from Juliet test suite. Sydr shows 95.59% overall accuracy. We make Sydr evaluation artifacts publicly available to facilitate results reproducibility. © 2021, CC BY-NC-ND.

关键词： Static analysis

Stochastic Gradient Methods with Preconditioned Updates

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Sadiev, Abdurakhmon Beznosikov, Aleksandr Almansoori, Abdulla Jasem Kamzolov, Dmitry Tappenden, Rachael Takáč, Martin Ivannikov Institute for System Programming Moscow Russia Moscow Institute of Physics and Technology Moscow Russia Mohamed bin Zayed University of Artificial Intelligence Abu Dhabi United Arab Emirates University of 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. Copyright © 2022, The Authors. All rights reserved.

关键词： Gradient methods

Interference Mitigation Using Uplink Power Control in 5G Relay-Based Heterogeneous Networks 2

学校读者我要写书评

暂无评论

Interference Mitigation Using Uplink Power Control in 5G Rel...

2nd International Conference on Emerging Smart Technologies and Applications, eSmarTA 2022

作者： Sultan, Jamil Jabbar, Waheb A. Al-Thobhani, Nashwan S. Al-Hetar, Abdulaziz Saif, Mobarak Sana'a Community College Telecommunication Engineering Technology Deptartment Yemen University of Modern Sceinces Computer Network Eng Dept Sana'a Yemen Birmingham City University School of Engineering and Built Environment BirminghamB4 7XG United Kingdom University of Modern Sciences Faculty of Engineering Sana'a Yemen School of Engineering Sana'a Yemen Sana'a Community College Computer Application Programming Dept Sana'a Yemen

ISBN: (数字)9781665461818

ISBN: (纸本)9781665461818

Heterogeneous network (HetNet) is an attractive solution for future cellular networks with high data rate and coverage requirements. In HetNets, small cells such as micro cells, pico cells, femto cells and relay node (RN) are added to the network of macro cells in the same region. A large number of low power RNs produces new cell edges with significant intra-cell and inter-cell interferences. In the uplink (UL) scenarios of time-division based HetNets with RN, the user equipment (UE) desired signal may be interfered by the transmissions of the co-channel UEs during the first time slot and by the transmissions of the co-channel UEs or RNs during the second time slot. The interference caused by the RNs may significantly degrade the UE signal. UL transmission power control (PC) is essential for mitigating interference and, as a result, enhancing the cell edge and overall system performance. This research proposes a PC algorithm in order to mitigate the UL interference in 5g relay-based HetNets. This research also investigates the UL performance of HetNets when PC is applied at the RNs. Simulation results indicate that UL PC at the RNs greatly reduces average interference and improves average UL signal-to-interference-plus-noise ratio (SINR) and average UL end-to-end throughput compared to the situation in which UL PC is not implemented. © 2022 IEEE.

关键词： Heterogeneous networks

Energy Efficiency Improvement System for AC Circuits

学校读者我要写书评

暂无评论

Energy Efficiency Improvement System for AC Circuits

IEEE NW Russia Young Researchers in Electrical and Electronic Engineering Conference (EIConRusNW)

作者： Vladimir Kulagin Kirill Selivanov Andrey Vlasov Dmitrii Aminev Lyudmila Bunina Sergey Bikovsky Moscow technical university of communications and informatics Moscow Russian Federation Department of Design and Technology of Electronic Devices Bauman Moscow State Technical University BMSTU Moscow Russian Federation Department of Development of software solutions and system programming Russian Technological University – MIREA Moscow Russian Federation Department of Electrical Engineering Systems Russian Technological University – MIREA Moscow Russian Federation

The problem of the rational use of energy resources remains constantly relevant and requires the search for new approaches. One of them is power control. In AC circuits, the authors see the most promising method of phase AC power control. Based on it, a power control module was developed. The structural and circuit diagrams of the developed device are presented to implement the proposed solution. The authors produced its experimental prototype and conducted experimental testing at various levels of regulated power. Savings when using the proposed power control module were calculated using the example of energy consumption in various sectors of energy-intensive systems. According to the data obtained, the module allows saving energy consumption without significant discomfort to the energy user. The calculated results allow concluding that the proposed AC power control module is fully operational, and its widespread use will significantly reduce the need for electricity.

关键词：

HIGHLY SMOOTH ZEROTH-ORDER METHODS FOR SOLVING OPTIMIZATION PROBLEMS UNDER THE PL CONDITION†

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Gasnikov, A.V. Lobanov, A.V. Stonyakin, F.S. Moscow Institute of Physics and Technology 141700 Dolgoprudny Institutskiy per. 9 Russia Skolkovo Institute of Science and Technology 121205 Moscow B. Boulevard 30 bld. 1 Russia Institute for System Programming of the RAS 125047 Moscow A. Solzhenitsyn st. 25 Russia V.I. Vernadsky Crimean Federal University 295007 Simferopol Prospekt Vernadskogo 4 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. Copyright © 2023, The Authors. All rights reserved.

关键词： Gradient methods

The Study of the Dispersion and Phase Velocity of Love Waves 41

学校读者我要写书评

暂无评论

The Study of the Dispersion and Phase Velocity of Love Waves

41st IEEE International Conference on Electronics and Nanotechnology, ELNANO 2022

作者： Porkuian, Olga Morkun, Volodymyr Morkun, Natalia Pikilnyak, Andrii Gaponenko, Alena Gaponenko, Irina Bobrov, Evhen Vesnin, Artem Volodymyr Dahl East Ukrainian National University Department of Programming and Mathematics Severodonetsk Ukraine Kryvyi Rih National University Department of Automation Computer Science and Technology Kryvyi Rih Ukraine Kryvyi Rih National University Department of Mechanical Engineering Kryvyi Rih Ukraine Kryvyi Rih National University R&d Department Kryvyi Rih Ukraine Kryvyi Rih National University Department of Motor Transport Kryvyi Rih Ukraine

ISBN: (数字)9781665469227

ISBN: (纸本)9781665469234

The research results of the Love wave propagation in the semi-space contact area and in a thin layer are presented. The dependences, which allow analyzing the effect of various conditions on the Love wave dispersion, as well as the dependence of the phase velocity on various parameters: the thickness of the layer, which is coated in semi-space;the relationship between the material characteristics of the media (densities, elastic moduli) are defined. The influence of film thickness on the value of phase velocity at a given frequency, but at different ratios between the density of the media are studied. © 2022 IEEE.

关键词： Dispersion (waves)

Context Classification in Dialog-Based Interaction

学校读者我要写书评

暂无评论

Context Classification in Dialog-Based Interaction

International Conference on Semantic Computing

作者： Alexander Wachtel Felix Eurich Dominik Fuchß Walter F. Tichy Chair for Programming Systems Karlsruhe Institute of Technology Karlsruhe Germany Karlsruhe Institute of Technology

ISBN: (数字)9781728163321

ISBN: (纸本)9781728163338

A difficulty in processing of the natural language is recognizing the context of a statement. However, since this contains implicit knowledge which we unconsciously use in our formulations, it is necessary to assign the context for correct interpretation. A context in our system is managed by a service that interprets and processes input and provides feedback to the end user. In this paper, we present a solution how an end user input can be assigned to such a service. For this purpose, we score the end user inputs by the system with an ensemble of 6 different classifiers that consider semantics as well as syntax. The system learns the user's input at run time and adapts his enunciation step by step. During the evaluation, the system was able to classified 87% of the user statements to the correct service. Far from perfect, this research might lead to fundamental changes in computer use.

关键词： Natural languages Syntactics History Semantics programming Software Unified modeling language

The use of incentives to promote Technical Debt management

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Besker, Terese Martini, Antonio Bosch, Jan Computer Science and Engineering Software Engineering Chalmers University of Technology Göteborg Sweden University of Oslo Programming and Software Engineering Oslo Norway

Context: When developing software, it is vitally important to keep the level of technical debt down since it is well established from several studies that technical debt can, e.g., lower the development productivity, decrease the developers' morale, and compromise the overall quality of the software. However, even if researchers and practitioners working in today's software development industry are quite familiar with the concept of technical debt and its related negative consequences, there has been no empirical research focusing specifically on how software managers actively communicate and manage the need to keep the level of technical debt as low as possible. Objective: This study aims to understand how software companies give incentives to manage Technical Debt. This is done by exploring how companies encourage and reward practitioners for actively keeping the level of technical debt down and whether the companies use any forcing or penalizing initiatives when managing technical debt. Method: In a first step, this paper reports the results of both an online survey provided quantitative data from 258 participants and interviews with 32 software practitioners. In a second step, this study set out to specifically provide a detailed assessment of additional and in-depth analysis of Technical Debt management strategies based on an encouraging mindset and attitude from both managers and technical roles to understand how, when and by whom such strategy is adopted in practice. Results: Our findings show that having a Technical Debt management strategy (specially based on encouragement) can significantly impact the amount of Technical Debt in the software. Conclusion: The result indicates that there is considerable unfulfilled potential to influence how software practitioners can further limit and reduce Technical Debt by adopting a strategy based explicitly on an encouraging mindset from managers where they also specifically dedicate time and resources for Technical De

关键词： Software design