the proceedings contain 32 papers. the special focus in this conference is on High-Performance computing Systems and technologies in Scientific Research, Automation of Control and Production. the topics include: Mathe...
ISBN:
(纸本)9783030941406
the proceedings contain 32 papers. the special focus in this conference is on High-Performance computing Systems and technologies in Scientific Research, Automation of Control and Production. the topics include: Mathematical Simulation of Coupled Elastic Deformation and Fluid Dynamics in Heterogeneous Media;numerical Modeling of Electric and Magnetic Fields Induced by External Source in Frequency Domain;improving the Heterogeneous computing Node Performance of the Desktop Grid When Searching for Orthogonal Diagonal Latin Squares;visual Metamodeling with Verification Based on Surrogate Modeling for Adaptive computing;recognition Algorithms Based on the Selection of 2D Representative Pseudo-objects;applied Interval Analysis of Big Data Using Linear Programming Methods;parallelcomputing in Problems of Classification of Teenagers Based on Analysis of Digital Traces;identification of Key Players in a Social Media Based on the Kendall-Wei Ranking;using Time Series and New Information technologies for Forecasting Sugarcane Production Indicators;Calculation of Activation Functions in FPGA-Based Neuroprocessors Using the Cordic Algorithm;software and Methodology for the Design of System Dynamics Models Based on the Situation-Activity Approach;architecture of an Intelligent Network Pyrometer for Building Information-Measuring and Mechatronic Systems;implementation of a Network-Centric Production Storage System in distributed High-Performance computing Systems;internet of things for Reducing Commercial Losses from Incorrect Activities of Personnel;method of Constructing the Assigned Trajectory of a Multi-link Manipulator Based on the "Programming by Demonstration" Approach;developing a Microprocessor-Based Equipment Control Panel for Rifle Sports Complexes;methods for Domain Adaptation of Automated Systems for Aspect Annotation of Customer Review Texts;testing Methods for Blockchain applications.
this work is devoted to an actual survey on advanced security and ensured user privacy for (Highly-) distributed Systems. the last term belongs to thin and thick apps, robot and mobile apps, (micro-)service-oriented a...
详细信息
ISBN:
(数字)9789532901351
ISBN:
(纸本)9798350390797
this work is devoted to an actual survey on advanced security and ensured user privacy for (Highly-) distributed Systems. the last term belongs to thin and thick apps, robot and mobile apps, (micro-)service-oriented applications, and IoT applications. the most dangerous vulnerabilities, intrusion analysis techniques and models, and countermeasures to increase the safety of the cyber-systems are discussed. AI-based methods are favored nowadays. However, generative models provide some new risks and vulnerabilities. EU legal regulations are considered. Several case studies (among others, for telemedicine and e-health) are examined.
In recent years, fuzzing has become the most popular and effective vulnerability mining technique due to high degree of automation and versatility. In order to improve the characteristics of fuzz testing such as blind...
In recent years, fuzzing has become the most popular and effective vulnerability mining technique due to high degree of automation and versatility. In order to improve the characteristics of fuzz testing such as blindness and inefficiency, a large number of studies have been conducted to optimize the design of each step. Since fuzzing is typically a computationally intensive process, and the performance improved by algorithm optimization is always limited on a single machine, parallelcomputing to improve the performance of fuzzing is of great research value. However, parallelization of fuzzing must face and overcome challenges such as task conflicts, scalability in distributed environments, data synchronization overhead and workload imbalance. In this paper, we set out to solve each challenge in parallelized fuzzing and propose ParaFuzz, a new parallelized fuzzing tool. ParaFuzz manage and distribute seeds centrally in a client/server architecture to avoid task conflicts, solve workload balancing problems through a request/response model. A unique global sharing mechanism is designed for different information characteristics, and a variant strategy selection mechanism is proposed to improve the efficiency of fuzzing at the task scheduling level. Results from ParaFuzz tests on the LAVA-M test set and two real-world applications show up to 66% improvement in code path discovery compared to AFL native parallel mode at 8-node parallel scale.
there are a huge amount of scientific and commercial applications written with a focus on sequential execution. Running such programs on multiprocessor systems is possible, but without taking advantage of these system...
详细信息
Data parallel frameworks become essential for training machine learning models. the classic Bulk Synchronous parallel (BSP) model updates the model parameters through pre-defined synchronization barriers. However, whe...
详细信息
ISBN:
(纸本)9781450391641
Data parallel frameworks become essential for training machine learning models. the classic Bulk Synchronous parallel (BSP) model updates the model parameters through pre-defined synchronization barriers. However, when a worker computes significantly slower than other workers, waiting for the slow worker will lead to excessive waste of computing resources. In this paper, we propose a novel proactive data-parallel (PDP) framework. PDP enables the parameter server to initiate the update of the model parameter. that is, we can perform the update at any time without pre-defined update points. PDP not only initiates the update but also determines when to update. the global decision on the frequency of updates will accelerate the training. We further propose asynchronous PDP to reduce the idle time caused by synchronizing parameter updates. We theoretically prove the convergence property of asynchronous PDP. We implement a distributed PDP framework and evaluate PDP with several popular machine learning algorithms including Multilayer Perceptron, Convolutional Neural Network, K-means, and Gaussian Mixture Model. Our evaluation shows that PDP can achieve up to 20X speedup over the BSP model and scale to large clusters.
Withthe rapid development of artificial intelligence technology, the need of large model training is becoming more and more urgent. However, the traditional large model training method has high communication cost, re...
详细信息
ISBN:
(数字)9798350352214
ISBN:
(纸本)9798350352221
Withthe rapid development of artificial intelligence technology, the need of large model training is becoming more and more urgent. However, the traditional large model training method has high communication cost, resulting in slow training speed and low efficiency. To solve this problem, this paper proposes a distributed large model training efficient communication scheme (DCoPa-DL) based on data compression and parallel communication. By compressing the training data and model gradient, and adopting parallel communication technology, the scheme effectively reduces the communication overhead and significantly improves the training speed. By verifying on the science and technology innovation infrastructure, the results show that the DCoPa-DL scheme can shorten the training time about 50%, compared withthe original scheme, increase the training throughput by about 1 times, and slightly improve the model accuracy.
A scalable bandwidth-adaptive on-chip storage network architecture is proposed to address the severe data conflict and low bus parallelism in existing multi-level storage, Crossbar, and NoC architectures in edge accel...
详细信息
ISBN:
(数字)9798331509453
ISBN:
(纸本)9798331520243
A scalable bandwidth-adaptive on-chip storage network architecture is proposed to address the severe data conflict and low bus parallelism in existing multi-level storage, Crossbar, and NoC architectures in edge accelerated computing. this architecture combines the advantages of centralized and distributed memory, allowing the memory hierarchy to be flexibly adapted based on current computational needs and memory access patterns to better accommodate multi-core processors and parallelcomputing requirements. this paper designs the result visualization method to visualize the stress of storing each node and bus. Furthermore, the full path test incentive is intended for the secondary sharing policy and router. the feasibility of this paper's architecture is verified through VCS simulation and visualization analysis.
In this paper, we study how to assemble a supply chain withthe avail of Blockchain to establish a secure trading environment. Blockchain is distributed, decentralized and a transparent mechanism which makes it suitab...
详细信息
parallel methods for solving saddle-type algebraic systems that are relevant for modeling processes and phenomena in the problems of electromagnetism, hydro-gas dynamics, elastoplasticity, filtration and other applica...
详细信息
Data security sharing in the Jointcloud has always been a difficult problem for researchers. Some cloud data sharing solutions have a series of problems, such as high communication overhead, difficult key management, ...
详细信息
ISBN:
(纸本)9781665434799
Data security sharing in the Jointcloud has always been a difficult problem for researchers. Some cloud data sharing solutions have a series of problems, such as high communication overhead, difficult key management, data tampering, single cloud failure, etc. We propose a data-sharing mechanism based on blockchain and threshold proxy re-encryption to solve the problems listed above. the proxy re-encryption algorithm we adopted is improved based on the Conditional Proxy Broadcast Re-Encryption. By combining the secret sharing scheme to realize trust decentralization. the re-encryption computing is delivered to multiple cloud service providers, which reduces the centralization problem caused by single cloud processing. At the same time, blockchain is used to manage metadata information and data access control rights to solve data loss and tampering problems. Experimental results show that our mechanism achieves low re-encryption computing overhead.
暂无评论