检索结果-内蒙古大学图书馆

8th IEEE Information technology and Mechatronics Engineering Conference, ITOEC 2025

作者： Li, Jiaxuan Zheng, Qilong School of Computer Science Hefei China National High Performance Computing Center Hefei China

ISBN: (纸本)9798331517915

Loops often dominate the execution time in high-performance computing, effective loop optimization is critical for overall performance. We propose a reinforcement learning-based framework that automatically discovers and composes transformations - including tiling, fusion, interchange, and unrolling - and evaluate the framework on a subset of Polybench benchmarks. Compared to the Polly compiler baseline, our approach achieves an average speedup of 2.46×, peaking at 7× on the jacobi-1d kernel, while also consistently outperforming a global greedy scheduling algorithm. By adaptively combining multiple transformations, the RL-based method exploits deeper synergies with minimal overhead once trained, thus alleviating the repeated manual tuning and hardware-specific adjustments required by conventional techniques. © 2025 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

A Graph-Based Method for AI Chip Operator Optimization Using Deep Learning 6

A Graph-Based Method for AI Chip Operator Optimization Using...

引用

6th IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference, IMCEC 2024

作者： Yan, Jiapeng Zheng, Qilong School of Computer Science Hefei China National High Performance Computing Center Hefei China

ISBN: (纸本)9798350316520

AI operators refer to reusable code programs encapsulated in AI chip frameworks that implement specific functions. To achieve high performance, it is necessary to combine hardware characteristics efficiently when programming. However, for operators with multiple computational steps, it is challenging to develop excellent scheduling strategies. To address this issue, this paper proposes a graph-based method for AI chip operator optimization. Firstly, establish bidirectional transformation relationship between operators and corresponding computation graphs. Then, use deepwalk and word2vec to convert operators' computation graphs into embedding representations, and optimize corresponding operators by annotating the nodes using graph neural network. Also, simple operator fusion can be achieved by fusing graphs of multiple operators and optimizing the fused operator. Through the creation of an operator dataset and related experiments within the Cambricon community framework, this method demonstrates superior optimization and fusion of element-wise operators compared to other simple tuning methods. © 2024 IEEE.

关键词： Graphic methods

来源：评论

学校读者我要写书评

暂无评论

FedDAD: Federated Domain Adaptation for Object Detection

引用

IEEE Access 2023年 11卷 51320-51330页

作者： Lu, Peggy Joy Jui, Chia-Yung Chuang, Jen-Hui National Yang Ming Chiao Tung University Department of Computer Science Hsinchu30010 Taiwan National Center for High-Performance Computing Data Science and Technology Division Taichung40763 Taiwan

Training an object detection model often requires numerous annotated images on a centralized host, which may violate user privacy and data confidentiality. Federated learning (FL) resolves this issue by allowing multiple clients, e.g., cameras, to collaboratively train a model while protecting user privacy. However, models trained with FL may fail to be generalized for new target domain due to domain shift when the data between source and target domains are statistically different. In this work, we formulate a real-world object detection problem as a source-free multi-domain adaptation problem in FL architecture. Moreover, we propose an adaptive FL algorithm, called FedDAD (Federated Domain Adaptive Detector), which aggregates models with dynamic attention targeting the unsupervised domain on server, and utilize instance-level alignment to alleviate the effects of scene variation on clients. Experimental results show that FedDAD improves the average precision (AP) by up to 10.05% and 19.15% compared to the popular FedAvg for specific object classes in the KAIST and MI3 datasets, respectively. © 2013 IEEE.

关键词： Image annotation

来源：评论

学校读者我要写书评

暂无评论

Mobility-Aware Deep Reinforcement Learning with Seq2seq Mobility Prediction for Offloading and Allocation in Edge computing

引用

IEEE Transactions on Mobile computing 2024年第6期23卷 6803-6819页

作者： Wu, Chao-Lun Chiu, Te-Chuan Wang, Chih-Yu Pang, Ai-Chun Academia Sinica Research Center for Information Technology Innovation Taipei115 Taiwan National Tsing Hua University Department of Computer Science Hsinchu300 Taiwan National Taiwan University Institute of Networking and Multimedia Taipei10617 Taiwan National Taiwan University High Performance and Scientific Computing Center Taipei10617 Taiwan

Mobile/multi-access edge computing (MEC) is developed to support the upcoming AI-aware mobile services, which require low latency and intensive computation resources at the edge of the network. One of the most challenging issues in MEC is service provision with mobility consideration. It has been known that the offloading decision and resource allocation need to be jointly handled to optimize the service provision efficiency within the latency constraints, which is challenging when users are in mobility. In this paper, we propose Mobility-Aware Deep Reinforcement Learning (M-DRL) framework for mobile service provision in the MEC system. M-DRL is composed of two parts: glimpse, a seq2seq model customized for mobility prediction to predict a sequence of locations just like a 'glimpse' of the future, and a DRL specialized in supporting offloading decisions and resource allocation in MEC. By integrating the proposed DRL and glimpse mobility prediction model, the proposed M-DRL framework is optimized to handle the MEC service provision with average 70% performance improvements. © 2002-2012 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Automatic APT Attack Reconstruction Supporting Lateral Movement 2nd

Automatic APT Attack Reconstruction Supporting Lateral Movem...

引用

2nd International Conference on Security and Information Technologies with AI, Internet computing and Big-data Applications, SITAIBA 2023

作者： Fan, Chun-I Wu, Wei-Chen Shie, Cheng-Han Kuo, Hsin-Nan Lee, Bo-Yi Department of Computer Science and Engineering National Sun Yat-sen University Kaohsiung Taiwan National Center for High-Performance Computing Hsinchu Taiwan Information Security Research Center National Sun Yat-sen University Kaohsiung Taiwan

ISBN: (纸本)9789819777853

This work proposes a framework for generating datasets that allows users to adjust the APT attack techniques within it. The framework utilizes the MITRE ATT&CK framework to label the attack traffic based on the Tactics, Techniques, and Procedures (TTP), which facilitating researchers in using the datasets for model training. Meanwhile, this work uses the framework to reconstruct the APT29 attack process and collect a dataset to validate the framework’s feasibility. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Advanced Persistent Threat (APT) Dataset generation The APT29 attack Threat intelligence

来源：评论

学校读者我要写书评

暂无评论

A Partitioning and Distributed Caching Approach Based on Adaptive Spectral Clustering for Big Data Streams 13

A Partitioning and Distributed Caching Approach Based on Ada...

引用

2023 13th International Workshop on computer science and Engineering, WCSE 2023

作者： Wang, Shun Zeng, Guo-Sun Department of Computer Science and Technology Tongji University China Tongji Branch National Engineering & Technology Center of High Performance Computer Shanghai201804 China

ISBN: (纸本)9789811879500

Big data streams with diversity are generally processed by parallel computing environments with multiple computational nodes. Before processing, the big data streams need to be partitioned into sub-streams and cached on each computational node for subsequent processing. Existing partitioning methods are difficult to process streams with diversity and high-dimensional characteristics. Partitioning with low quality leads to unreasonable cache placing, which in turn leads to more data migration, lower computational efficiency, and smaller velocity of big data streams on a processing system. Inspired by the advantages of spectral clustering in identifying arbitrary manifolds, an approach of partitioning for big data streams based on spectral clustering is proposed, which transforms the partitioning of streams received during each micro window into the clustering of similarity graphs. We formulate an optimization problem for data items received during each micro window. Then, we present an algorithm to optimize the similarity graphs. With the characteristics of data streams changing gradually in adjacent windows, a distributed caching algorithm based on stream partitioning is presented for continuous windows. Experimental analysis shows that the proposed method can significantly improve the velocity and efficiency of the system for stream processing. © WCSE *** rights reserved.

关键词： Big data

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning for Automated Pragma-Based Loop Optimization

Reinforcement Learning for Automated Pragma-Based Loop Optim...

引用

IEEE Information technology and Mechatronics Engineering Conference (ITOEC)

作者： Jiaxuan Li Qilong Zheng School of Computer Science University of Science and Technology of China (USTC) Hefei China National High Performance Computing Center University of Science and Technology of China (USTC) Hefei China

ISBN: (数字)9798331529482

ISBN: (纸本)9798331529499

Loops often dominate the execution time in high-performance computing, effective loop optimization is critical for overall performance. We propose a reinforcement learning–based framework that automatically discovers and composes transformations—including tiling, fusion, interchange, and unrolling—and evaluate the framework on a subset of Polybench benchmarks. Compared to the Polly compiler baseline, our approach achieves an average speedup of 2.46×, peaking at 7× on the jacobi-1d kernel, while also consistently outperforming a global greedy scheduling algorithm. By adaptively combining multiple transformations, the RL-based method exploits deeper synergies with minimal overhead once trained, thus alleviating the repeated manual tuning and hardware-specific adjustments required by conventional techniques.

关键词： Productivity Mechatronics Scheduling algorithms high performance computing Reinforcement learning Manuals Benchmark testing Kernel Optimization Tuning

来源：评论

学校读者我要写书评

暂无评论

Efficient Collusion-Resisting Secure Sum Protocol

引用

Chinese Journal of Electronics 2023年第3期20卷 407-413页

作者： Youwen ZHU Liusheng HUANG Wei YANG Xing YUAN Department of Computer Science and Technology National High Performance Computing Center at Hefei University of Science and Technology of China Hefei China Suzhou Institute for Advanced Study University of Science and Technology of China Suzhou China

Secure sum protocol is a significant secure multiparty computation protocol and it has various applications in privacy-preserving distributed multiparty computation. However, most existing secure sum protocols rarely considered how to resist underlying collusion which is a significant practical problem. Urabe et al. proposed a collusion-resistant secure sum protocol, but too much cost of communication and computation results in its low performance efficiency. In this paper, we propose security definitions to measure secure multiparty computation protocol's capability of resisting potential collusion. Then, we precisely analyze several previous secure sum protocols' capability of resisting collusion. In addition, considering realistic requirement to resist collusion and performance efficiency needs, we present a novel collusion-resisting secure sum protocol. Theoretical analysis and experimental results confirm that our secure sum protocol is efficient and has strong capability of resisting potential collusion such that it is much superior to previous ones. The communication overheads and computation complexity of our scheme both are linearity of the number of participants. Besides, our protocol's capability of resisting collusion is adjustable according to different security needs.

关键词： Protocols Costs Linearity Resists Computational efficiency Security Computational complexity

来源：评论

学校读者我要写书评

暂无评论

Systematic Survey on Big Data Analytics and Artificial Intelligence for COVID-19 Containment

引用

computer Systems science & Engineering 2023年第11期47卷 1793-1817页

作者： Saeed M.Alshahrani Jameel Almalki Waleed Alshehri Rashid Mehmood Marwan Albahar Najlaa Jannah Nayyar Ahmed Khan Department of Computer Science College Computing and Information TechnologyShaqra UniversityShaqraSaudi Arabia Department of Computer Science College of Computer in Al-LithUmm Al-Qura UniversityMakkahSaudi Arabia High-Performance Computing Center King Abdulaziz UniversityJeddahSaudi Arabia

Artificial Intelligence(AI)has gained popularity for the containment of COVID-19 pandemic *** AI techniques provide efficient mechanisms for handling pandemic *** methods,protocols,data sets,and various validation mechanisms empower the users towards proper decision-making and procedures to handle the *** so many tools,there still exist conditions in which AI must go a long *** increase the adaptability and potential of these techniques,a combination of AI and Bigdata is currently gaining *** paper surveys and analyzes the methods within the various computational paradigms used by different researchers and national governments,such as China and South Korea,to fight against this *** process of vaccine development requires multiple medical *** process requires analyzing datasets from different parts of the *** learning and the Internet of Things(IoT)revolutionized the field of disease diagnosis and disease *** accurate observations from different datasets across the world empowered the process of drug development and drug *** overcome the issues generated by the pandemic,using such sophisticated computing paradigms such as AI,Machine Learning(ML),deep learning,Robotics and Bigdata is essential.

关键词： COVID-19 IoT artificial intelligence big data coronavirus deep learning robotics machine learning

来源：评论

学校读者我要写书评

暂无评论

A Graph-Based Method for AI Chip Operator Optimization Using Deep Learning

A Graph-Based Method for AI Chip Operator Optimization Using...

引用

IEEE Advanced Information Management,Communicates,Electronic and Automation Control Conference (IMCEC)

作者： Jiapeng Yan Qilong Zheng School of Computer Science University of Science and Technology of China (USTC) Hefei China National High Performance Computing Center University of Science and Technology of China (USTC) Hefei China

ISBN: (数字)9798350316537

ISBN: (纸本)9798350316544

AI operators refer to reusable code programs encapsulated in AI chip frameworks that implement specific functions. To achieve high performance, it is necessary to combine hardware characteristics efficiently when programming. However, for operators with multiple computational steps, it is challenging to develop excellent scheduling strategies. To address this issue, this paper proposes a graph-based method for AI chip operator optimization. Firstly, establish bidirectional transformation relationship between operators and corresponding computation graphs. Then, use deepwalk and word2vec to convert operators’ computation graphs into embedding representations, and optimize corresponding operators by annotating the nodes using graph neural network. Also, simple operator fusion can be achieved by fusing graphs of multiple operators and optimizing the fused operator. Through the creation of an operator dataset and related experiments within the Cambricon community framework, this method demonstrates superior optimization and fusion of element-wise operators compared to other simple tuning methods.

关键词： Deep learning Training Codes Computational modeling AI accelerators Programming Rendering (computer graphics)

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：