检索结果-内蒙古大学图书馆

Exploiting the Memory-Compute-Coupling Feature for CIM Accelerator Design Optimization

IEEE Transactions on computer-Aided Design of Integrated Circuits and Systems 2025年

作者： Wu, Yongkun Wang, Xiaomeng Chen, Jia Zhu, Zhenhua He, Jingyu Dong, Pingcheng Tan, Yonghao Zhao, Xin Chang, Liang Wang, Yu Tu, Fengbin Tsui, Chi-Ying Cheng, Kwang-Ting The Hong Kong University of Science and Technology Department of Electronic and Computer Engineering Hong Kong The Hong Kong University of Science and Technology AI Chip Center for Emerging Smart Systems (ACCESS) Hong Kong Hong Kong University of Science and Technology Hong Kong Tsinghua University Beijing 100084 China University of Electronic Science and Technology of China School of Information and Communication Engineering Chengdu 611731 China Tsinghua University Department of Electronic Engineering Beijing 100084 China

SRAM computing-in-memory (CIM) accelerators have evolved as a promising solution to the memory wall problem in neural network (NN) models. By integrating memory and compute resources in each macro, CIM accelerators offer massive in-situ computing parallelism and large memory capacity, enabling spatial mapping with layer fusion and potentially keeping layers stationary in CIM. However, CIM’s memory-compute coupling (MCC) feature poses challenges in designing CIM accelerators. From an architecture aspect, designers must balance CIM’s memory and compute resources by optimizing the macro’s memory-compute ratio (MCR) configuration across diverse scenarios. From a mapping aspect, conventional mappings, which allocate each macro exclusively to each layer, face two major problems: a layer-fusion dilemma (the accelerator suffers from excessive memory access due to layer replications or performance degradation due to load imbalance) and a layer-eviction issue (storing layers stationary in CIM is usually infeasible due to limited CIM capacity). To address these challenges, this paper introduces MCC-DSE, an MCC-aware Design Space Exploration framework for architecture-mapping co-optimization of CIM accelerators. We also propose a three-axis CIM division mapping, which interleaves multiple layers in each macro to concurrently optimize memory access and performance during layer fusion as well as reserves a part of CIM memory in each macro for layer pinning. Compared to baseline architecture and mapping, MCC-DSE shows a 1.4x∼8.3x EDP reduction across various workloads and chip areas. Moreover, MCC-DSE provides insights into CIM accelerator optimization, such as selecting optimal MCR and configuring CIM dynamically for different scenarios. © 1982-2012 IEEE.

关键词： Computing-in-memory design space exploration mapping memory-compute coupling

来源：评论

学校读者我要写书评

暂无评论

Seamless Graph Task Scheduling over Dynamic Vehicular Clouds: A Hybrid Methodology for Integrating Pilot and Instantaneous Decisions

引用

IEEE Transactions on Services Computing 2025年

作者： Guo, Bingshuo Liwang, Minghui Xia, Xiaoyu Li, Li Jiao, Zhenzhen Hosseinalipour, Seyyedali Wang, Xianbin Xiamen University School of Informatics Fujian China Rmit University School of Computing Technologies MelbourneVIC Australia Tongji University Department of Control Science and Engineering Shanghai China Beijing Teleinfo Technology Co. Ltd. Caict IF-Labs China University at Buffalo-SUNY Department of Electrical Engineering NY United States Western University Department of Electrical and Computer Engineering ON Canada

Vehicular clouds (VCs) play a crucial role in the Internet-of-Vehicles (IoV) ecosystem by securing essential computing resources for a wide range of tasks. This paPertackles the intricacies of resource provisioning in dynamic VCs for computation-intensive tasks, represented by undirected graphs for parallel processing over multiple vehicles. We model the dynamics of VCs by considering multiple factors, including varying communication quality among vehicles, fluctuating computing capabilities of vehicles, uncertain contact duration among vehicles, and dynamic data exchange costs between vehicles. Our primary goal is to obtain feasible assignments between task components and nearby vehicles, called templates, in a timely manner with minimized task completion time and data exchange overhead. To achieve this, we propose a hybrid graph task scheduling (P-HTS) methodology that combines offline and online decision-making modes. For the offline mode, we introduce an approach called risk-aware pilot isomorphic subgraph searching (RA-PilotISS), which predicts feasible solutions for task scheduling in advance based on historical information. Then, for the online mode, we propose time-efficient instantaneous isomorphic subgraph searching (TE-InstaISS), serving as a backup approach for quickly identifying new optimal scheduling template when the one identified by RA-PilotISS becomes invalid due to changing conditions. Through comprehensive experiments, we demonstrate the superiority of our proposed hybrid mechanism compared to state-of-the-art methods in terms of various evaluative metrics, e.g., time efficiency such as the delay caused by seeking for possible templates and task completion time, as well as cost function, upon considering different VC scales and graph task topologies. © 2025 IEEE.

关键词： Undirected graphs

来源：评论

学校读者我要写书评

暂无评论

Cancer type and survival prediction based on transcriptomic feature map

引用

computers in Biology and Medicine 2025年第Pt A期192卷 110267-110267页

作者： Yan, Ming Dong, Zirou Zhu, Zhaopo Qiao, Chengliang Wang, Meizhi Teng, Zhixia Xing, Yongqiang Liu, Guojun Liu, Guoqing Cai, Lu Meng, Hu Inner Mongolia Key Laboratory of Life Health and Bioinformatics College of Life Science and Technology Inner Mongolia University of Science and Technology Baotou014010 China College of Information and Computer Engineering Northeast Forestry University Harbin China Center for Medical Genetics & Hunan Key Laboratory School of Life Sciences Central South University Huna Changsha410008 China

This study achieved cancer type and survival time prediction by transforming transcriptomic features into feature maps and employing deep learning models. Using transcriptomic data from 27 cancer types and survival data from 10 types in the TCGA database, a pan-cancer transcriptomic feature map was constructed through data cleaning, feature extraction, and visualization. Using Inception network and gated convolutional modules yielded a pan-cancer classification accuracy of 91.8 %. Additionally, by extracting 31 differential genes from different cancer feature maps, an interaction network diagram was drawn, identifying two key genes, ANXA5 and ACTB. These genes are potential biomarkers related to cancer progression, angiogenesis, metastasis, and treatment resistance. Survival prediction analysis on 10 cancer types, combined with feature maps and data amplification, cancer survival prediction accuracy reached from 0.75 to 0.91. This transcriptomic feature map provides a novel approach for cancer omics analysis, to facilitate personalized treatments and reflecting individual differences. © 2025 The Authors

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Integrating Dynamic Access Control in Hybrid Cloud Environments for Enhanced Security and Load Balancing

Integrating Dynamic Access Control in Hybrid Cloud Environme...

引用

3rd IEEE International Conference on Automation and Computation, AUTOCOM 2025

作者： Swarup, Lakshya Raghotham, Saritha Srinivasmurthy Singh, B.P. Albert Mayan, J. Aneja, Aseem Barisal, Swadhin Ku. Chitkara University Chitkara Centre for Research and Development Himachal Pradesh Baddi 174103 India School of Management - Ug Management Jain (Deemed to be University) Karnataka Bangalore India Maharishi School of Engineering & Technology Maharishi University of Information Technology Uttar Pradesh India Sathyabama Institute of Science and Technology Department of Computer Science and Engineering Tamil Nadu Chennai India Chitkara University Centre of Research Impact and Outcome Punjab Rajpura 140417 India Siksha 'O' Anusandhan (Deemed to be University) Department of Computer Applications Odisha Bhubaneswar India

ISBN: (纸本)9798331542375

Hybrid cloud environments come into play more each month, this service is becoming increasingly important for security and load balancing. Hybrid cloud environments are too fluid to manage with traditional access control methods and load-balancing solutions. This is where dynamic access control integration comes in. Dynamic access control is a process that enables organizations to define and enforce access controls based on the context of the request, not merely user identities. This could be applied in a hybrid cloud environment to implement access controls based on the whereabouts of users, the kind of data, and the threat level. This is intended to add an extra layer of security as we can limit the users who have permission to access specific resources. You can add dynamic access control to your load-balancing solutions for resource efficiency and performance improvements. It also reduces the possibility of service outage, as more accurate granular access control based on lower visibility to higher certain users making fewer requests at a time than another set can simplify algorithms that tombstone or down-replicated out-of-date servers and thus carve their load across racks of machines. © 2025 IEEE.

关键词： Efficiency Hybrid Load Balancing Lower Visibility Specific Resources

来源：评论

学校读者我要写书评

暂无评论

Personalized Federated Learning for Cellular VR: Online Learning and Dynamic Caching

arXiv

引用

arXiv 2025年

作者： Tharakan, Krishnendu S. Dahrouj, Hayssam Kouzayha, Nour ElSawy, Hesham Al-Naffouri, Tareq Y. The School of Computing Queen's University Kingston Canada The Department of Electrical Engineering University of Sharjah Sharjah United Arab Emirates The Division of Computer Electrical and Mathematical Sciences and Engineering King Abdullah University of Science and Technology Thuwal23955-6900 Saudi Arabia

Delivering an immersive experience to virtual reality (VR) users through wireless connectivity offers the freedom to engage from anywhere at any time. Nevertheless, it is challenging to ensure seamless wireless connectivity that delivers real-time and high-quality videos to the VR users. This paper proposes a field of view (FoV) aware caching for mobile edge computing (MEC)-enabled wireless VR network. In particular, the FoV of each VR user is cached/prefetched at the base stations (BSs) based on the caching strategies tailored to each BS. Specifically, decentralized and personalized federated learning (DP-FL) based caching strategies with guarantees are presented. Considering VR systems composed of multiple VR devices and BSs, a DP-FL caching algorithm is implemented at each BS to personalize content delivery for VR users. The utilized DP-FL algorithm guarantees a probably approximately correct (PAC) bound on the conditional average cache hit. Further, to reduce the cost of communicating gradients, one-bit quantization of the stochastic gradient descent (OBSGD) is proposed, and a convergence guarantee of O(1/√T) is obtained for the proposed algorithm, where T is the number of iterations. Additionally, to better account for the wireless channel dynamics, the FoVs are grouped into multicast or unicast groups based on the number of requesting VR users. The performance of the proposed DP-FL algorithm is validated through realistic VR head-tracking dataset, and the proposed algorithm is shown to have better performance in terms of average delay and cache hit as compared to baseline algorithms. © 2025, CC BY.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Advanced Cryptocurrency Price Prediction: Incorporating Time-Reversed ARIMA and BVIC

Advanced Cryptocurrency Price Prediction: Incorporating Time...

引用

International Conference on Ubiquitous Information Management and Communication (IMCOM)

作者： Minh Tuan Luc Roman Senkerik Tran Khanh Dang Dept. of Computer Science Faculty of Electrical Engineering and Computer Science VSB-Technical University of Ostrava Ostrava-Poruba Czech Repulic Center for Applied Information Technology Ton Duc Thang University Ho Chi Minh city Vietnam Institute for Experiential Technology Van Lang University Ho Chi Minh city Vietnam Faculty of Information Technology School of Technology Van Lang University Ho Chi Minh city Vietnam

ISBN: (数字)9798331507817

ISBN: (纸本)9798331507824

The highly volatile nature of cryptocurrencies, particularly Bitcoin, exhibits extreme price volatility, making accurate price prediction a challenging task for traditional forecasting models. This paper proposes an enhanced ARIMA model, called ARIMA_T, which incorporates time-reversed data and optimizes model selection using the Backward Validated Information Criterion (BVIC). The ARIMA_T model aims to improve forecasting accuracy by capturing both linear and non-linear patterns in the data that standard ARIMA models often overlook. We apply this model to predict Bitcoin prices over short and medium-term horizons, specifically 1-day, 7-days, and 30-days ahead predictions. The dataset spans from April 2013 to February 2024. Experimental results demonstrate that ARIMA_T consistently outperforms the traditional ARIMA model across all forecast horizons, as evidenced by lower Root Mean Square Error (RMSE) values. The ARIMA_T model provides more accurate and robust predictions, particularly for short-term forecasting, making it a valuable tool for navigating the volatile cryptocurrency market. This study also discusses the limitations of both models in handling longer-term predictions and suggests future work in integrating ARIMA_T with machine learning techniques for further improvements.

关键词： Accuracy Uncertainty Navigation Bitcoin Predictive models Data models Robustness Forecasting Root mean square Standards

来源：评论

学校读者我要写书评

暂无评论

NI-GDBA: Non-Intrusive Distributed Backdoor Attack Based on Adaptive Perturbation on Federated Graph Learning 25

NI-GDBA: Non-Intrusive Distributed Backdoor Attack Based on ...

引用

Proceedings of the ACM on Web Conference 2025

作者： Ken Li Bin Shi Jiazhe Wei Bo Dong School of Computer Science and Technology Xi'an Jiaotong University Xi'an China and Ministry of Education Key Laboratory of Intelligent Networks and Network Security Xi'an Jiaotong University Xi'an China School of Computer Science and Technology Xi'an Jiaotong University Xi'an China and Shaanxi Province Key Laboratory of Big Data Knowledge Engineering Xi'an Jiaotong University Xi'an China School of Distance Education Xi'an Jiaotong University Xi'an China and Shaanxi Province Key Laboratory of Big Data Knowledge Engineering Xi'an Jiaotong University Xi'an China

ISBN: (纸本)9798400712746

Federated Graph Learning (FedGL) is an emerging Federated Learning (FL) framework that learns the graph data from various clients to train better Graph Neural Networks(GNNs) model. Owing to concerns regarding the security of such framework, numerous studies have attempted to execute backdoor attacks on FedGL, with a particular focus on distributed backdoor attacks. However, all existing methods posting distributed backdoor attack on FedGL only focus on injecting distributed backdoor triggers into the training data of each malicious client, which will cause model performance degradation on original task and is not always effective when confronted with robust federated learning defense algorithms, leading to low success rate of attack. What's more, the backdoor signals introduced by the malicious clients may be smoothed out by other clean signals from the honest clients, which potentially undermining the performance of the *** address the above significant shortcomings, we propose a non-intrusive graph distributed backdoor attack(NI-GDBA) that does not require backdoor triggers to be injected in the training data. Our attack trains an adaptive perturbation trigger generator model for each malicious client to learn the natural backdoor from the GNN model downloading from the server with the malicious client's local data. In contrast to traditional distributed backdoor attacks on FedGL via trigger injection in training data, our attack on different datasets such as Molecules and Bioinformatics have higher attack success rate, stronger persistence and stealth, and has no negative impact on the performance of the global GNN model. We also explore the robustness of NI-GDBA under different defense strategies, and based on our extensive experimental studies, we show that our attack method is robust to current federated learning defense methods, thus it is necessary to consider non-intrusive distributed backdoor attacks on FedGL as a novel threat that requires custom de

关键词： backdoor attacks

来源：评论

学校读者我要写书评

暂无评论

Integrating Blockchain technology for Secure and Decentralized Network Communication

Integrating Blockchain Technology for Secure and Decentraliz...

引用

3rd IEEE International Conference on Automation and Computation, AUTOCOM 2025

作者： Nagpal, Manish Chakraborty, Srija Gupta, Mohit Raghu, N. Vigneshwari, S. Chaudhary, Pavan Centre of Research Impact and Outcome Chitkara University Punjab Rajpura 140417 India Centre for Data Sciences Siksha 'O' Anusandhan (Deemed to be University) Odisha Bhubaneswar India Chitkara Centre for Research and Development Chitkara University Himachal Pradesh Baddi 174103 India School of Engineering and Technology Jain (Deemed to be University) Karnataka Bangalore India Sathyabama Institute of Science and Technology Department of Computer Science and Engineering Tamil Nadu Chennai India Maharishi School of Engineering & Technology Maharishi University of Information Technology Uttar Pradesh India

ISBN: (纸本)9798331542375

It is a distributed and decentralized database that protects transparent transaction records. While this technology has been mainly related to crypto currencies, it has a potential application in improving network communication. For instance, the centralized nature of communication can make room for vulnerabilities, including data leakage and single points of failure in a traditional client-server model. You will not have a single point of failure in network effects as communication can be decentralized with block chain. Block chain technology is distributed by nature. Hence, the data is shared and stored on multiple nodes, making it almost impossible for an attacker to modify it. Lastly, Block chain technology offers encryption and authentication for safe device communication without intermediaries. This mitigates the chance of man-in-the-middle attacks and thus helps with data confidentiality and integrity. Besides, using smart contracts in block chain allows attendees to communicate as much as possible without human interference. The introduction of block chain technology in network communication can provide a secure and decentralized solution between client nodes and servers to maintain data privacy and integrity. © 2025 IEEE.

关键词： Communication Integrity Interference Transparent Transaction Vulnerabilities

来源：评论

学校读者我要写书评

暂无评论

CLEARING SECTIONS OF LATTICE LIABILITY NETWORKS

arXiv

引用

arXiv 2025年

作者： Ghrist, Robert Gould, Julian Lopez, Miguel Riess, Hans Department of Mathematics and Electrical & Systems Engineering University of Pennsylvania Philadelphia PA Department of Mathematics University of Pennsylvania Philadelphia PA Applied Mathematics & Computational Science University of Pennsylvania Philadelphia PA School of Electrical and Computer Engineering Georgia Institute of Technology Atlanta GA

Modern financial networks involve complex obligations that transcend simple monetary debts: multiple currencies, prioritized claims, supply chain dependencies, and more. We present a mathematical framework that unifies and extends these scenarios by recasting the classical Eisenberg-Noe model of financial clearing in terms of lattice liability networks. Each node in the network carries a complete lattice of possible states, while edges encode nominal liabilities. Our framework generalizes the scalar-valued clearing vectors of the classical model to lattice-valued clearing sections, preserving the elegant fixed-point structure while dramatically expanding its descriptive power. Our main theorem establishes that such networks possess clearing sections that themselves form a complete lattice under the product order. This structure theorem enables tractable analysis of equilibria in diverse domains, including multi-currency financial systems, decentralized finance with automated market makers, supply chains with resource transformation, and permission networks with complex authorization structures. We further extend our framework to chain-complete lattices for term structure models and multivalued mappings for complex negotiation systems. Our results demonstrate how lattice theory provides a natural language for understanding complex network dynamics across multiple domains, creating a unified mathematical foundation for analyzing systemic risk, resource allocation, and network *** Codes 91G40, 06B23, 91D30, 68Q85 Copyright © 2025, The Authors. All rights reserved.

关键词： Lattice theory

来源：评论

学校读者我要写书评

暂无评论

Activating Associative Disease-Aware Vision Token Memory for LLM-Based X-ray Report Generation

arXiv

引用

arXiv 2025年

作者： Wang, Xiao Wang, Fuling Wang, Haowen Jiang, Bo Li, Chuanfu Wang, Yaowei Tian, Yonghong Tang, Jin Information Materials and Intelligent Sensing Laboratory of Anhui Province Anhui University Hefei230601 China Anhui Provincial Key Laboratory of Multimodal Cognitive Computation Anhui University Hefei230601 China School of Computer Science and Technology Anhui University Hefei230601 China The First Affiliated Hospital of Anhui University of Chinese Medicine Hefei230022 China Peng Cheng Laboratory Shenzhen China Harbin Institute of Technology Shenzhen China National Engineering Laboratory for Video Technology School of Electronics Engineering and Computer Science Peking University Beijing China

X-ray image based medical report generation achieves significant progress in recent years with the help of the large language model, however, these models have not fully exploited the effective information in visual image regions, resulting in reports that are linguistically sound but insufficient in describing key diseases. In this paper, we propose a novel associative memory-enhanced X-ray report generation model that effectively mimics the process of professional doctors writing medical reports. It considers both the mining of global and local visual information and associates historical report information to better complete the writing of the current report. Specifically, given an X-ray image, we first utilize a classification model along with its activation maps to accomplish the mining of visual regions highly associated with diseases and the learning of disease query tokens. Then, we employ a visual Hopfield network to establish memory associations for disease-related tokens, and a report Hopfield network to retrieve report memory information. This process facilitates the generation of high-quality reports based on a large language model and achieves state-of-the-art performance on multiple benchmark datasets, including the IU X-ray, MIMIC-CXR, and Chexpert Plus. The source code of this work is released on https://***/Event-AHU/Medical_Image_Analysis. Copyright © 2025, The Authors. All rights reserved.

关键词： Associative storage

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：