检索结果-内蒙古大学图书馆

Proceedings of the 60th Annual ACM/IEEE Design Automation Conference

作者： Yu Zhu Zhenhua Zhu Guohao Dai Fengbin Tu Hanbo Sun Kwang-Ting Cheng Huazhong Yang Yu Wang Department of Electronic Engineering BNRist Tsinghua University Beijing China Qingyuan Research Institute Shanghai Jiao Tong University Shanghai China Department of Electronic and Computer Engineering The Hong Kong University of Science and Technology Hong Kong

ISBN: (纸本)9798350323481

Processing-in-memory (PIM) architectures have shown great abilities for neural network (NN) acceleration on edge devices that demand low latency under severe area constraints. Heterogeneous PIM architectures with different PIM implementation approaches such as RRAM-based PIM and SRAM-based PIM can further improve the performance. However, the automatic generation of heterogeneous PIM architectures faces the following two unresolved problems. First, existing work has not considered the design for heterogeneous PIM-based NN accelerators with multiple memory technologies. Second, for PIM with insufficient memory on edge devices, it is challenging to find the optimal runtime weight scheduling strategy in an O(L!) optimization space for the NN with L *** this paper, we propose PIM-HLS, an automatic hardware generation tool for heterogeneous PIM-based NN accelerators. Aiming at the problems above, we first point out that heterogeneous PIM can improve the performance under severe area constraints. Then we optimize the architectures for each NN layer by taking the advantage of different memory technologies. We also define the optimization problem of runtime weight scheduling and mapping for the first time, and propose a dynamic-programming-based weight scheduling algorithm to reduce the optimization space to O(L2). We implement PIM-HLS to automatically generate the hardware code and the instructions. Results show that we achieve an averagely 5.9× speedup with 72.8% less area compared with state-of-the-art PIM designs.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Fine-Tuning and Deploying Large Language Models Over Edges: Issues and Approaches

arXiv

引用

arXiv 2024年

作者： Dong, Yanjie Zhang, Haijun Li, Chengming Guo, Song Leung, Victor C.M. Hu, Xiping Artificial Intelligence Research Institute Guangdong-Hong Kong-Macao Joint Laboratory for Emotional Intelligence and Pervasive Computing Shenzhen MSU-BIT University Shenzhen518172 China The Beijing Engineering and Technology Research Center for Convergence Networks and Ubiquitous Services University of Science and Technology Beijing Beijing China The Department of Computer Science and Engineering The Hong Kong University of Science and Technology Hong Kong The School of Medical Technology Beijing Institute of Technology Beijing100811 China

Since the invention of GPT2-1.5B in 2019, large language models (LLMs) have transitioned from specialized models to versatile foundation models. The LLMs exhibit impressive zero-shot ability, however, require fine-tuning on local datasets and significant resources for deployment. Traditional fine-tuning techniques with the first-order optimizers require substantial GPU memory that exceeds mainstream hardware capability. Therefore, memory-efficient methods are motivated to be investigated. Model compression techniques can reduce energy consumption, operational costs, and environmental impact so that to support sustainable artificial intelligence advancements. Additionally, large-scale foundation models have expanded to create images, audio, videos, and multi-modal contents, further emphasizing the need for efficient deployment. Therefore, we are motivated to present a comprehensive overview of the prevalent memory-efficient fine-tuning methods over the network edge. We also review the state-of-the-art literatures on model compression to provide a vision on deploying LLMs over the network edge. © 2024, CC BY.

关键词： Edge computing

来源：评论

学校读者我要写书评

暂无评论

Fully Quantum Hash Function

arXiv

引用

arXiv 2024年

作者： Banerjee, Shreya Meena, Harshita Tripathy, Somanath Panigrahi, Prasanta K. Center for Quantum Science and Technology Siksha’O’ Anusandhan University Odisha Bhubaneswar751030 India Department of Computer Science and Engineering Indian Institute of Technology Bihar Patna801106 India Department of Physical Sciences Indian Institute of Science Education and Research Kolkata West Bengal Mohanpur741246 India

We introduce a novel, fully quantum hash (FQH) function within the quantum walk on a cycle framework. We incorporate the deterministic quantum computation with single qubit to replace classical post-processing, thus increasing the inherent security. Further, our proposed hash function exhibits zero collision rate and high reliability. We further show that it provides > 50% avalanche in average, and is highly sensitive to the initial conditions. We show comparisons of several performance metrics for the proposed FQH on different settings as well as with existing protocols to prove its efficacy. FQH requires minimal quantum resources to produce a large hash value, providing security against the birthday attack. This innovative approach thus serves as an efficient hash function and lays the foundation for potential advancements in quantum cryptography by integrating the fully quantum hash generation protocol. © 2024, CC BY.

关键词： Hash functions

来源：评论

学校读者我要写书评

暂无评论

Digital-Twin-Driven PI Parameter Evaluation Method for Grid-Connected Converters 4

Digital-Twin-Driven PI Parameter Evaluation Method for Grid-...

引用

4th International Conference on Power engineering, ICPE 2023

作者： Cheng, Xu Chen, Jianfu Chen, Yong Zhao, Xiaoyan Zhou, Zhipeng Yang, Qifan Dai, Ning Yi Zhuhai Power Supply Bureau and DC Power Distribution and Consumption Technology Research Center Guangdong Power Grid Co. Ltd. Zhuhai China Zhuhai Power Supply Bureau Guangdong Power Grid Co. Ltd. Zhuhai China University of Macau SKL of Internet of Things for Smart City Department of Electrical and Computer Engineering Macao China University of Macau and Zhuhai UM Science & Technology Research Institute SKL of Internet of Things for Smart City Department of Electrical and Computer Engineering Macao China

ISBN: (纸本)9798350340020

Digital twin technology describes and models the characteristics, behavior, formation process and real-world performance of physical objects by using digital technologies such as sensors, Internet of Things (IoT), and artificial intelligence (AI). It is one of the key technologies to realize real-time perception and operation optimization of electric power equipment. Proportional-integral (PI) modules are the most commonly used blocks in the control system of power converters. But choosing the parameters with the best control effect requires repeated experiments. Therefore, this paper proposes a data-driven model that can evaluate the control performance of different PI parameters accurately and efficiently. First, the concept of digital twins and a general architecture applicable to the key power electronic equipment are demonstrated. Then, taking the grid-connected converter as an example, a data-driven model based on artificial neural network (ANN) and Levenberg-Marquardt (LM) algorithm is built to realize the evaluation of PI controller parameters. Comparisons with simulation models are also provided to show the validity of the digital-twin driven control parameter evaluation method. © 2023 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

Performance Analysis of Economic Model Predictive Control on a Solar-Powered Direct Contact Membrane Distillation System

引用

IFAC-PapersOnLine 2022年第33期55卷 13-19页

作者： Hichem Bessafa Ghania Khodja Chakir Messaoud Taous-Meriem Laleg-Kirati Electrical and Computer Engineering Program. Computer Electrical and Mathematical Science and Engineering (CEMSE) division. King Abdullah University of Science and Technology (KAUST) Ecole Polytechnique of Algiers Algeria National Institute for Research in Digital Science and Technology (INRIA) Paris-Saclay France

This paper proposes an economic model predictive control (EMPC) design for a Direct Contact Membrane Distillation powered by a solar collector system which aims at enhancing its economical performances. A differential algebraic equations-based model is used for the design of the EMPC control. Moreover, a nonlinear observer is developed for the estimation of the unmeasured state. A neural network is proposed to predict the unknown solar irradiance for future horizon where a solar model provides temperature predictions. The proposed control design has been validated in simulation using data provided by a partial differential equation-based model mimicking the real plant.

关键词： Direct contact membrane distillation Economic model predictive control differential algebraic equations nonlinear observer Long-short term memory neural network

来源：评论

学校读者我要写书评

暂无评论

Electrical Tuning of Neutral and Charged Excitons with 1-nm Gate

arXiv

引用

arXiv 2023年

作者： Almutlaq, Jawaher Wang, Jiangtao Li, Linsen Li, Chao Dang, Tong Bulović, Vladimir Kong, Jing Englund, Dirk Research Laboratory of Electronics Massachusetts Institute of Technology CambridgeMA02139 United States Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology CambridgeMA02139 United States

Electrical control of individual spins and photons in solids is key for quantum technologies, but scaling down to small, static systems remains challenging. Here, we demonstrate nanoscale electrical tuning of neutral and charged excitons in monolayer WSe2 using 1-nm carbon nanotube gates. Electrostatic simulations reveal a confinement radius below 15 nm, reaching the exciton Bohr radius limit for few-layer dielectric spacing. In situ photoluminescence spectroscopy shows gate-controlled conversion between neutral excitons, negatively charged trions, and biexcitons at 4 K. Important for quantum information processing applications, our measurements indicate gating of a local 2D electron gas in the WSe2 layer, coupled to photons via trion transitions with binding energies exceeding 20 meV. The ability to deterministically tune and address quantum emitters using nanoscale gates provides a pathway towards large-scale quantum optoelectronic circuits and spin-photon interfaces for quantum networking. © 2023, CC BY.

关键词： Selenium compounds

来源：评论

学校读者我要写书评

暂无评论

Detecting and Preventing Brute Force Attacks on Mobile Devices Using Advanced Encryption Techniques 3

Detecting and Preventing Brute Force Attacks on Mobile Devic...

引用

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

作者： Franklin, Ramya G. Kaur, Savinder Kumari, Binita Gupta, Mohit Gopalakrishna, K. Maharishi, Trapty Agarwal 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 Rajpura140417 India Department of Computer Applications Odisha Bhubaneswar India Chitkara University Chitkara Centre for Research and Development Himachal Pradesh Baddi174103 India Karnataka Bangalore India School of Engineering & Technology Maharishi University of Information Technology Uttar Pradesh India

ISBN: (纸本)9798331542375

Brute pressure attacks are a not unusual approach utilized by malicious actors to gain unauthorized rights of entry to touchy records on mobile devices. Those assaults involve trying out multiple combinations of usernames and passwords until the precise one is located. This poses a huge risk to mobile security, as those devices often incorporate a tremendous quantity of personal and private information. The proposed solution is to detect and save brute force attacks on cellular gadgets by enforcing superior encryption techniques. Those techniques contain complex algorithms and methods to protect touchy records from being accessed by way of unauthorized events. This may make it easier for attackers to decipher the encrypted documents, even though they're capable of getting entry to them. The device will constantly display the get entry to attempts and locate any unusual styles or suspicious pastimes. If a brute pressure assault is detected, the device will robotically block further entry attempts and cause an alert to the consumer. It'll additionally encrypt the information in the tool to save it from being compromised. This answer will appreciably increase the security of cell gadgets because it will make it almost impossible for hackers to benefit from unauthorized admission to sensitive data. It's going to supply user's peace of thoughts, understanding that their statistics are covered, and offer an extra layer of security to cell gadgets. © 2025 IEEE.

关键词： Sensitive data

来源：评论

学校读者我要写书评

暂无评论

Vector Control of Permanent Magnet Synchronous Motor

Vector Control of Permanent Magnet Synchronous Motor

引用

Intelligent Systems for Cybersecurity (ISCS), International Conference on

作者： Harshal Patil Yogesh Lohumi Praful V. Nandankar J. Leo Amalraj Amit Barve Ramya Maranan Balaji Institute of Technology & Management (BITM) Sri Balaji University Pune Maharashtra India Department of Computer Science & Engineering Graphic Era Deemed to be University Dehradun Uttarakhand India Department of Electrical Engineering Government College of Nagpur Maharashtra India Science and Humanities R.M.K College of Engineering and Technology Chennai Tamil Nadu India Department of Computer Science and Engineering Faculty of Engineering and Technology Parul Institute of Technology Parul University Gujarat India Department of Research and Innovation Saveetha School of Engineering SIMATS Chennai Tamil Nadu India

ISBN: (数字)9798350375237

ISBN: (纸本)9798350375244

Experimental measurements were conducted to validate the vector control (VC) algorithm's functionality with a synchronous motor (SM). Various scenarios were tested, including start-up, reversal, and responses to speed changes. Stability and regulation quality were assessed under load, with deliberate de-excitation testing to prevent motor damage. Transitioning to torque control, responses to torque changes during start-up and reversal were observed. Despite hardware limitations causing waveform discontinuities, the VC structure proved effective for laboratory purposes. Parameter designs of individual regulators were validated through measured data. The system's capability to handle rapid transients was challenged, necessitating longer recording times on the oscilloscope. Overall, the study confirms the functionality of individual system blocks and the practical feasibility of VC for synchronous drives, highlighting the need to address waveform discontinuities and hardware limitations in future implementations.

关键词： Torque Heuristic algorithms Torque control Control systems Hardware Regulation Vectors

来源：评论

学校读者我要写书评

暂无评论

Design and Development of an Agricultural Mobile Application using Machine Learning 2

Design and Development of an Agricultural Mobile Application...

引用

2nd International Conference on Technological Advancements in Computational sciences, ICTACS 2022

作者： Krishna, Vempati Tamrakar, Ashish Kumar Banala, Rajesh Saritha, Damera Rao, A.L.N. Buddhi, Dharam Tkr College of Engineering & Technology Department of Computer Science and Engineering Hyderabad India Rsr Rungta College of Engineering and Technology Department of CSE Chhattisgarh Bhilai India Telangana Hyderabad India Cmr Engineering College Department of Aiml Telangana Hyderabad India Lloyd Institute of Engineering and Technology Department of CSE Uttar Pradesh Greater Noida India Uttaranchal University Division of Research & Innovation Dehradun India

ISBN: (数字)9781665476577

ISBN: (纸本)9781665476577

Machine learning algorithms such as KNN and SVM can provide assistance with a variety of issues, including determining what crops should be planted when, as well as determining when the field requires additional water and fertilizer. The proposed system is intended to collect data on the current condition of the soil and make use of that data in order to establish the types of nutrients that are present in the soil. Farmers will be able to identify pest damage to their crops using camera sensor modules for the internet of things. They will be able to take the appropriate actions now that they have the ability to. Through the use of the app, the farmer is able to receive notifications and other information regarding crops based on the conditions of the soil and the weather. The types of soil, crops, nitrogen, potassium, and phosphorus are a few examples of the types of information that fall under this category. In addition to the characteristics of the soil and the weather, farmers can also base their decisions on the kind of crops they grow based on these elements. Because of this, the farmer is given the ability to take the appropriate measures to reduce crop loss and increase crop yield. © 2022 IEEE.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Methods, Models and Hardware-Software Complex of Distributed Monitoring Based on Iot and Blockchain technology

Methods, Models and Hardware-Software Complex of Distributed...

引用

International Russian Smart Industry Conference (SmartIndustryCon)

作者： Dmitry V. Goncharov Vyacheslav I. Fedorov Oleg O. Ivashchuk Dept of Information and Robotic Systems Institute of Engineering Technology and Natural Sciences Belgorod State National Research University Belgorod Russia Dept of Computer science Caspian University of Technology and Engineering named after Sh. Yessenov Aktau Kazakhstan

This research work explores the possibility of simultaneous use of Industry 4.0 technologies, namely the Internet of Things (IoT) and distributed systems (Blockchain) for collecting, processing, storing and confirming the adequacy of environmental data characterizing the state of sources of anthropogenic impact significantly distributed throughout the study area. It also describes the dynamics of the quality of natural environments during the dispersion and accumulation of pollutants. This approach is especially important for the implementation of market relations in the field of carbon quotas giving the possibility of redistributing the area of responsibility of production enterprises. The authors have developed a subsystem of an automated environmental safety management system, including a hardware and software complex that, based on Internet of Things technology, allows for the collection and integration of heterogeneous environmental information received from various sensors with its further storage and processing in a cloud system. The proposed subsystem, based on distributed registry technology, ensures openness and accessibility of the environmental data. It suggests decentralization of subsequent management decisions, so that it includes a specialized digital platform developed by the authors. The corresponding models and algorithms are presented in the research work, and their software implementation is developed.

关键词：

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：