检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Yin, Chenyang Bai, Zhenyu Venkatram, Pranav Aggarval, Shivam Li, Zhaoying Mitra, Tulika School of Electronic Engineering and Computer Science Peking University China School of Computing National University of Singapore Singapore

Large Language Model (LLM) deployment on edge devices is typically constrained by the need for off-chip memory access, leading to high power consumption and limited throughput. Ternary quantization for LLMs is promising in maintaining model accuracy while reducing memory footprint. However, existing accelerators have not exploited this potential for on-chip inference. We present TerEffic, an FPGA-based accelerator that carefully co-designs memory architecture and computational units to unlock highly efficient LLM inference with fully on-chip execution. Through weight compression, custom computational units, and memory hierarchy optimization, we achieve unprecedented efficiency by eliminating off-chip memory bandwidth bottlenecks. We propose two architectural variants: a fully on-chip design for smaller models and an HBM-assisted design for larger ones. When evaluated on a 370M parameter model with single-batch inference, our on-chip design achieves 12,700 tokens/sec (149× higher than NVIDIA’s Jetson Orin Nano) with a power efficiency of 467 tokens/sec/W (19× better than Jetson Orin Nano). The HBM-assisted design provides 521 tokens/sec on a 2.7B parameter model (2× higher than NVIDIA’s A100) with 33W power consumption, achieving a power efficiency of 16 tokens/sec/W (8× better than A100). © 2025, CC BY.

关键词： Integrated circuit design

来源：评论

学校读者我要写书评

暂无评论

Cyberbullying Image Classification using Artificial Intelligence for Safer Online Platform

Cyberbullying Image Classification using Artificial Intellig...

引用

Pervasive computing and Social Networking (ICPCSN), International Conference on

作者： M. Vamshi Krishna R. Asish Verma P. Kirubanantham Department of Computing Technologies School of Computing College of Engineering and Technology SRM Institute of Science and Technology Chennai Kattankulathur India

ISBN: (数字)9798350386349

ISBN: (纸本)9798350386356

The toll that cyberbullying takes on victims' life is incalculable because a person's approach to it is highly individualized. For some victims, the message might be harsh, but for others, it might be normal. It is quite difficult to find reliable content because of the ambiguities in cyberbullying communications. Several research works have reportedly used a text- content-based methodology to answer this question. However, picture-based cyberbullying detection is now receiving less attention. Enhancing a methodology that permits the prevention of social media posts on image-based cyberbullying issues is the aim of this job. This study provides support for an automated version that searches a malicious social media network for a cyber-symbol image, relying just on switch mastery. By applying fashion knowledge, it is feasible to infer hidden contextual functions from texts containing cyberbullying. Our test includes two photo datasets: pictures of cyberbullying and pictures of non-bullying. The datasets might be helpful to Destiny researchers in expanding their research. It's challenging to find a pleasant and suitable version to recognize bullying photographs, therefore experiment with both transfer the experiment's findings showed that switch learning models are an excellent method for predicting conversations that are only based on images and involve cyberbullying.

关键词： Training Accuracy Image recognition Computational modeling Cyberbullying Switches Safety

来源：评论

学校读者我要写书评

暂无评论

Towards Sustainable Mango Cultivation: Automated Severity Classification of Mango Rust Disease using CNN-SVM 4

Towards Sustainable Mango Cultivation: Automated Severity Cl...

引用

4th International Conference on Advances in Electrical, computing, Communication and Sustainable Technologies, ICAECT 2024

作者： Kaur, Arshleen Kukreja, Vinay Nisha Chandran, S. Garg, Navin Sharma, Rishabh Chitkara University Institute of Engineering and Technology Chitkara University Punjab India School of Computing Graphic Era Hill University Uttarakhand Dehradun248002 India Computer Science and Engineering Graphic Era Deemed to Be University Uttarakhand Dehradun248002 India

ISBN: (纸本)9798350343670

The fungal pathogen Puccinia psidii is responsible for the disease known as mango rust, which poses a severe risk to mango farming all over the world. It is essential to efficient illness management to have both accurate and early disease diagnosis as well as a classification of the disease's severity. Convolutional Neural Network (CNN) and Support Vector Machines (SVM) are two types of artificial neural networks. In this study, we provide a unique way to solve this difficulty by introducing a hybrid model that combines the feature extraction skills of CNN with the discriminative capacity of SVM. In the first part of our research, we conduct an exhaustive analysis of conventional approaches to illness evaluation, image-based detection methods, and the revolutionary function that deep learning plays in plant pathology. We provide a description of our methodology, which includes the gathering of data, the preprocessing of that data, and the structure of the hybrid CNN-SVM model. The findings present compelling evidence of the model's superior performance, revealing an overall accuracy of 97.02%. The ability of the model to differentiate between the five unique severity levels of mango rust disease is demonstrated by class-specific metrics such as precision, recall, and F1-score. A comprehensive confusion matrix provides insights into the performance of the model, indicating how well it can reduce the number of incorrect classifications. In addition, our hybrid model outperforms other methods already used in the industry, making it a potentially useful step forward in the mango rust disease severity classification process. The findings of this research have significant consequences, including the elimination of subjectivity and wasteful use of resources in disease management, in addition to contributing to the maintenance of high-quality mango crops. © 2024 IEEE.

关键词： Agriculture Technology Disease Severity classification Fungal Pathogens Image Preprocessing Mango Rust Disease

来源：评论

学校读者我要写书评

暂无评论

SecureML: Machine Learning-Powered Malware Detection

SecureML: Machine Learning-Powered Malware Detection

引用

International Conference on Advances in Information Technology (ICAIT)

作者： Animay Awasthi Mohd Ahsan Vanusha D Delsi Robinsha S Department of ComputingTechnologies School of Computing SRM Institute of Science and Technology Kattankulathur Tamilnadu India Department of Computing Technologies Faculty of Engineering and Technology School of Computing SRM Institute of Science and Technology Kattankulathur Tamilnadu India

ISBN: (数字)9798350383867

ISBN: (纸本)9798350383874

As malware threats become increasingly complex, the development of advanced detection systems is necessary to protect digital assets. SecureML Using Machine Learning presents an innovative approach to combat the escalating threat of malicious software by leveraging machine learning algorithms. This research paper outlines the design, implementation, and evaluation of SecureML, emphasizing its effectiveness in detecting both known and novel malware strains. Through comprehensive feature selection techniques and algorithmic comparison, SecureML identifies crucial features and selects the Random Forest algorithm as the most effective for malware detection, achieving an impressive accuracy rate of 99.40% on the test set. The system's low false positive and false negative rates validate its robustness and reliability in accurately distinguishing between benign and malicious files. SecureML demonstrates significant potential in bolstering cybersecurity defenses and safeguarding digital assets against evolving malware threats.

关键词： Resistance Machine learning algorithms Accuracy Software algorithms Feature extraction Malware Robustness

来源：评论

学校读者我要写书评

暂无评论

Accurate and Optimized Labelling of Fashion Products Through Attention Based Spiking Neural Networks

Accurate and Optimized Labelling of Fashion Products Through...

引用

International Conference on Advanced computing and Communication Systems (ICACCS)

作者： N Vithyatharshana Yedhoti Thrinayani Yogeshwar Nitin Kulkarni Rimjhim Padam Singh Sneha Kanchan Dept. of Computer Science and Engineering Amrita School of Computing Bengaluru Amrita Vishwa Vidyapeetham India Department of Computer Science & Engineering Amrita School of Computing Bengaluru Department of Internet Engineering and Computer science Universiti Tunku Abdul Rahman Malaysia

ISBN: (数字)9798350384369

ISBN: (纸本)9798350384376

Automation of garment classification has become the backbone of e-commerce, influencing product search, personalization, and customer experience. This paper uses the Fashion MNIST dataset, a rich collection of clothing images to investigate the potential of Spiking Neural Networks (SNNs) for accurate and efficient classification. SNNs are an excellent choice for widespread usage and real-world integration due to their low power consumption and biomimicry. With the intention of revolutionizing product search, personalization, and customer experience, SNNs are made to address style differences, perspectives, and illumination. In order to fully utilize Fashion MNIST, this paper suggests strategic improvements to the SNN architecture. By addressing key challenges and introducing novel optimizations, our research contributes to a more robust and intelligent future for automated fashion product labelling. An SNN model with 93.6% accuracy has been developed as the base model for all the applications for the same. This data-driven approach leads the way to flawless product searches, highly customized recommendations, and a more intelligent fashion technology in future.

关键词： Accuracy Power demand Clothing Lighting Spiking neural networks Computer architecture Encoding Labeling Electronic commerce Optimization

来源：评论

学校读者我要写书评

暂无评论

FDGLib: A Communication Library for Efficient Large-Scale Graph Processing in FPGA-Accelerated Data Centers

引用

Journal of Computer science & Technology 2021年第5期36卷 1051-1070页

作者： Yu-Wei Wu Qing-Gang Wang Long Zheng Xiao-Fei Liao Hai Jin Wen-Bin Jiang Ran Zheng Kan Hu National Engineering Research Center for Big Data Technology and System School of Computer Science and Technology Huazhong University of Science and TechnologyWuhan 430074China Services Computing Technology and System Laboratory School of Computer Science and Technology Huazhong University of Science and TechnologyWuhan 430074China Cluster and Grid Computing Laboratory School of Computer Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China

With the rapid growth of real-world graphs,the size of which can easily exceed the on-chip(board)storage capacity of an accelerator,processing large-scale graphs on a single Field Programmable Gate Array(FPGA)becomes *** multi-FPGA acceleration is of great necessity and *** cloud providers(e.g.,Amazon,Microsoft,and Baidu)now expose FPGAs to users in their data centers,providing opportunities to accelerate large-scale graph *** this paper,we present a communication library,called FDGLib,which can easily scale out any existing single FPGA-based graph accelerator to a distributed version in a data center,with minimal hardware engineering *** provides six APIs that can be easily used and integrated into any FPGA-based graph accelerator with only a few lines of code *** the torus-based FPGA interconnection in data centers,FDGLib also improves communication efficiency using simple yet effective torus-friendly graph partition and placement *** interface FDGLib into AccuGraph,a state-of-the-art graph *** results on a 32-node Microsoft Catapult-like data center show that the distributed AccuGraph can be 2.32x and 4.77x faster than a state-of-the-art distributed FPGA-based graph accelerator ForeGraph and a distributed CPU-based graph system Gemini,with better scalability.

关键词： data center accelerator graph processing distributed architecture communication optimization

来源：评论

学校读者我要写书评

暂无评论

AIDEN - Artificial Intelligence and Digital Entity Network

AIDEN - Artificial Intelligence and Digital Entity Network

引用

Advances in computing, Communication and Applied Informatics (ACCAI), International Conference on

作者： Megha V Kawad Sudhanshu Himanshu Makwana P. Kirubanantham Department of Computing Technologies School of Computing College of Engineering and Technology SRM Institute of Science and Technology Kattankulathur Chennai India

ISBN: (数字)9798350389449

ISBN: (纸本)9798350389456

In the past few years, we have witnessed a rapid surge in technological advancements, which has significantly accelerated the evolution of Artificial Intelligence (AI) systems [1], [2]. This has fundamentally transformed the manner in which humans and machines communicate and interact with each other [3], [4]. The advent of advanced AI technologies has paved the way for the development of personal AI assistants that can emulate human intelligence and behavior [5], [6]. Our paper provides an in-depth exploration into the design and execution of ‘AIDEN: Artificial Intelligence and Digital Entity Network’. AIDEN is an innovative project that aims to create a personal AI assistant. The primary objective of AIDEN is to enhance user experiences by facilitating interactions that are as natural and human-like as possible [7], [8]. This groundbreaking project represents a significant stride in the field of AI, pushing the boundaries of what machines can achieve in terms of mimicking human intelligence and interaction [9], [10].

关键词： Measurement Accuracy Computational modeling Human intelligence Predictive models Data models Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

Comprehensive Methods for Secure Medical Image Transfer: Encryption and Standardization Techniques

Comprehensive Methods for Secure Medical Image Transfer: Enc...

引用

science, Technology, engineering and Management (ICSTEM), International Conference on

作者： K. Priyadharshini Hitesh Nandi Harsh Parmar Dept. of Computing Technologies School of Computing College of Engineering and Technology SRM Institute of Science and Technology Chennai India

ISBN: (数字)9798350376913

ISBN: (纸本)9798350388473

The primary goal of the process is to securely transfer medical images. This involves utilizing MRI datasets for processing and applying resizing and filtering techniques such as Conv2D and Gaussian blur to standardize the images. Additionally, an attack model is developed to enhance information security, starting with a PGD attack to access model weights and followed by a DAG attack to increase loss functions. Encryption is then performed using a Chaos-based encryption model to create a chaotic map and the Henon map for image conversion. A randomly generated key encryption method is utilized to generate a key for secure transfer. The encrypted image is displayed, and in the receiver section, the process is reversed for decryption by inputting the key generated in the encryption model. Finally, the original medical image is retrieved along with parameters such as MSE, MAE, RMSE, and F2 score to assess the accuracy of the recovered image and ensure its correctness.

关键词： Accuracy Filtering Magnetic resonance imaging Information security Standardization Receivers Encryption

来源：评论

学校读者我要写书评

暂无评论

Multi-Modal Emotion and Gesture Recognition System for Humans and Animals Using Computer Vision and Deep Learning 5

Multi-Modal Emotion and Gesture Recognition System for Human...

引用

5th International Conference on IoT Based Control Networks and Intelligent Systems, ICICNIS 2024

作者： Nair, Rekha R Babu, Tina Kishore, S. Shivaswamy, Smitha Aruna, S. Bhargavi, Manchikalapudi Alliance School of Advanced Computing Alliance University Department of Computer Science and Engineering Bengaluru India Department of Computer Application Karnataka Mangaluru India Acharya Institute of Technology Department of Computer Applications Bengaluru India School of Science and Computer Studies CMR University Bangalore India Hyderabad India

ISBN: (纸本)9798331518097

Emotion recognition through facial expressions and gesture analysis has emerged as a key area of research in computer vision and human-computer interaction. The ability to automatically detect and classify human emotions has wide-ranging applications, from enhancing user experiences to improving mental health monitoring and surveillance systems. This project builds on this background to develop a comprehensive system for recognizing emotions and gestures in both humans and animals. The system integrates facial emotion recognition, hand gesture interpretation, and animal emotion detection into a single platform. Using advanced deep learning models and computer vision libraries, the system classifies facial expressions into categories such as happiness, sadness, anger, fear, surprise, and neutral. The gesture recognition component interprets hand signals to convey messages like 'need help' or 'need water.' An innovative aspect of this work is the extension of emotion recognition to animals, particularly focusing on species like dogs and monkeys. Implemented primarily in Python, the system leverages libraries such as OpenCV and DeepFace. By combining these different modalities of nonverbal communication analysis, the project presents a versatile solution with potential applications in human-computer interaction, mental health monitoring, security, and animal welfare. © 2024 IEEE.

关键词： Gesture recognition

来源：评论

学校读者我要写书评

暂无评论

Software Re-engineering using Cloud computing Platform 5

Software Re-Engineering using Cloud Computing Platform

引用

5th International Conference on Smart Systems and Inventive Technology, ICSSIT 2023

作者： Raja, A. Ravi Raghunandan, K.R. Dodmane, Radhakrishna Dixit, Asmita Kumar, M. Sunil Kapila, Dhiraj Velagapudi Ramakrishna Siddhartha Engineering College Ece Department Vijayawada India Department of Computer Science and Engineering Nitte India Abes Engineering College Department of Informaton Technology Ghaziabad India Mohan Babu University School of Computing Department of Computer Science and Engineering Vijayawada India Lovely Professional University Department of Computer Science and Engineering Phagwara India

ISBN: (纸本)9781665474672

Economic and technological progress in the cloud are the main topics of this article. The essay examines the question of whether or not it makes financial sense to build software in the cloud, vs in-house. This research work provides an overview of the current model-based re-engineering practises. A novel and more thorough method is outlined in Business Process Model and Notation (BPMN) with some example activities, inputs, and outputs. As an integral element of the primary re-engineering process, reverse engineering is also something to be considered. Therefore, it expands on the concept of 'SBSE for the cloud,' providing actionable formulations of problems associated with cloud computing. The system has been built in the cloud to track the health of analytical services while they are being called into use. As an example of analytical services, a number of implementations of data mining methods are hosted from a number of packages. Their Quality of Service (QoS) values are also sampled by running them on certain real datasets. Based on the analysis of the KSEA National Math Competition event management system, this project will suggest some guidelines for software reengineering using AWS cloud computing services and web application technologies. Using these best practises, the informational gaps in software reengineering are bridged for web applications built in with cloud computing frameworks. © 2023 IEEE.

关键词： Cloud computing

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：