检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Liu, Hou-I Galindo, Marco Xie, Hongxia Wong, Lai-Kuan Shuai, Hong-Han Li, Yung-Yui Cheng, Wen-Huang Department of Electronics and Electrical Engineering National Yang Ming Chiao Tung University Hsinchu300 Taiwan College of Computer Science and Technology Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education Jilin University Changchun130000 China Faculty of Computing and Informatics Multimedia University Cyberjaya63100 Malaysia Hon Hai Research Institute Taipei114 Taiwan Department of Computer Science and Information Engineering National Taiwan University Taipei106 Taiwan

Over the past decade, the dominance of deep learning has prevailed across various domains of artificial intelligence, including natural language processing, computer vision, and biomedical signal processing. While there have been remarkable improvements in model accuracy, deploying these models on lightweight devices, such as mobile phones and microcontrollers, is constrained by limited resources. In this survey, we provide comprehensive design guidance tailored for these devices, detailing the meticulous design of lightweight models, compression methods, and hardware acceleration strategies. The principal goal of this work is to explore methods and concepts for getting around hardware constraints without compromising the model's accuracy. Additionally, we explore two notable paths for lightweight deep learning in the future: deployment techniques for TinyML and Large Language Models. Although these paths undoubtedly have potential, they also present significant challenges, encouraging research into unexplored areas. Copyright © 2024, The Authors. All rights reserved.

关键词： computer hardware

来源：评论

学校读者我要写书评

暂无评论

MRIGAT: Many-to-many Recommendation using Interaction based knowledge Graph Attention 22

MRIGAT: Many-to-many Recommendation using Interaction based ...

引用

3rd International Symposium on Artificial Intelligence for Medical sciences, ISAIMS 2022

作者： Gao, Feng Yuan, Kai Gu, Jinguang Liu, Yun Department of Computer Science and Technology Wuhan University of Science and Technology China The Key Laboratory of Rich-Media Knowledge Organization Service of Digital Publishing Content Institute of Scientific and Technical Information of China China Guangxi Eco-engineering Vocational & Technical College China 1st Affiliated Hospital Dept. of Medical Informatics School of Biomedical Engineering & Informatic Nanjing Medical University China

ISBN: (纸本)9781450398442

Recommendation systems combining Graph Neural Networks and knowledge Graphs have been successfully applied in various domains. However, most existing approaches only consider one-to-one or one-to-many user-item interactions and cannot cater to many-to-many recommendation scenarios, such as providing prescriptions based on clinical diagnosis and medical history. Providing such a recommendation requires considering the implicit interaction within input user features as well as output candidates. In this paper, we propose a two-stage knowledge graph attention aggregation mechanism that helps recommend drug combinations for a patient based on his conditions. First, a disease-drug interaction graph is constructed and integrated with the medical domain knowledge, forming a collaborative knowledge graph. Secondly, an intra-feature attention aggregation is performed to obtain the representations of diseases and drugs based on drug-drug and disease-disease interactions, respectively. Thirdly, an inter-feature attention aggregation is performed using the disease-drug interaction to better represent a user's condition. Finally, the user's condition representation is concatenated with other user features to generate the final user representation for the prescription recommendation. Experiments with realistic datasets show that our approach can outperform existing recommendation systems by 4.3% and 6.1% in precision and recall, respectively. © 2022 ACM.

关键词： knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Evaluating Energy Consumption for Routing Selection using the Multi-Routing Clustering Protocol using Timeslot Transmission in Dynamic Path Selection in Wireless Sensor Networks

引用

Procedia computer science 2025年 252卷 251-259页

作者： M. Prakash J. Abinesh P. Malarvizhi J. Jeba Emilyn A. Sam Thamburaj D. Vinod Kumar Department of Computer Science and Engineering Knowledge Institute of Technology Salem-637504 Tamil Nadu India Department of Information Technology Sona College of Technology Salem Tamil Nadu India Vinayaka Mission’s College of Physiotherapy Vinayaka Mission’s Research Foundation (Deemed to be University) Salem Tamil Nadu India Depertment of Biomedical Engineering Vinayaka Mission’s Kirupananda Variyar Engineering College Vinayaka Mission’s Research Foundation (Deemed to be University) Salem Tamil Nadu India

Wireless sensor networks (WSNs) are cutting-edge technology that can be used in many fields requiring critical information. However, limited resource constraints, contextual connectivity, and lifecycle requirements drive designers to seek more efficient WSN infrastructures. Unbalanced energy utilization of sensor hubs during information steering in Wireless Sensor Networks (WSN). Accordingly, one of the principal configuration difficulties of remote sensor networks is to limit the energy utilization of sensor hubs. Therefore, many routing schemes are designed to efficiently utilize sensor nodes’ limited Energy. These schemes generally use low-power paths to transmit data. It turns out that using the same path is suboptimal given the network lifetime. To overcome this problem, a new method is introduced that uses a Multi-Route Clustering Protocol Using Timeslot Transmission (MRCP-TTDPS) with dynamic path selection. It consumes Energy on path selection. In the first step, multipath routing is configured to detect path quality, so this method uses Multipath Optimized Routing (MPRS) to send high-quality path data. The second stage uses Cluster-Based Optimal Path Selection (CORS) to establish the best energy path. The third stage develops energy consumption models. The amount of the hubs with the most noteworthy leftover Energy is chosen as the group top of each round. Every standard hub partakes in a group framed by neigh exhausting bunch heads. Every sensor hub sends identified information to the bunch head in each round. The group head sends the data to the base station. The results show that the model is better than the path model regarding path identifier energy utilization and path quality.

关键词： Wireless Sensor Networks (WSN) Multi-Route clustering Protocol using Timeslot Transmission in Dynamic path selection (MRCP-TTDPS) Energy data transfer multipath routing Quality path sensor nodes

来源：评论

学校读者我要写书评

暂无评论

Label-aware Multi-level Contrastive Learning for Cross-lingual Spoken Language Understanding

arXiv

引用

arXiv 2022年

作者： Liang, Shining Shou, Linjun Pei, Jian Gong, Ming Zuo, Wanli Zuo, Xianglin Jiang, Daxin College of Computer Science and Technology Jilin University China Key Laboratory of Symbolic Computation and Knowledge Engineering MOE China STCA Microsoft Department of Electrical & Computer Engineering Duke University United States

Despite the great success of spoken language understanding (SLU) in high-resource languages, it remains challenging in low-resource languages mainly due to the lack of labeled training data. The recent multilingual code-switching approach achieves better alignments of model representations across languages by constructing a mixed-language context in zero-shot cross-lingual SLU. However, current code-switching methods are limited to implicit alignment and disregard the inherent semantic structure in SLU, i.e., the hierarchical inclusion of utterances, slots, and words. In this paper, we propose to model the utterance-slot-word structure by a multi-level contrastive learning framework at the utterance, slot, and word levels to facilitate explicit alignment. Novel code-switching schemes are introduced to generate hard negative examples for our contrastive learning framework. Furthermore, we develop a label-aware joint model leveraging label semantics to enhance the implicit alignment and feed to contrastive learning. Our experimental results show that our proposed methods significantly improve the performance compared with the strong baselines on two zero-shot cross-lingual SLU benchmark datasets. © 2022, CC BY-NC-ND.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Softsign Gaussian Deep Belief Neural Network (SGDBNN): A Novel Technique for Improving Pedestrian Detection Accuracy and Processing Efficiency

Softsign Gaussian Deep Belief Neural Network (SGDBNN): A Nov...

引用

International Conference on Image Information Processing

作者： N. P. G. Bhavani R V Sudha K.Shanmuga Priya S R Arun Raj Vinay Kumar Yadav Allam Balaram Department of Electronics and Communication Engineering Saveetha School of Engineering Saveetha Institute of Medical And Technical Sciences Tamilnadu India Department of Computer Science and Business Systems Knowledge Institute of Technology Kakapalayam Selliampalayam Tamil Nadu India Department of Information Technology KSR Institute for Engineering and Technology Tiruchengode TamilNadu India Department of Electronics & Communication Engineering U.B.D.T College of Engineering Davanagere Karnataka India Glocal School of Science and Technology Glocal University Uttar Pradesh India Department of Computer Science and Engineering MLR Institute of Technology Hyderabad India

ISBN: (数字)9798350371406

ISBN: (纸本)9798350371413

Pedestrian detection is a critical task in computer vision with the aim of accurately identifying pedestrians in images or video frames. In this work, we tackle the issue of enhancing pedestrian detection precision while reducing processing time. To achieve this, we propose a novel technique called Softsign Gaussian Deep Belief Neural Network (SGDBNN) for pedestrian detection. The proposed SGDBNN technique leverages the power of recurrent deep neural networks to handle the complexities of pedestrian detection. The input layer receives images as input, which are then processed in the hidden layer. Here, a Gaussian activation function is employed to extract relevant features, effectively detecting pedestrian objects in the images. The incorporation of deep learning and recurrent behavior in the neural networks results in reduced incorrect classifications, leading to a lower false positive rate and improved pedestrian detection accuracy. The softsign activation function at the hidden layer allows the proposed technique to efficiently compare extracted features with pre-stored templates, enabling effective pedestrian identification in the images. By effectively utilizing the recurrent behavior and advanced activation functions, the SGDBNN demonstrates superior performance compared to existing methods, such as Deep Convolutional Neural Networks and Extended Deep Model. Experimental evaluations illustrate the effectiveness of the proposed SGDBNN technique. It achieves a significant increase in Positive Detection Accuracy (PDA) of up to 20% and 16%, while simultaneously reducing the False Positive Rate (FPR) by up to 18% and 3% compared to existing Deep Convolutional Neural Networks and Extended Deep Model, respectively. Additionally, the proposed SGDBNN reduces training time by up to 30% and 12% and decreases memory requirements by up to 47% and 30% when compared to existing Deep Convolutional Neural Networks and Extended Deep Model, respectively. Likewise, the SGDBNN tec

关键词： Training Deep learning computer vision Pedestrians Computational modeling Memory management Information processing

来源：评论

学校读者我要写书评

暂无评论

Learning generalizable agents via saliency-guided features decorrelation 23

Learning generalizable agents via saliency-guided features d...

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Sili Huang Yanchao Sun Jifeng Hu Siyuan Guo Hechang Chen Yi Chang Lichao Sun Bo Yang Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education and School of Artificial Intelligence Jilin University China Department of Computer Science University of Maryland College Park School of Artificial Intelligence Jilin University China Lehigh University Bethlehem Pennsylvania Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education

In visual-based Reinforcement Learning (RL), agents often struggle to generalize well to environmental variations in the state space that were not observed during training. The variations can arise in both task-irrelevant features, such as background noise, and task-relevant features, such as robot configurations, that are related to the optimal decisions. To achieve generalization in both situations, agents are required to accurately understand the impact of changed features on the decisions, i.e., establishing the true associations between changed features and decisions in the policy model. However, due to the inherent correlations among features in the state space, the associations between features and decisions become entangled, making it difficult for the policy to distinguish them. To this end, we propose Saliency-Guided Features Decorrelation (SGFD) to eliminate these correlations through sample reweighting. Concretely, SGFD consists of two core techniques: Random Fourier Functions (RFF) and the saliency map. RFF is utilized to estimate the complex non-linear correlations in high-dimensional images, while the saliency map is designed to identify the changed features. Under the guidance of the saliency map, SGFD employs sample reweighting to minimize the estimated correlations related to changed features, thereby achieving decorrelation in visual RL tasks. Our experimental results demonstrate that SGFD can generalize well on a wide range of test environments and significantly outperforms state-of-the-art methods in handling both task-irrelevant variations and task-relevant variations.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Energy-Efficient Cluster in Wireless Sensor Network Using Life Time Delay Clustering Algorithms

Research Square

引用

Research Square 2021年

作者： Kumar, V. Jayapandian, N. Balasubramanie, P. Department of Computer Science and Engineering Knowledge Institute of Technology Salem India Department of Computer Science and Engineering Christ University Bangalore India Department of Computer Science and Engineering Kongu Engineering College Perundurai India

Through Wireless Sensor Networks (WSN) formation, industrial and academic communities have seen remarkable development in recent decades. One of the most common techniques to derive the best out of wireless sensor networks is to upgrade the operating group. The most important problem is the arrangement of optimal number of sensor nodes as clusters to discuss clustering method. In this method, new client nodes and dynamic methods are used to determine the optimal number of clusters and cluster heads which are to be better organized and proposed to classify each round. Parameters of effective energy use and the ability to decide the best method of attachments are included. The Problem coverage find changeability network route due to which traffic and delays keep the performance to be very high. A newer version of Gravity Analysis Algorithm (GAA) is used to solve this problem. This proposed new approach GAA is introduced to improve network lifetime, increase system energy efficiency and end delay performance. Simulation results show that modified GAA performance is better than other networks and it has more advanced Life Time Delay Clustering Algorithms-LTDCA protocols. The proposed method provides a set of data collection and increased throughput in wireless sensor networks. © 2021, CC BY.

关键词： Clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Avant-Grade framework Routing Optimization Based on Established Network Lifetime and Energy Management in WSN

Research Square

引用

Research Square 2021年

Wireless Sensor Network (WSN) is known to be a highly resource constrained class of network where energy consumption is one of the prime concerns. The existing system of LTDC has more drawbacks in network routing optimization lifetime and energy level improvement needed. In this research, a cross layer design methodology was adopted to design an energy efficient routing protocol entitled Avant-Grade framework Routing Optimization [AGFRO]. AGFRO is designed to minimize energy consumed in each node by reducing the amount of time in which a sensor node is in an idle listening state and reducing the average communication distance over the network. The performance of the proposed system has been critically evaluated in the context of network lifetime, throughput, and energy consumption of the network per individual basis and per data packet basis. The research results were analyzed and benchmarked against the well-known AGFRO protocols. The outcomes show a significant improvement in the WSN in terms of energy efficiency and the overall performance of WSN. © 2021, CC BY.

关键词： Energy utilization

来源：评论

学校读者我要写书评

暂无评论

Optimal Linear Deception Attacks on Remote State Estimation with Constrained Alarm Rates: A Low-Dimensional Case

Optimal Linear Deception Attacks on Remote State Estimation ...

引用

IEEE Conference on Decision and Control

作者： Jun Shang Hanwen Zhang Jing Zhou Tongwen Chen Department of Control Science and Engineering Shanghai Research Institute for Intelligent Autonomous Systems National Key Laboratory of Autonomous Intelligent Unmanned Systems and Frontiers Science Center for Intelligent Autonomous Systems Tongji University Shanghai China Key Laboratory of Knowledge Automation for Industrial Processes of Ministry of Education School of Automation and Electrical Engineering University of Science and Technology Beijing Beijing China Department of Electrical and Computer Engineering University of Alberta Edmonton AB Canada

ISBN: (数字)9798350316339

ISBN: (纸本)9798350316346

This study addresses linear attacks on remote state estimation within the context of a constrained alarm rate. Smart sensors, which are equipped with local Kalman filters, transmit innovations instead of raw measurements through a wireless communication network. This transmission is vulnerable to malicious data interception and manipulation by attackers. The aim of this research is to identify the optimal attack strategy that degrades the system performance while adhering to stealthiness constraints. A notable innovation of this paper is the direct association of the attack’s stealthiness with the alarm rate, diverging from traditional approaches that rely on the covariance of the innovation or the Kullback–Leibler divergence, which are conventional metrics that have been extensively explored in previous studies. Our findings reveal that the optimal attack strategy exhibits some structural characteristics in systems of low dimensions. The performance of the proposed attack strategy is demonstrated through numerical examples.

关键词： Wireless communication Measurement Technological innovation System performance Kalman filters State estimation Intelligent sensors

来源：评论

学校读者我要写书评

暂无评论

A New Embedded Surveillance System for Reducing COVID-19 Outbreak in Elderly Based on Deep Learning and IoT

A New Embedded Surveillance System for Reducing COVID-19 Out...

引用

2020 International Conference on Data Analytics for Business and Industry: Way Towards a Sustainable Economy, ICDABI 2020

作者： Othman, Nashwan Adnan Al-Dabagh, Mustafa Zuhaer Nayef Aydin, Ilhan Knowledge University College of Science Department of Computer Science Erbil Kurdistan Region446015 Iraq Firat University Computer Engineering Department Elazig Turkey

ISBN: (纸本)9781728196756

As a result of the fast spread of Coronavirus disease (COVID-19) throughout the world, it became urgent to evolve an aided-intelligent system to help healthcare organizations to control and early detect COVID-19 outbreak, especially after massive development in the computing. Artificial Intelligence and Deep Learning methods could introduce real assistance to many healthcare organizations in this global problem by monitoring, detecting, and reporting on infected persons in early-stage. As is well known, the elderly are most vulnerable to the effects of COVID-19, therefore, we aim through this paper is to present the embedded system has the ability to detect and report on the elderly in the endemic areas. An age estimation from a facial image based on deep learning methods and Internet of Things is also applied to send notification to mobile or any other device systems that the application is embedded on it based on IoT. Depending on the experimental results on the proposed system through the Mean Average Error rate, the proposed system gave better or equivalent to the results of the state-of-the-art techniques, and the prediction results of an average accuracy achieved 89.45%. Finally, the proposed system has the capacity to help reduce the intensity of spread COVID-19 by identified older people and prevent them from prohibited in dangerous areas. © 2020 IEEE.

关键词： COVID-19

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：