检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Wang, Na Zhan, Yu-Wei Chen, Zhen-Duo Wang, Yongxin Luo, Xin Xu, Xin-Shun School of Software Shandong University Jinan China Department of Computer Science and Technology Tsinghua University Beijing China School of Computer Science and Technology Shandong Jianzhu University Jinan China

ISBN: (纸本)9798350368741

Weakly-supervised online hashing has garnered significant attention recently, yet several challenges remain unresolved, such as how to effectively denoise tags, and how to efficiently learn hash functions in dynamic online scenarios. To tackle these challenges, we propose a novel method named Tag-Aware Weakly-supervised Online Hashing with enhanced joint representation (TA-WOH). Our method creates an enhanced joint representation with CLIP-based features in order to reduce the tag noise. Additionally, we introduce a tag association and noise model for improved similarity matrix and hash code learning. A novel mapping mechanism is developed to align joint representations with the optimal tag space, enhancing both the accuracy and robustness of the model. The computational complexity of TA-WOH is dependent solely on the size of the incoming data, ensuring scalability and efficiency for large-scale datasets. Extensive experiments on two datasets demonstrate that our method surpasses several state-of-the-art methods in both accuracy and efficiency. © 2025 IEEE.

关键词： Learning to Hash Online Hashing Social Image Retrieval

来源：评论

学校读者我要写书评

暂无评论

Improved GMM High-Resolution Remote Sensing Image Land Cover Classification 4

Improved GMM High-Resolution Remote Sensing Image Land Cover...

引用

4th International Signal Processing, Communications and Engineering Management Conference, ISPCEM 2024

作者： Chunyan, Wang Kaixin, Fu Xiang, Wang School of Software Liaoning Technical University Huludao China Department of Computer Science and Technology Dalian University of Technology Dalian China

ISBN: (纸本)9798331528676

To address the increased complexity of surface cover, increased heterogeneity within homogeneous regions, and increased similarity between different regions in high-resolution remote sensing images, which lead to increased classification difficulty, a supervised learning method based on a double neighborhood relationship Gaussian Regression Mixture Model (GRMM) is proposed. Firstly, supervised sampling of image regions is conducted, fitting histograms using the least squares method to establish Gaussian Mixture Models (GMM) for each land cover to represent the complex gray-scale features of land cover. Secondly, local spatial information of adjacent pixels is introduced into the image gray-scale space to construct a Gaussian Regression Model (GRM). Finally, in the membership degree space, the neighborhood relationship is processed again to achieve classification decision. The kappa coefficients of GRMM on synthetic images and real high-resolution remote sensing images respectively reaches 97.2% and 98.5%. Compared to existing mainstream models, GRMM demonstrates strong classification efficiency, noise reduction capability, and generalization ability, with clear classification boundaries, effectively improved the land cover classification capability of remote sensing images. ©2024 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Unleashing the Power of LLMs in Court View Generation by Stimulating Internal Knowledge and Incorporating External Knowledge

Unleashing the Power of LLMs in Court View Generation by Sti...

引用

2024 Findings of the Association for Computational Linguistics: NAACL 2024

作者： Liu, Yifei Wu, Yiquan Li, Ang Zhang, Yating Sun, Changlong Lu, Weiming Wu, Fei Kuang, Kun School of Software Technology Zhejiang University China College of Computer Science and Technology Zhejiang University China Alibaba Group China

ISBN: (纸本)9798891761193

Court View Generation (CVG) plays a vital role in the realm of legal artificial intelligence, which aims to support judges in crafting legal judgment documents. The court view consists of three essential judgment parts: the charge-related, law article-related, and prison term-related parts, each requiring specialized legal knowledge, rendering CVG a challenging task. Although Large Language Models (LLMs) have made remarkable strides in language generation, they encounter difficulties in the knowledge-intensive legal domain. Actually, there can be two types of knowledge: internal knowledge stored within LLMs' parameters and external knowledge sourced from legal documents outside the models. In this paper, we decompose court views into different parts, stimulate internal knowledge, and incorporate external information to unleash the power of LLMs in the CVG task. To validate our method, we conduct a series of experiment results on two real-world datasets LAIC2021 and CJO2022. The experiments demonstrate that our method is capable of generating more accurate and reliable court views. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

CATC: A task scheduling based on computing-aware and task classification in Pervasive Edge Computing 22

CATC: A task scheduling based on computing-aware and task cl...

引用

22nd IEEE International Symposium on Parallel and Distributed Processing with Applications, ISPA 2024

作者： Yang, JiaHui Zhang, Yang Zhang, ShuKui Zhu, Mingyu Wang, YingYing Soochow University School of Computer Science and Technology Suzhou215006 China Global Institute of Software Technology Suzhou215163 China

ISBN: (纸本)9798331509712

With no infrastructure, the battlefield environment has characteristics such as device heterogeneity, uneven resource distribution, and dynamic node movement. Additionally, the sensing tasks produced by devices are complex and delay-sensitive, making it challenging to complete the tasks within the time limit if a centralized unified scheduling strategy is employed. In order to fully utilize the resources in the battlefield environment, optimize the load situation in the battlefield scenario, and guarantee the stability of task offloading, this paper proposes a scheduling strategy based on computing-aware and task classification (CATC). We first design a distributed computing-aware mechanism to ensure the detection of surrounding resources and load conditions by the generating task devices. Based on this, we classify the battlefield tasks into multi-attribute tasks according to the actual situation, and develop an offloading strategy for different task types, taking into account computation, storage, transmission, and load situation. The simulation results demonstrate that the proposed CATC strategy has a high task completion rate and can satisfy the definite delay requirements of delay-sensitive tasks. © 2024 IEEE.

关键词： Computation offloading

来源：评论

学校读者我要写书评

暂无评论

Machine Learning Model for Recipe Optimization of College Students Based on Simulated Annealing 3

Machine Learning Model for Recipe Optimization of College St...

引用

3rd International Conference on Data Analytics, Computing and Artificial Intelligence, ICDACAI 2024

作者： Sheng, Peizhi Chen, Xuhao School of Computer and Information Science Software Southwest University Chongqing China School of Engineering and Technology Southwest University Chongqing China

ISBN: (纸本)9798331519032

This paper first made a comprehensive dietary nutrition evaluation of the two recipes provided, and conducted a visual analysis of the differences of various nutrient components by querying the "Table of Chinese Food Ingredients". The nutrients involved included major daily productivity nutrients such as protein and fat, and non-productivity nutrients such as calcium, iron and zinc, which were used as collaborative evaluation indicators. Secondly, in order to specifically evaluate the daily diet of two male and female students, we pointed out the problems in the diet according to the Basic requirements for Analysis, Evaluation and Optimal Design of Dietary Recipes, and then adjusted and improved them, and made a comprehensive dietary nutrition evaluation again. Finally, the diet after adjustment has been improved to basically meet the nutritional requirements. Then, two models are established. Firstly, a model based on linear programming algorithm is established to solve the problem, and a model based on simulated annealing algorithm is established as a backup scheme and innovative ideas. Then, we used Model 1 to optimize male and female college birthday meal recipes with different goals (most economical, highest AAS, and dual goals). Finally, we also compare these three kinds of daily diet recipes with different objective optimization, and get the conclusion that the daily diet recipes with dual objective optimization have the highest cost performance. Finally, in order to realize the optimal design of weekly balanced diet, the paper extends the optimization model of daily diet based on problem 2. In order to solve the problem that the extended optimization model is difficult to solve due to too many decision variables, we propose a random elimination strategy to solve the weekly recipe scheme step by step according to the time granularity. After the design of daily recipes for each day is completed, the food of the day is eliminated from the daily recipes of the n

关键词： Simulated annealing

来源：评论

学校读者我要写书评

暂无评论

Towards Location- and Orientation-Independent RFID Authentication with COTS Devices

Towards Location- and Orientation-Independent RFID Authentic...

引用

2023 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events, PerCom Workshops 2023

作者： Zhu, Yinan Duan, Chunhui Ding, Xuan School of Software and BNRist Tsinghua University China School of Computer Science and Technology Beijing Institute of Technology China

ISBN: (纸本)9781665453813

To authenticate tags against counterfeiting, RF fin-gerprinting technique is widely exploited. However, the status of tagged item is always changed by movement, rotation and other operations. When the item's location or orientation changes, the capability of past methods would be severely affected and their supported authentication ranges are fairly small. To overcome this challenge, we propose the first-of-its-kind method FreeAuth to achieve certain location- and orientation-independent RFID authentication without any customized devices, by attaching a tag-pair and hopping the frequency channels and transmission powers. The key insight of FreeAuth lies in an implicit fin-gerprint matching scheme where the distance-frequency-power and orientation-frequency-power relationships are leveraged to circumvent these two negative factors. We implement a prototype of FreeAuth with COTS devices and the experiments demonstrate that FreeAuth is able to achieve around 0.8m and 0.6m ranges along 2D axes, in which the authentication accuracy is over 80%. The average effective authentication range of FreeAuth can outperform the state-of-the-art method by 11.67 x. © 2023 IEEE.

关键词： Authentication

来源：评论

学校读者我要写书评

暂无评论

COLARE: Commit Classification via Fine-grained Context-aware Representation of Code Changes 31

COLARE: Commit Classification via Fine-grained Context-aware...

引用

31st IEEE International Conference on software Analysis, Evolution and Reengineering, SANER 2024

作者： Zeng, Qunhong Zhang, Yuxia Sun, Zeyu Guo, Yujie Liu, Hui School of Computer Science & Technology Beijing Institute of Technology Beijing China Institute of Software Chinese Academy of Sciences Beijing China

ISBN: (纸本)9798350330663

Commit classification for maintenance activities is of critical importance for both industry and academia. State-of-the-art approaches either treat code changes as plain text or rely on manually identified features. Directly applying the most advanced model of code change representation into commit classification faces two limitations: (1) coarse-grained diff comparison neglects the distance of modified code lines;(2) missing key context information of hunk modification and file categories. This study proposes a novel classification model, COLARE, which compares code changes at the hunk level, takes fine-grained features based on categories of changed files, and aggregates with the representation of commit messages. The evaluation results show that our model outperforms state-of-the-art techniques by 7.24% and 7.35% in accuracy and macro F1 score, respectively. We also manually labeled a multi-language dataset and evaluated our approach, The results further confirm that our approach achieves the best performance over three baselines, including ChatGPT (3.5). The evaluation of the ablation study demonstrates the effectiveness of the major components in our technique. © 2024 IEEE.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

DDA-NG: Fusing Non-negative Matrix Factorization and Graph Tranformer to Enhance Potential Dmg-Disease Associations 5

DDA-NG: Fusing Non-negative Matrix Factorization and Graph T...

引用

5th International Conference on computer Engineering and Intelligent Control, ICCEIC 2024

作者： Shi, Jie Su, Ying Lv, Xiaoyi Tian, Xuecong School of Software Xinjiang University Xinjiang Urumqi China School of Computer Science and Technology Xinjiang University Xinjiang Urumqi China

ISBN: (纸本)9798331507992

Predicting drug-disease relationships is crucial in the biomedical field, but traditional laboratory experiments are time-consuming and costly, often lacking predictive accuracy and efficiency. This paper presents the DDA-NG model, which integrates Non-negative Matrix Factorization (NMF) and Graph Transformer (GT) to predict drug-disease associations. The model utilizes NMF to extract features from drugs and diseases, capturing global structural information. It then integrates various similarity data and captures higher-order structural relationships through the Graph Transformer. By fusing features from both modules, DDA-NG enhances feature diversity and prediction accuracy. Experimental results show significant performance improvements across multiple datasets, demonstrating its strong capability to identify potential drugs for diseases. © 2024 IEEE.

关键词： Non-negative matrix factorization

来源：评论

学校读者我要写书评

暂无评论

Analysis and Improvement of an Authentication Scheme for Fog Computing Services

引用

Wuhan University Journal of Natural sciences 2024年第3期29卷 209-218页

作者： HUO Yuyan KANG Baoyuan NIU Shufang LI Anqian ZUO Xinyu School of Software Tiangong UniversityTianjin 300387China School of Computer Science and Technology Tiangong UniversityTianjin 300387China

Fog computing utilizes devices in the edge network to transmit data with very low latency and supports high mobility. However, fog computing inherits security and privacy problems from cloud computing. Therefore, various privacy schemes for fog computing have been proposed to prevent different types of attacks. Recently, Weng et al proposed a fog computing authentication scheme;after analyzing, we found that Weng et al's scheme cannot resist user tracking attack and user impersonation attack. Then, we propose an improved scheme through adding a password, modifying the calculation method of Ei, and adding timestamps. In addition, we also compare the improved scheme with existing authentication schemes in terms of security and computational efficiency. The results show that the improved scheme is more secure and has less computation.

关键词： authentication scheme fog computing security

来源：评论

学校读者我要写书评

暂无评论

Boosting Adversarial Training with Learnable Distribution

引用

computers, Materials & Continua 2024年第3期78卷 3247-3265页

作者： Kai Chen Jinwei Wang James Msughter Adeke Guangjie Liu Yuewei Dai School of Electronics and Information Engineering Nanjing University of Information Science and TechnologyNanjing210044China Key Laboratory of Intelligent Support Technology for Complex Environments Ministry of EducationNanjing210044China School of Computer and Software Nanjing University of Information Science and TechnologyNanjing210044China Nanjing Center for Applied Mathematics Nanjing211135China

In recent years,various adversarial defense methods have been proposed to improve the robustness of deep neural *** training is one of the most potent methods to defend against adversarial ***,the difference in the feature space between natural and adversarial examples hinders the accuracy and robustness of the model in adversarial *** paper proposes a learnable distribution adversarial training method,aiming to construct the same distribution for training data utilizing the Gaussian mixture *** distribution centroid is built to classify samples and constrain the distribution of the sample *** natural and adversarial examples are pushed to the same distribution centroid to improve the accuracy and robustness of the *** proposed method generates adversarial examples to close the distribution gap between the natural and adversarial examples through an attack algorithm explicitly designed for adversarial *** algorithm gradually increases the accuracy and robustness of the model by scaling ***,the proposed method outputs the predicted labels and the distance between the sample and the distribution *** distribution characteristics of the samples can be utilized to detect adversarial cases that can potentially evade the model *** effectiveness of the proposed method is demonstrated through comprehensive experiments.

关键词： Adversarial training feature space learnable distribution distribution centroid

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：