检索结果-内蒙古大学图书馆

arXiv 2025年

作者： He, Changpeng Lu, Yang Ai, Bo Dobre, Octavia A. Ding, Zhiguo Niyato, Dusit School of Computer Science and Technology Beijing Jiaotong University Beijing100044 China School of Electronics and Information Engineering Beijing Jiaotong University Beijing100044 China Faculty of Engineering and Applied Science Memorial University St. John’sNLA1C 5S7 Canada Department of Electrical Engineering and Computer Science Khalifa University Abu Dhabi127788 United Arab Emirates College of Computing and Data Science Nanyang Technological University Singapore639798 Singapore

This paper investigates the graph neural network (GNN)-enabled beamforming design for interference channels. We propose a model termed interference channel GNN (ICGNN) to solve a quality-of-service constrained energy efficiency maximization problem. The ICGNN is two-stage, where the direction and power parts of beamforming vectors are learned separately but trained jointly via unsupervised learning. By formulating the dimensionality of features independent of the transceiver pairs, the ICGNN is scalable with the number of transceiver pairs. Besides, to improve the performance of the ICGNN, the hybrid maximum ratio transmission and zero-forcing scheme reduces the output ports, the feature enhancement module unifies the two types of links into one type, the subgraph representation enhances the message passing efficiency, and the multi-head attention and residual connection facilitate the feature extracting. Furthermore, we present the over-the-air distributed implementation of the ICGNN. Ablation studies validate the effectiveness of key components in the ICGNN. Numerical results also demonstrate the capability of ICGNN in achieving near-optimal performance with an average inference time less than 0.1 ms. The scalability of ICGNN for unseen problem sizes is evaluated and enhanced by transfer learning with limited fine-tuning cost. The results of the centralized and distributed implementations of ICGNN are illustrated. © 2025, CC BY.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Generalized Kullback-Leibler Divergence Loss

arXiv

引用

arXiv 2025年

作者： Cui, Jiequan Zhu, Beier Xu, Qingshan Tian, Zhuotao Qi, Xiaojuan Yu, Bei Zhang, Hanwang Hong, Richang College of Computing & Data Science Nanyang Technological University Singapore Department of Computer Science & Engineering The Chinese University of Hong Kong Sha Tin Hong Kong The University of Hong Kong Hong Kong Harbin Institution of Technology China Hefei University of Technology China

In this paper, we delve deeper into the Kullback–Leibler (KL) Divergence loss and mathematically prove that it is equivalent to the Decoupled Kullback-Leibler (DKL) Divergence loss that consists of 1) a weighted Mean Square Error (wMSE) loss and 2) a Cross-Entropy loss incorporating soft labels. Thanks to the decoupled structure of DKL loss, we have identified two areas for improvement. Firstly, we address the limitation of KL loss in scenarios like knowledge distillation by breaking its asymmetric optimization property along with a smoother weight function. This modification effectively alleviates convergence challenges in optimization, particularly for classes with high predicted scores in soft labels. Secondly, we introduce class-wise global information into KL/DKL to reduce bias arising from individual samples. With these two enhancements, we derive the Generalized Kullback–Leibler (GKL) Divergence loss and evaluate its effectiveness by conducting experiments on CIFAR-10/100, ImageNet, and vision-language datasets, focusing on adversarial training, and knowledge distillation tasks. Specifically, we achieve new state-of-the-art adversarial robustness on the public leaderboard — RobustBench and competitive knowledge distillation performance across CIFAR/ImageNet models and CLIP models, demonstrating the substantial practical merits. Our code is available at https://***/jiequancui/DKL. © 2025, CC BY-NC-SA.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

Comparative Analysis of Outlier Elimination Algorithms on Pulsar Stars dataset

Comparative Analysis of Outlier Elimination Algorithms on Pu...

引用

Emerging Smart Computing and Informatics (ESCI), Conference on

作者： Tejaswini Katale Waidehi Gautam Aakash Chotrani Aishwarya Katale Parikshit N. Mahalle Gitanjali R. Shinde Department of Computer Engineering Vishwakarma Institute of Technology Pune Maharashtra India Department of Information Technology University of Cumberlands Williamsburg KY USA Department of Business Analytics Mercer University Atlanta Georgia USA Artificial Intelligence and Data Science Vishwakarma Institute of Information Technology Pune Maharashtra India Computer Science and Engineering Vishwakarma Institute of Information Technology Pune Maharashtra India

Outliers are uncommon observations that are dispersed from most of the other data Points. These points, which can be abnormally high or low, can significantly affect the entire analysis and forecasting of the data. The purpose of this Research Paper is to examine the implementation of Outlier Elimination Algorithms on Pulsar Stars dataset. It is attributed as an HTRU2 (High Time Resolution Universe Survey) project, which focuses on identifying new pulsar candidates based on the data from Parkes radio telescope in Australia. On this dataset, a total of 6 Outlier Elimination Algorithms are implemented; Z-Score, DBSCAN(Density-based spatial clustering of applications with noise), Mahalanobis Distance, Median Absolute Deviation, Multivariate Normal Distribution for N Dimension, Angle-based Outlier Detection (ABOD). After removal of the identified Outliers, to determine which Outlier Technique was more effective, 4 Machine Learning Techniques are deployed - Random Forest, Decision Tree, Naive Bayes and XGBoost. Furthermore, Accuracies, Recall, Precision and F1 Score values were compared. Every deployed Outlier algorithm must have a consistent quantity of Outliers in order to be evaluated for efficacy. For the same around 2600–2800 number of Outliers are eliminated for each Technique. Multivariate Normal Distribution for N Dimension Outlier Algorithm with Random Forest yields Highest Accuracy of 98.2%. ABOD with Naive Bayes gives Highest Recall Value of 0.93, Z-Score with Random Forest results in Highest Precision value of 0.96 and ABOD with Random Forest yields Highest F1-Score value of 0.92. From the information provided above, it is also clear that Random Forest produces the greatest Accuracy, Precision, and F1-Score values among all the Machine Learning algorithms that have been deployed. Therefore, for improved outcomes, the Random Forest technique should be utilized after the outliers are removed.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing Diabetic Retinopathy Screening with Sequential Deep Learning Models

Enhancing Diabetic Retinopathy Screening with Sequential Dee...

引用

International Conference on Image Information Processing

作者： Vankadari Mohith Kalpana Raja I. R. Oviya Department of Computer Science and Engineering (AIE) Amrita School of Computing Amrita Vishwa Vidyapeetham Chennai India Section for Biomedical Informatics and Data Science School of Medicine Yale University New Haven USA

ISBN: (数字)9798350371406

ISBN: (纸本)9798350371413

Diabetic retinopathy is a serious medical disorder that, if left untreated, can result in visual impairment or blindness. The precise and timely categorization of its severity is critical for appropriate medical management. This research describes an automated method for determining the severity of diabetic retinopathy using deep learning algorithms. The suggested methodology adopts a two-step process for identifying diabetic retinopathy and then categorizing its severity. In this technique, two pre-trained deep learning models, InceptionResNetV2 and DenseNet121, are used. For training and assessment purposes, the “APTOS 2019 Blindness Detection” dataset which consists of five severity classes is used as the foundation. To overcome class imbalance, data augmentation approaches are used to improve the model’s performance. The results of the experiment show that the strategy is effective, with an accuracy of 83.10 % and an F1 score of 82.77 %. This study contributes to the development of automated systems for assessing diabetic retinopathy, perhaps enhancing early detection and treatment.

关键词： Deep learning Training Diabetic retinopathy Visual impairment Transfer learning Blindness Information processing

来源：评论

学校读者我要写书评

暂无评论

Improved Deep Matrix Factorization for Signal Recovery in Random Frequency Diversity Array Radar

Improved Deep Matrix Factorization for Signal Recovery in Ra...

引用

Signal, Information and data Processing (ICSIDP), IEEE International Conference on

作者： Dayan Guo Zijian Lin Tianyao Huang Yue Qi Jie He Xiao-Ping Steven Zhang Xueqian Wang Gang Li School of Computer & Communication Engineering University of Science and Technology Beijing Beijing China Institute of Data and Information Tsinghua Shenzhen International Graduate School Shenzhen China Department of Electronic Engineering Tsinghua University Beijing China

ISBN: (数字)9798331515669

ISBN: (纸本)9798331515676

Motivated by the success of recent deep learning researches on radar, we consider a deep learning based matrix factorization method for suppressing the high range-angle side-lobes in random frequency diversity array (FDA) radars. While the existing deep matrix factorization approach is designed for real-valued situations and signals of FDA radars are complex-valued cases, we propose two methods to extend the existing approach into complex-valued framework: 1) adjusting the neural network structures and activation functions for complex-valued operations, and 2) constructing a real-valued representation of the original complex-valued input. Simulation results demonstrate that the proposed methods outperform traditional ones in multi-target scenarios.

关键词： Deep learning Simulation Neural networks Estimation Radar data processing Frequency diversity

来源：评论

学校读者我要写书评

暂无评论

eCeLLM: Generalizing Large Language Models for E-commerce from Large-scale, High-quality Instruction data

arXiv

引用

arXiv 2024年

作者： Peng, Bo Ling, Xinyi Chen, Ziru Sun, Huan Ning, Xia Department of Computer Science and Engineering The Ohio State University United States Translational Data Analytics Institute The Ohio State University United States Department of Biomedical Informatics The Ohio State University United States

With tremendous efforts in developing effective e-commerce models, conventional e-commerce models show limited success in generalist e-commerce modeling, and suffer from unsatisfactory performance on new users and new products – a typical out-of-domain generalization challenge. Meanwhile, large language models (LLMs) demonstrate outstanding performance in generalist modeling and out-of-domain generalizability in many fields. Toward fully unleashing their power for e-commerce, in this paper, we construct ECInstruct, the first open-sourced, large-scale, and high-quality benchmark instruction dataset for e-commerce. Leveraging ECInstruct, we develop eCeLLM, a series of e-commerce LLMs, by instruction-tuning general-purpose LLMs. Our comprehensive experiments and evaluation demonstrate that eCeLLM models substantially outperform baseline models, including the most advanced GPT-4, and the state-of-the-art task-specific models in in-domain evaluation. Moreover, eCeLLM exhibits excellent generalizability to out-of-domain settings, including unseen products and unseen instructions, highlighting its superiority as a generalist e-commerce model. Both the ECInstruct dataset and the eCeLLM models show great potential in empowering versatile and effective LLMs for e-commerce. ECInstruct and eCeLLM models are publicly accessible through https://***/eCeLLM/. © 2024, CC BY.

关键词： Marketplaces

来源：评论

学校读者我要写书评

暂无评论

No-Regret Linear Bandits under Gap-Adjusted Misspecification

arXiv

引用

arXiv 2025年

作者： Liu, Chong Qiao, Dan Yin, Ming Bogunovic, Ilija Wang, Yu-Xiang Department of Computer Science University at Albany State University of New York AlbanyNY12222 United States Department of Computer Science and Engineering University of California La Jolla San DiegoCA92093 United States Department of Computer Science University of California Santa Barbara Santa BarbaraCA93106 United States Department of Electronic and Electrical Engineering University College London LondonWC1E 7JE United Kingdom Halıcıoğlu Data Science Institute University of California La Jolla San DiegoCA92093 United States

This work studies linear bandits under a new notion of gap-adjusted misspecification and is an extension of Liu et al. (2023). When the underlying reward function is not linear, existing linear bandits work usually relies on a uniform misspecification parameter ϵ that measures the sup-norm error of the best linear approximation. This results in an unavoidable linear regret whenever ϵ > 0. We propose a more natural model of misspecification which only requires the approximation error at each input x to be proportional to the suboptimality gap at x. It captures the intuition that, for optimization problems, near-optimal regions should matter more and we can tolerate larger approximation errors in suboptimal regions. Quite surprisingly, we show that the classical LinUCB algorithm — designed for the realizable case — is automatically robust against such ρ-gap-adjusted misspecification with parameter ρ diminishing at O(1/(d√log T)). It achieves a near-optimal O(√T) regret for problems that the best-known regret is almost linear in time horizon T. We further advance this frontier by presenting a novel phased elimination-based algorithm whose gap-adjusted misspecification parameter ρ = O(1/√d) does not scale with T. This algorithm attains optimal O(√T) regret and is deployment-efficient, requiring only log T batches of exploration. It also enjoys an adaptive O(log T) regret when a constant suboptimality gap exists. Technically, our proof relies on a novel self-bounding argument that bounds the part of the regret due to misspecification by the regret itself, and a new inductive lemma that limits the misspecification error within the suboptimality gap for all valid actions in each batch selected by G-optimal design. Copyright © 2025, The Authors. All rights reserved.

关键词： Errors

来源：评论

学校读者我要写书评

暂无评论

Evaluating the Effectiveness of Object Detection Algorithms for Autonomous Parking

Evaluating the Effectiveness of Object Detection Algorithms ...

引用

computer, Electronics & Electrical engineering & their Applications (IC2E3), International Conference on

作者： Utkarsh Sharma Krishnan Subramonian Aaron Dsouza Sughosh P Savitha G Girisha S Department of Data Science and Computer Applications Manipal Institute of Technology Manipal Academy of Higher Education Manipal India Department of Civil Engineering Manipal Institute of Technology Manipal Academy of Higher Education Manipal India

ISBN: (数字)9798350388534

ISBN: (纸本)9798350388541

In response to the challenges posed by increasing urbanization, traditional parking systems face numerous issues, such as inaccurate slot detection and manual tracking, which exacerbate traffic congestion and environmental strain. This paper presents an autonomous parking system as a solution to these challenges. By leveraging advanced object detection algorithms, including YOLO-NAS, YOLOv8, and DETR models, this system revolutionizes parking space identification, freeing drivers from the task of manual searching. Our research focused on identifying the most suitable object detection algorithm for our dataset and objectives. Utilizing the ‘PKLot’ dataset, we trained and fine-tuned our models, ensuring consistency in hyperparameters across algorithms to isolate their impact on performance. We evaluated algorithmic efficacy using metrics like mean average precision (mAP) and latency, measuring accuracy in discerning occupied and vacant spaces alongside real-time processing efficiency. Our evaluation results highlight the performance differences among algorithms, guiding future research. This research advocates for an intelligent, autonomous approach to parking management, aligning with goals of optimizing operations and promoting sustainable urban development.

关键词： Training Accuracy Urban areas Training data Object detection Manuals Object recognition Traffic congestion Testing Strain

来源：评论

学校读者我要写书评

暂无评论

On the Effectiveness of Function-Level Vulnerability Detectors for Inter-Procedural Vulnerabilities

On the Effectiveness of Function-Level Vulnerability Detecto...

引用

International Conference on Software engineering (ICSE)

作者： Zhen Li Ning Wang Deqing Zou Yating Li Ruqian Zhang Shouhuai Xu Chao Zhang Hai Jin National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security Cluster and Grid Computing Lab School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan China Jin YinHu Laboratory Wuhan China Department of Computer Science University of Colorado Colorado Springs Colorado Springs Colorado USA Institute for Network Sciences and Cyberspace Tsinghua University Beijing China School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China

ISBN: (数字)9798400702174

ISBN: (纸本)9798350382143

Software vulnerabilities are a major cyber threat and it is important to detect them. One important approach to detecting vulnerabilities is to use deep learning while treating a program function as a whole, known as function-level vulnerability detectors. However, the limitation of this approach is not understood. In this paper, we investigate its limitation in detecting one class of vulnerabilities known as inter-procedural vulnerabilities, where the to-be-patched statements and the vulnerability-triggering statements belong to different functions. For this purpose, we create the first Inter -Procedural Vulnerability dataset (InterPVD) based on C/C++ open-source software, and we propose a tool dubbed VulTrigger for identifying vulnerability-triggering statements across functions. Experimental results show that VulTrigger can effectively identify vulnerability-triggering statements and inter-procedural vulnerabilities. Our findings include: (i) inter-procedural vulnerabilities are prevalent with an average of 2.8 inter-procedural layers; and (ii) function-level vulner-ability detectors are much less effective in detecting to-be-patched functions of inter-procedural vulnerabilities than detecting their counterparts of intra-procedural vulnerabilities.

关键词： Deep learning Detectors Open source software Software engineering

来源：评论

学校读者我要写书评

暂无评论

Notice of Removal: Fourier Diffusion Models: A Method to Control MTF and NPS in Score-Based Stochastic Image Generation

引用

IEEE Transactions on Medical Imaging 2023年 PP卷 1-1页

作者： Tivnan, Matthew Teneggi, Jacopo Lee, Tzu-Cheng Zhang, Ruoqiao Boedeker, Kirsten Cai, Liang Gang, Grace J. Sulam, Jeremias Stayman, J. Webster Department of Biomedical Engineering Johns Hopkins University in Baltimore MD USA Department of Computer Science and Mathematical Institute for Data Science Johns Hopkins University in Baltimore MD USA Canon Medical Research USA in Vernon Hills IL USA Canon Medical Systems Corporation Otawara Japan Department of Radiology University of Pennsylvania in Philadelphia PA USA Department of Biomedical Engineering and Mathematical Institute for Data Science Johns Hopkins University in Baltimore MD USA

Removed.

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：