检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Yuan, Cheng Liu, Zhening Lv, Jiashu Shao, Jiawei Jiang, Yufei Zhang, Jun Li, Xuelong of China Telecom China Department of Electronic and Computer Engineering Hong Kong University of Science and Technology Hong Kong School of Software and Microelectronics Peking University China School of Electronic and Information Engineering Harbin Institute of Technology Shenzhen China

With the rapid development of large multimodal models (LMMs), multimodal understanding applications are emerging. As most LMM inference requests originate from edge devices with limited computational capabilities, the predominant inference pipeline involves directly forwarding the input data to an edge server which handles all computations. However, this approach introduces high transmission latency due to limited uplink bandwidth of edge devices and significant computation latency caused by the prohibitive number of visual tokens, thus hindering delay-sensitive tasks and degrading user experience. To address this challenge, we propose a task-oriented feature compression (TOFC) method for multimodal understanding in a device-edge co-inference framework, where visual features are merged by clustering and encoded by a learnable and selective entropy model before feature projection. Specifically, we employ density peaks clustering based on K nearest neighbors to reduce the number of visual features, thereby minimizing both data transmission and computational complexity. Subsequently, a learnable entropy model with hyperprior is utilized to encode and decode merged features, further reducing transmission overhead. To enhance compression efficiency, multiple entropy models are adaptively selected based on the characteristics of the visual features, enabling a more accurate estimation of the probability distribution. Comprehensive experiments on seven visual question answering benchmarks validate the effectiveness of the proposed TOFC method. Results show that TOFC achieves up to 60% reduction in data transmission overhead and 50% reduction in system latency while maintaining identical task performance, compared with traditional image compression methods. Copyright © 2025, The Authors. All rights reserved.

关键词： Image compression

来源：评论

学校读者我要写书评

暂无评论

Compact NSGA-II for Multi-objective Feature Selection

Compact NSGA-II for Multi-objective Feature Selection

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Sevil Zanjani Miyandoab Shahryar Rahnamayan Azam Asilian Bidgoli Department of Electrical Computer and Software Engineering Ontario Tech University Oshawa ON Canada Department of Engineering Brock University St. Catharines ON Canada Faculty of Science Wilfrid Laurier University Waterloo ON Canada

Feature selection is an expensive challenging task in machine learning and data mining aimed at removing irrelevant and redundant features. This contributes to an improvement in classification accuracy, as well as the budget and memory requirements for classification, or any other post-processing task conducted after feature selection. In this regard, we define feature selection as a multi-objective binary optimization task with the objectives of maximizing classification accuracy and minimizing the number of selected features. In order to select optimal features, we have proposed a binary Compact NSGA-II (CNSGA-II) algorithm. Compactness represents the population as a probability distribution to enhance evolutionary algorithms not only to be more memory-efficient but also to reduce the number of fitness evaluations. Instead of holding two populations during the optimization process, our proposed method uses several Probability Vectors (PVs) to generate new individuals. Each PV efficiently explores a region of the search space to find non-dominated solutions instead of generating candidate solutions from a small population as is the common approach in most evolutionary algorithms. To the best of our knowledge, this is the first compact multi-objective algorithm proposed for feature selection. The reported results for expensive optimization cases with a limited budget on five datasets show that the CNSGA-II performs more efficiently than the well-known NSGA-II method in terms of the hypervolume (HV) performance metric requiring less memory. The proposed method and experimental results are explained and analyzed in detail.

关键词：

来源：评论

学校读者我要写书评

暂无评论

An efficient network operation automation scheme using network status information for local 5G networks 14

An efficient network operation automation scheme using netwo...

引用

14th International Conference on Information and Communication Technology Convergence, ICTC 2023

作者： Yoon, Mahnsuk Seo, Jihun Cho, Kyucheol Future Mobile Communication Research Center Gumi Korea Republic of Korea Transport Institute Center for Connected & Automated Driving Research Department of the Fourth Industrial Revolution & Transport Sejong Korea Republic of Kyungpook National University School of Computer Secience and Engineering Software Education Center Daegu Korea Republic of

ISBN: (纸本)9798350313277

5G SA communication technology is being recognized as a crucial technology shaping the future of wireless communication. Compared to previous generations of networks, 5G offers ultra-high-speed, low latency, and massive connectivity, presenting both enterprises and individual users with opportunities for new business models and an enhanced quality of life. One of the most significant changes between the conventional mobile communication environment and 5G is the activation of various application cases in B2B services based on the local 5G network. The local 5G network, tailored to specific sites, requires effective management of high-volume data generated in dense areas and necessitates minimizing manual intervention by administrators. The ZSM (Zero-touch Service Management) technology is gaining attention to meet these requirements. ZSM focuses on automated operations for managing large systems like commercial mobile communication networks. It is anticipated that ZSM will also be applied to the local 5G network for efficient management and operation of tailored services. This paper proposes a network management technique that minimizes administrator intervention in the local 5G network environment. The proposed network management technique collects statistical logs from local 5G network base stations and core systems, supports real-time network status checks as well as traffic predictions using machine learning. Additionally, it offers automated operation techniques for the local 5G network by incorporating functions such as disconnecting network access for devices that are factors of malfunction within the specialized network. © 2023 IEEE.

关键词： Automation of operations Local 5G Networks Network Management Zero-touch Service Management

来源：评论

学校读者我要写书评

暂无评论

Sequential Image Storytelling Model Based on Transformer Attention Pooling

Sequential Image Storytelling Model Based on Transformer Att...

引用

International Conference on Image and Vision Computing New Zealand, IVCNZ

作者： Zainy M. Malakan Ghulam Mubashar Hassan Ajmal Mian Department of Computer Science and Software Engineering The University of Western Australia Perth Australia Department of Data Science Umm Al-Qura University Makkah Kingdom of Saudi Arabia

The Visual Storytelling Task (VST) extends beyond describing a single image, such as image captioning, to sequential image descriptions in the form of a coherent story. However, such descriptions present challenges in handling varying language styles, relational role-modeling, consistency, and events not evident in individual images. A common limitation in existing approaches is their inability to fully describe relations and visual changes, leading to stories devoid of linguistic cohesion between multiple sentences. To address this challenge, we introduce a novel framework, Sequential Image Storytelling Model (SISM), based on the Transformer architecture. Our model contextualizes input images by dividing them into multiple 16 × 16 patches and associates them with language content using the encoder-decoder technique. It incorporates cross-attention and attention pooling to identify the most relevant relationships between input images. Our proposed model achieves state-of-the-art performance on the recently published SSID dataset and performs competitively on VIST dataset, achieving the top score on BLEU-1 metric.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Distributed Lossy Computation with Structured Codes: From Discrete to Continuous Sources

Distributed Lossy Computation with Structured Codes: From Di...

引用

2023 IEEE International Symposium on Information Theory, ISIT 2023

作者： Pastore, Adriano Lim, Sung Hoon Feng, Chen Nazer, Bobak Gastpar, Michael CTTC/CERCA Ispic Research Unit Castelldefels Spain Hallym University School of Software Chuncheon Korea Republic of University of British Columbia School of Engineering KelownaBC Canada Boston University Ece Department BostonMA United States Epfl School of Computer and Communication Sciences Lausanne Switzerland

ISBN: (纸本)9781665475549

This paper considers the problem of distributed lossy compression where the goal is to recover one or more linear combinations of the sources at the decoder, subject to distortion constraints. For certain configurations, it is known that codes with algebraic structure can outperform i.i.d. codebooks. For the special case of finite-alphabet sources, recent work has demonstrated how to incorporate joint typicality decoding alongside linear encoding and binning. This work takes a discretization approach to extend this rate region to include both integer- and real-valued sources. As a case study, the rate region is evaluated for the Gaussian case. The resulting joint-typicality-based rate region recovers and generalizes the best-known rate region for this scenario, based on lattice encoding and sequential decoding. © 2023 IEEE.

关键词： Signal encoding

来源：评论

学校读者我要写书评

暂无评论

Cybersecurity Threat Detection using Machine Learning on Network Traffic

Cybersecurity Threat Detection using Machine Learning on Net...

引用

International Conference on ent Computing and Control Systems (ICICCS)

作者： R. Rithi N. Nithya A. Akila R. Saravanakumar M. Vargheese S. Jeya Lakshmi Master of Computer Applications Karpagam Academy of Higher Education Coimbatore Tamil Nadu India Department of Computer Applications Karpagam Academy of Higher Education Coimbatore Tamil Nadu India Department of Artificial Intelligence and Data Science R.M.K College of Engineering and Technology Tamil Nadu India Iconix Software Solution Tirunelveli Tamil Nadu India Department of Computer Science and Engineering PSN College of Engineering and Technology Tirunelveli Tamil Nadu India

ISBN: (数字)9798331512088

ISBN: (纸本)9798331512095

Web phishing is a prevalent online attack in which victims are tricked into revealing their personal information by visiting fraudulent websites. Phishers often use URL shortening tools to disguise malicious links, making it easier to steal sensitive data such as usernames, passwords, and banking details. These phishing websites are designed to mimic legitimate ones, making detection more challenging. As phishing techniques continue to evolve rapidly alongside technological advancements, the need for effective anti-phishing systems has become increasingly crucial. Machine learning has emerged as a powerful tool for recognizing and mitigating phishing scams. This article provides a comprehensive overview of machine learning techniques for phishing detection and their effectiveness in combating online threats.

关键词： Uniform resource locators Accuracy Reviews Phishing Scalability Machine learning Telecommunication traffic Passwords Threat assessment Real-time systems

来源：评论

学校读者我要写书评

暂无评论

Boosting the Prediction of Brain Tumor Using Two Stage BiGait Architecture

Boosting the Prediction of Brain Tumor Using Two Stage BiGai...

引用

Proceedings of the Digital Image Computing: Technqiues and Applications (DICTA)

作者： Saif-ur-Rehman Khan Zia Khan Md Zakir Hossain Nicanor Mayumu Farhana Yasmin Younas Aziz School of Computer Science and Engineering Central South University Changsha China School of Elec Eng Comp and Math Sci (EECMS) Curtin University Perth Australia Department of Computer and Software Nanjing University of information Science and Technology Nanjing

ISBN: (数字)9798350379037

ISBN: (纸本)9798350379044

Deep learning (DL) techniques hold immense promise for revolutionizing medical diagnostics, including brain tumor detection. Detecting malignancies in the brain is fraught with challenges that carry critical implications for accurate and timely diagnosis. Early detection is paramount, and our paper presents a novel approach to address this issue. We introduce the BiGait ensemble transfer learning (TL) architecture tailored for brain tumor classification tasks. Our proposed methodology is a two-stage process that harnesses the power of TL. In the initial phase, we employ two established models, ResNet50 and EfficientNetB1, and fine-tune them using the Kaggle multiclass dataset. This includes the incorporation of supplementary feature layers. We also leverage a stack ensemble solution to combine these feature layers, enhancing the model's overall predictive capabilities. Our approach is designed to shift the input distribution positively, effectively mitigating overfitting and reducing hardware overhead. To validate the effectiveness of our method, we conducted extensive experiments and evaluation. The dataset utilized in this work, publicly available on Kaggle contains 3,264 brain MRIs categorized into four groups: normal (500 images), meningioma (937 images), glioma (926 images), and pituitary tumor (901 images). The experimental results demonstrate that our model surpasses traditional single-model approaches, achieving an impressive accuracy rate of 96% and exhibiting superior generalization capabilities. This advancement in brain tumor classification holds great promise for early diagnosis and improved patient outcomes.

关键词： Accuracy Transfer learning Brain tumors computer architecture Predictive models Brain modeling Reliability Medical diagnosis Residual neural networks Overfitting

来源：评论

学校读者我要写书评

暂无评论

An Image Classification Approach to Pre-Determine Extent of Development of Post-Operative Necrosis in Skin Flaps

An Image Classification Approach to Pre-Determine Extent of ...

引用

Annual International Conference of the IEEE engineering in Medicine and Biology Society (EMBC)

作者： G. Spacagna T. Fevens J. Barralet Department of Surgery McGill University Montreal Canada Department of Computer Science and Software Engineering Concordia University Montreal Canada Faculty of Dentistry McGill University Montreal Canada

Purpose - Clinicians and care workers currently cannot determine the final extent of necrosis once it begins. If the area is relatively small, it may heal like a normal wound, whereas re-operation may be the only option if it spreads. This paper represents work toward a predictive algorithm using full-thickness random rat skin flap photographs to determine whether the tissue will develop irretrievable *** - Using post-surgery images taken over a series of days of ischemic flaps, features were extracted, selected, and input into a classification algorithm to see if it could provide information on the future condition of the flaps. We split our data into two groups: flaps that underwent normal healing and slow healing. When consulting with a specialist, it was observed that the resulting dermal damage was not severe when a flap had ≤ 40% necrosis over its length on the final day. Three classifiers were implemented: K-Nearest Neighbor (KNN), Random Forest, and Support Vector Machine (SVM).Results - The trained KNN was able to correctly determine whether a flap developed a necrotic area larger or less than 40% of its length with an accuracy of 91% using only images one day post-surgery. Under leave-one-out cross-validation, the Random Forest and SVM achieved accuracies of 82% and 79.5%, respectively, using images spanning ten days *** - We have shown that a classifier can accurately determine whether ischemic skin flaps will develop severe necrotic *** Relevance— An algorithm that assesses early timepoint images and predicts necrosis spread is not only a vital tool for patients and clinicians but is also an extremely important tool to accelerate research into necrosis reduction strategies that ultimately may find application for more life-threatening necrosis-related conditions.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Preliminary Study on the Reproducibility of Fix Templates in Static Analysis Tool

Preliminary Study on the Reproducibility of Fix Templates in...

引用

Conference on software engineering Education & Training (CSEE&T)

作者： Sohyun Kim Youngkyoung Kim Eunseok Lee Department of Software Sungkyunkwan University Suwon Korea Department of Electrical and Computer Engineering Sungkyunkwan University Suwon Korea College of Computing and Informatics Sungkyunkwan University Suwon Korea

Automated Program Repair (APR) automatically generates patches for identified defects. As a result, APR can encourage novice students to learn coding by providing appropriate patches. Since students have their coding conventions, we should be able to deal with many types of defects. Existing studies have used predefined RuleId-driven templates to automatically fix defects in Static Analysis Tools(SATs). However, the community periodically adds, deletes, or changes SAT’s RuleIds. This is difficult for existing RuleId-based templates to reflect those changes immediately. Existing studies only cover about 10 RuleIds, making it difficult to address all defects faced by all students. Therefore, it is necessary to establish appropriate criteria for classifying templates. The SAT has a predefined format for how Error Messages are written, and since Error Messages contain fixing actions, defects with similar Error Messages tend to have similar fixing actions. These characteristics of the Error Message are suitable for reproducible template classification criteria. Our preliminary study demonstrated that by classifying patterns based on Error Messages, we could effectively address various defects, including those in different programming languages, using a single template. This means that if a newly added RuleId corresponds to an Error Message format already in the predefined Error Message-based template, it can be modified without additional effort. We plan to construct reproducible templates for each Error Message and provide ongoing patching of defects to students.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Deep boundary-aware semantic image segmentation

Deep boundary-aware semantic image segmentation

引用

作者： Wu, Huisi Li, Yifan Chen, Le Liu, Xueting Li, Ping College of Computer Science and Software Engineering Shenzhen University Shenzhen China Department of Computing The Hong Kong Polytechnic University Hong Kong

While extensive research efforts have been made in semantic image segmentation, the state-of-the-art methods still suffer from blurry boundaries and mismatched objects due to the insufficient multiscale adaptability. In this paper, we propose a two-branch convolutional neural network (CNN) approach to capture the multiscale context and the boundary information with the two branches, respectively. To capture the multiscale context, we propose to embed self-attention mechanism to the atrous spatial pyramid pooling network. To capture the boundary information, we propose to fuse the low-level features in boundary feature extraction for refining the extracted boundaries via a feature fusion layer (FFL). With FFL, our method can improve the segmentation result with clearer boundaries. A new loss function is proposed which contains a segmentation loss and a boundary loss. Experiments show that our method can predict the boundaries of objects more clearly and have better performance for small-scale objects. © 2021 John Wiley & Sons, Ltd.

关键词： Semantic Segmentation

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：