检索结果-内蒙古大学图书馆

Separate Source Channel Coding Is Still What You Need:An LLM-Based Rethinking

ZTE Communications 2025年第1期23卷 30-44页

作者： REN Tianqi LI Rongpeng ZHAO Mingmin CHEN Xianfu LIU Guangyi YANG Yang ZHAO Zhifeng ZHANG Honggang College of Information Science and Electronic Engineering Zhejiang UniversityHangzhou 310027China Shenzhen CyberAray Network Technology Co. Ltd.Shenzhen 518000China China Mobile Research Institute Beijing 100053China The Internet of Things Thrust The Hong Kong University of Science and Technology(Guangzhou)Guangzhou 511453China Zhejiang Lab Hangzhou 311121China Faculty of Data Science City University of MacaoMacao 999078China

Along with the proliferating research interest in semantic communication(Sem Com),joint source channel coding(JSCC)has dominated the attention due to the widely assumed existence in efficiently delivering information ***,this paper challenges the conventional JSCC paradigm and advocates for adopting separate source channel coding(SSCC)to enjoy a more underlying degree of freedom for *** demonstrate that SSCC,after leveraging the strengths of the Large Language Model(LLM)for source coding and Error Correction Code Transformer(ECCT)complemented for channel coding,offers superior performance over *** proposed framework also effectively highlights the compatibility challenges between Sem Com approaches and digital communication systems,particularly concerning the resource costs associated with the transmission of high-precision floating point *** comprehensive evaluations,we establish that assisted by LLM-based compression and ECCT-enhanced error correction,SSCC remains a viable and effective solution for modern communication *** other words,separate source channel coding is still what we need.

关键词： separate source channel coding(SSCC) joint source channel coding(JSCC) end-to-end communication system Large Language Model(LLM) lossless text compression Error Correction Code Transformer(ECCT)

来源：评论

学校读者我要写书评

暂无评论

Multilingual Hate Speech Detection: Innovations in Optimized Deep Learning for English and Arabic Hate Speech Detection

引用

SN Computer science 2025年第3期6卷 1-16页

作者： AL-Sukhani, Hassan Bsoul, Qusay Elhawary, Abdelrahman H. Nasr, Ziad M. Mansour, Ahmed E. Batyha, Radwan M. Alqadi, Basma S. Alqurni, Jehad Saad Alfagham, Hayat Madbouly, Magda M. Cybersecurity Department Faculty of Science and Information Technology Jadara University Irbid Jordan Cybersecurity Department College of Computer Sciences and Informatics Amman Arab University Amman Jordan Faculty of Engineering Ain Shams University Cairo Egypt Faculty of Computer Science Misr International University Cairo Egypt Faculty of Information Technology Applied Science Private University Amman 11931 Jordan Computer Science Department College of Computer and Information Sciences Al Imam Mohammad Ibn Saud Islamic University Riyadh Saudi Arabia Department of Educational Technologies College of Education Imam Abdulrahman Bin Faisal University P.O. Box 1982 Dammam Saudi Arabia Department of MIS College of Applied Studies and Community Service Imam Abdulrahman Bin Faisal University P.O. Box 1982 Dammam Saudi Arabia Institute of Graduate Studies & Research Alexandria University Alexandria Egypt Faculty of Computer and Data Science Alexandria University Alexandria Egypt

This paper presents the development of a multilingual hate speech detection model that effectively processes and classifies content in both Arabic and English. The study leverages both traditional machine learning models, such as K-Nearest Neighbors (KNN), Naive Bayes, and Support Vector Machines (SVM), as well as advanced deep learning models, specifically Bi-directional Long Short-Term Memory (Bi-LSTM) networks. A key challenge addressed is the classification of mixed-language content, which is common on social media platforms in the MENA region. To enhance detection performance, preprocessing techniques were applied to the text data, and the Synthetic Minority Over-sampling Technique (SMOTE) was used to balance the dataset. The results show that the Bi-LSTM model outperformed traditional machine learning approaches, particularly in identifying hate speech across multiple languages. The proposed model demonstrates superior accuracy and robustness in handling mixed-language input, providing a more effective solution for real-world hate speech detection tasks. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： Class imbalance Deep learning Hate speech detection Machine learning Mixed Arabic and English training Multilingual NLP Natural language processing (NLP) Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

LossControl: Defending Membership Inference Attacks by Controlling the Loss

LossControl: Defending Membership Inference Attacks by Contr...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Bo Yang Hongwei Yang Renhao Lu Hui He Weizhe Zhang Haoyu He Rahul Yadav Faculty of Computing Harbin Institute of Technology Harbin China Department of Data Science and AI Faculty of Information Technology Monash University Melbourne Australia College of Computer Science and Technology Harbin Engineering University Harbin China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Machine learning models are vulnerable to membership inference attacks (MIAs), where adversaries attempt to predict whether specific samples are part of the model’s training set. Previous studies have demonstrated a strong correlation between the distinguishability of training and testing loss distributions and the model’s susceptibility to MIAs. Motivated by existing results, we propose a novel training framework called LossControl, which focuses on manipulating loss to mitigate privacy leaks. In LossControl, we first utilize Soft-label Training to replace the general learning process, which facilitates model training while improving generalization. Next, we monitor overfitting samples during the training process and prevent further loss reduction by applying our designed Loss Ascent to these samples without sacrificing model performance. Through extensive evaluations across four diverse datasets (including images, medical data, and transaction records), our method consistently outperforms defense mechanisms against state-of-the-art attacks and achieves optimal model performance in most experiments, demonstrating LossControl’s superior resilience against MIAs and its ability to strike an unparalleled balance between privacy and utility.

关键词： Training Privacy Process control Machine learning Predictive models Signal processing Speech processing Testing Resilience Overfitting

来源：评论

学校读者我要写书评

暂无评论

Enhancing Speech Emotion Recognition With Conditional Emotion Feature Diffusion and Progressive Interleaved Learning Strategy

IEEE Transactions on Audio, Speech and Language Processing

引用

IEEE Transactions on Audio, Speech and Language Processing 2025年 33卷 1787-1800页

作者： Yang Liu Xin Chen Zhichao Peng Yongwei Li Xingfeng Li Peng Song Masashi Unoki Zhen Zhao School of Information Science and Technology Qingdao University of Science and Technology Qingdao China School of Information Hunan University of Humanities Science and Technology Loudi China CAS Key Laboratory of Behavioral Science Institute of Psychology Chinese Academy of Sciences Beijing China Faculty of Data Science City University of Macau Macau China School of Computer and Control Engineering Yantai University Yantai China School of Information Science Japan Advanced Institute of Science and Technology Nomi Japan

Speech emotion recognition (SER) aims to identify the speaker's emotional states in specific utterances accurately. However, existing methods still face feature confusion when attempting to recognize certain emotions because traditional acoustic feature extraction methods fail to capture dynamic emotional changes, blurring emotional boundaries. Additionally, existing classification networks (CNs) are constrained by fixed learning strategies, hindering their ability to capture subtle emotional nuances and resulting in label confusion. To address these two issues, we introduce 3D multiresolution modulation filtered cochleogram (MMCG) features by computing the deltas and delta-deltas of MMCG features to enhance the dynamic emotional changes and produce distinct emotional boundaries. We then customize a conditional emotion feature diffusion (CEFD) module, which progressively diffuses features based on emotional context to retain emotional nuances effectively and reduce reliance on conditioned information. In addition, a confidence filtering module is used to filter diffused features based on confidence-based posterior probabilities to ensure enhanced feature discrimination. We design a flexible training strategy named the progressive interleaved learning strategy (PILS) to learn further complex emotional nuances, which consists of two alternating stages: fine-tuning the CN parameters and supervising the CEFD output. Testing on the IEMOCAP, CASIA, and EMODB corpora demonstrates significant performance improvements in SER.

关键词： Feature extraction Emotion recognition Training Three-dimensional displays Diffusion models Speech enhancement Electronic mail Speech recognition Semantics Face recognition

来源：评论

学校读者我要写书评

暂无评论

From User Preferences to Accurate Predictions: Enhancing Movie Recommendation Systems with Neural Collaborative Filtering and Sentiment Analysis

引用

SN Computer science 2025年第3期6卷 1-26页

作者： Bsoul, Qusay Zawaideh, Firas Alqadi, Basma S. Almusfar, Latifa Abdullah Khalaf, Osamah Ibrahim Alattas, Ahmed Saleh Alali, Muath AbdElminaam, Diaa Salama Cybersecurity Department College of Computer Sciences and Informatics Amman Arab University Amman 11953 Jordan Cybersecurity Department Faculty of Science and Information Technology Jadara University Irbid Jordan Computer Science Department College of Computer and Information Sciences Al Imam Mohammad Ibn Saud Islamic University Riyadh Saudi Arabia College of Business Administration Imam Abdulrahman Bin Faisal University P.O. Box 1982 Dammam 31441 Saudi Arabia Department of Solar Al-Nahrain Research Center for Renewable Energy Al-Nahrain University Jadriya Baghdad Iraq Information Science Department Faculty of Arts and Humanities King Abdulaziz University Jeddah Saudi Arabia Data Science and Artificial Intelligence Faculty of Information Technology Applied Science Private University Amman Jordan Jadara Research Center Jadara University Irbid 21110 Jordan MEU Research Unit Middle East University Amman 11831 Jordan

This paper introduces an advanced movie recommendation system that combines neural collaborative filtering (NCF) with sentiment analysis to improve personalization, accuracy, and robustness in recommending movies. By integrating user preferences with the sentiment polarity of reviews, the system refines recommendations, effectively addressing challenges such as the cold start problem, data sparsity, and limited semantic understanding. Unlike traditional collaborative and content-based filtering methods, which often struggle with diversity and accuracy, our approach utilizes deep learning within NCF to reveal hidden patterns in user behavior, enhancing recommendations’ relevance and precision. The proposed system is evaluated using standard performance metrics, including RMSE, MSE, and MAE, demonstrating its superior performance over conventional filtering techniques. Its enhanced scalability and adaptability position it as a promising tool for personalized content delivery in digital entertainment, with considerable potential for large-scale, dynamic recommendation environments. This research contributes to the existing knowledge on recommendation systems and offers new insights into improving content personalization and user satisfaction. The novelty and scientific value of this work lie in applying deep learning to tackle the challenges of accurate content recommendation in the rapidly evolving digital media landscape. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： Collaborative filtering Content-based filtering Cosine and person similarities Deep learning Evaluation metrics MAE Movie recommendation system NCF RMSE Surprise library

来源：评论

学校读者我要写书评

暂无评论

Understanding the spatio-temporal dynamics of ecosystem services under multiple future scenarios to assess the progress of Sustainable Development Goals implementation

引用

Journal of Geographical sciences 2025年第4期35卷 745-762页

作者： ZHANG Ze JIANG Weiguo LING Ziyan PENG Kaifeng WU Zhifeng LI Zhuo State Key Laboratory of Remote Sensing and Digital Earth Science Faculty of Geographical SciencesBeijing Normal UniversityBeijing 100875China School of Information and Control Engineering Qingdao University of TechnologyQingdao 266520ShandongChina College of Geographic and Environmental Sciences Tianjin Normal UniversityTianjin 300387China School of Geography and Remote Sensing Guangzhou UniversityGuangzhou 510006China National Marine Data and Information Service Tianjin 300171China

Ecosystem services in urban agglomerations are the environmental conditions under which human survival and development are *** assessment of ecosystem services and complex interactions can contribute positively to the achievement of the Sustainable Development Goals(SDGs)for urban ***,studies on the future contribution of multi-scenario ecosystem services to the SDGS are *** pronovel integrated modeling framework that integrates the CLUES,InVEST,SOM,and GWR approaches to address the complex relationship between ecosystem services over a long“past-present-future”time *** construct a novel ecosystem service bundle-based approach for measuring urban agglomerations progress towards achieving ecologically relevant sustainable development goals at multiple *** the future scenario,the water yield(WY),habitat quality(HQ),and soil conservation(SC)show similar spatial patterns,with comparable spatial grids,while carbon stock(CS)remains predominantly unchanged and the ecological protection scenario(EPS)improves more *** high-synergy regions are mainly distributed in bundle 4,and most of the trade-off regions appear in bundles 1 and *** the last 30 years,all but the water-related SDGs are declining in bundle 1 of the two urban agglomerations,which are 15%higher in the Guangxi Beibu Gulf(GBG)than in the Guangdong-Hong Kong-Macao Greater Bay Area(GBA).From 2020 to 2035,the three scenarios demonstrate that the optimization of the SDGs progresses most effectively under the future ecological protection scenario(EPS).In particular,bundles 3 and 4 are significantly *** critical new knowledge can be used in sustainable ecosystem management and decision-making in urban agglomerations.

关键词： Sustainable Development Goals(SDGs) ecosystem service bundle trade-offs and synergies InVEST model self-organizing mapping(SOM)algorithm urban agglomeration in south China

来源：评论

学校读者我要写书评

暂无评论

MedScale-Former: Self-guided multiscale transformer for medical image segmentation

引用

Medical Image Analysis 2025年 103卷 103554页

作者： Karimijafarbigloo, Sanaz Azad, Reza Kazerouni, Amirhossein Merhof, Dorit Faculty of Informatics and Data Science University of Regensburg Regensburg Germany Faculty of Electrical Engineering and Information Technology RWTH Aachen University Aachen Germany School of Electrical Engineering Iran University of Science and Technology Tehran Iran Fraunhofer Institute for Digital Medicine MEVIS Bremen Germany

Accurate medical image segmentation is crucial for enabling automated clinical decision procedures. However, existing supervised deep learning methods for medical image segmentation face significant challenges due to their reliance on extensive labeled training data. To address this limitation, our novel approach introduces a dual-branch transformer network operating on two scales, strategically encoding global contextual dependencies while preserving local information. To promote self-supervised learning, our method leverages semantic dependencies between different scales, generating a supervisory signal for inter-scale consistency. Additionally, it incorporates a spatial stability loss within each scale, fostering self-supervised content clustering. While intra-scale and inter-scale consistency losses enhance feature uniformity within clusters, we introduce a cross-entropy loss function atop the clustering score map to effectively model cluster distributions and refine decision boundaries. Furthermore, to account for pixel-level similarities between organ or lesion subpixels, we propose a selective kernel regional attention module as a plug and play component. This module adeptly captures and outlines organ or lesion regions, slightly enhancing the definition of object boundaries. Our experimental results on skin lesion, lung organ, and multiple myeloma plasma cell segmentation tasks demonstrate the superior performance of our method compared to state-of-the-art approaches. © 2025

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

DE-RALBA: dynamic enhanced resource aware load balancing algorithm for cloud computing

引用

PeerJ Computer science 2025年 11卷 e2739-e2739页

作者： Hussain, Altaf Aleem, Muhammad Rehman, Atiq Ur Arshad, Umer Department of Computer Science KICSIT Campus Institute of Space Technology Islamabad Pakistan School of Computing Mathematics and Data Science Coventry University London United Kingdom Faculty of Information Technology Department of Computer Science University of Central Punjab UCP Lahore Pakistan

Cloud computing provides an opportunity to gain access to the large-scale and high-speed resources without establishing your own computing infrastructure for executing the high-performance computing (HPC) applications. Cloud has the computing resources (i.e., computation power, storage, operating system, network, and database etc.) as a public utility and provides services to the end users on a pay-as-you-go model. From past several years, the efficient utilization of resources on a compute cloud has become a prime interest for the scientific community. One of the key reasons behind inefficient resource utilization is the imbalance distribution of workload while executing the HPC applications in a heterogenous computing environment. The static scheduling technique usually produces lower resource utilization and higher makespan, while the dynamic scheduling achieves better resource utilization and load-balancing by incorporating a dynamic resource pool. The dynamic techniques lead to increased overhead by requiring a continuous system monitoring, job requirement assessments and real-time allocation decisions. This additional load has the potential to impact the performance and responsiveness on computing system. In this article, a dynamic enhanced resource-aware load balancing algorithm (DE-RALBA) is proposed to mitigate the load-imbalance in job scheduling by considering the computing capabilities of all VMs in cloud computing. The empirical assessments are performed on CloudSim simulator using instances of two scientific benchmark datasets (i.e., heterogeneous computing scheduling problems (HCSP) instances and Google Cloud Jobs (GoCJ) dataset). The obtained results revealed that the DE-RALBA mitigates the load imbalance and provides a significant improvement in terms of makespan and resource utilization against existing algorithms, namely PSSLB, PSSELB, Dynamic MaxMin, and DRALBA. Using HCSP instances, the DE-RALBA algorithm achieves up to 52.35% improved resources

关键词： Scheduling algorithms

来源：评论

学校读者我要写书评

暂无评论

Robust Adversarial Defense Based on Non-Transferability of Attack Across Foundation Models

Robust Adversarial Defense Based on Non-Transferability of A...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Koshiro Toishi Keisuke Maeda Ren Togo Takahiro Ogawa Miki Haseyama School of Engineering Hokkaido University Japan Research Emergence Department Data-Driven Interdisciplinary Japan Faculty of Information Science and Technology Hokkaido University Japan

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

This paper presents a novel adversarial defense method that exploits non-transferability of attack across foundation models. Existing adversarial training methods have insufficient robustness to adversarial examples under powerful adversarial attacks in a white-box setting. We clarify that there is no attack transferability between various foundation models through preliminary experiments and propose a defense method that combines the non-transferability with adversarial training. The proposed method constructs a common embedding space of different models and detects adversarial examples by leveraging the different outputs from each model. Based on the detection of adversarial examples, the proposed method can adaptively select a robust model depending on the input images based on non-transferability of attack between models. Experimental results show that the proposed method improves the accuracy of zero-shot class classification for adversarial examples against each model.

关键词： Training Adaptation models Accuracy Foundation models Signal processing Feature extraction Robustness Acoustics Speech processing Glass box

来源：评论

学校读者我要写书评

暂无评论

Design of a Smart Home Door Security System with Face Detection and Smart Bell using ESP32-CAM

Design of a Smart Home Door Security System with Face Detect...

引用

Mobile Computing and Sustainable Informatics (ICMCSI), International Conference on

作者： Nurhusna Salsabila Apri Siswanto Luhur Bayuaji Informatics Department Faculty of Engineering Universitas Islam Riau Pekanbaru Indonesia Fac. Data Science and Information Technology INTI International University Malaysia

ISBN: (数字)9798331522667

ISBN: (纸本)9798331522674

This research introduces the design and implementation of an advanced IoT-based home security system aimed at improving both convenience and safety for homeowners. The system integrates a smart doorbell with a mobile application, enabling real-time alerts for guest arrivals. Key features include facial recognition for automated door unlocking and notification of unrecognized visitors. When an unknown face is detected, the system sends an image along with a notification to the homeowner via a Telegram bot. The architecture incorporates the ESP32-CAM microcontroller for face detection, a solenoid lock for automatic door access, and an ultrasonic sensor to detect visitor presence. Extensive testing under various environmental conditions was conducted to assess the system's performance, particularly in terms of face recognition accuracy and notification delivery. The results indicate that the system significantly enhances home security by ensuring timely notifications and automating door access for verified individuals. While the system performs reliably in well-lit environments, challenges remain in low-light conditions. Software testing confirmed the robustness of the underlying logic. This innovative IoT solution overcomes the limitations of traditional security systems, offering a more secure, user-friendly, and efficient alternative for modern home security.

关键词： Software testing Face recognition System performance Smart homes Chatbots Real-time systems Software reliability Security Face detection Solenoids

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：