检索结果-内蒙古大学图书馆

A novel attention-based deep learning model for improving sentiment classification after the case of the 2023 Kahramanmaras/Turkey earthquake on Twitter

引用

PeerJ computer Science 2025年 11卷 e2881-e2881页

作者： Aslan, Serpil Yildirim, Muhammed Software Engineering Malatya Turgut Ozal University Malatya Turkey Computer Engineering Malatya Turgut Ozal University Malatya Turkey

Twitter has emerged as one of the most widely used platforms for sharing information and updates. As users freely express their thoughts and emotions, a vast amount of data is generated, particularly in the aftermath of disasters, which can be collected quickly and directly from individuals. Traditionally, earthquake impact assessments have been conducted through field studies by non-governmental organizations (NGOs), a process that is often time-consuming and costly. Sentiment analysis (SA) on Twitter presents a valuable research area, enabling the extraction and interpretation of real-time public perceptions. In recent years, attention-based methods in deep learning networks have gained significant attention among researchers. This study proposes a novel sentiment classification model, MConv-BiLSTM-GAM, which leverages an attention mechanism to analyze public sentiment following the 7.8 and 7.5 Mw earthquakes that struck Kahramanmaraş, Turkey. The model employs the FastText word embedding technique to convert tweets into vector representations. These vectorized inputs are then processed by a hybrid model integrating convolutional neural networks (CNNs) and recurrent neural networks (RNNs) with a global attention mechanism. This ensures careful consideration of semantic dependencies in sentiment classification. The proposed model operates in three stages: (i) MConv—Local Contextual Feature Extraction, (ii) bidirectional long short-term memory (BiLSTM)—sequence learning, and (iii) Global Attention Mechanism (GAM)—Attention Mechanism. Experimental results demonstrate that the model achieves an accuracy of 93.32%, surpassing traditional deep learning models in the literature by approximately 3%. This research aims to provide objective insights to policymakers and decision-makers, facilitating adequate support for individuals and communities affected by disasters. Moreover, analyzing public sentiment during earthquakes contributes to understanding societal responses an

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

Critical Role of Individual Cell Temperature Monitoring in Mitigating Thermal Runaway and Reducing Accelerated Degradation in Lithium-Ion Batteries

Critical Role of Individual Cell Temperature Monitoring in M...

引用

Annual IEEE Conference on Applied Power Electronics Conference and Exposition (APEC)

作者： Mohit Sharma Akash Samanta William Locke Sheldon Williamson Department of Electrical Computer and Software Engineering Ontario Tech University Oshawa Canada

ISBN: (数字)9798331516116

ISBN: (纸本)9798331516123

Thermal runaway in lithium-ion batteries (LIBs) is a critical risk, potentially leading to fires and explosions. This phenomenon can be triggered by overcharging, short-circuiting, physical damage, manufacturing defects, and overheating. Effective mitigation of thermal runaway relies on precise and close monitoring of individual cells within the battery pack, a task often complicated by the complexity and cost of implementing temperature sensors on each cell in automotive applications. This study demonstrates the importance of individual cell temperature monitoring by intentionally including an unhealthy cell (C#12) in a 14-cell LIB 21700 module equipped with an automotive-grade battery management system (BMS) from NXP®. Experimental results reveal that despite efficient BMS balancing, the weak cell exhibited a significantly higher temperature rise, with a final temperature difference of 6°C and a voltage difference of over 200 mV compared to healthy cells. These findings underscore the potential for thermal failure and runaway if individual cell temperatures are not closely monitored. Additionally, reliance on voltage-based control alone can lead to suboptimal battery pack utilization, as evidenced by a 5% capacity loss due to the weak cell. This research highlights the necessity of monitoring the temperature of each cell to prevent thermal runaway and ensure efficient battery performance.

关键词： Temperature measurement Temperature sensors Lithium-ion batteries Degradation Accuracy Automotive applications Battery management systems Voltage control Monitoring Thermal degradation

来源：评论

学校读者我要写书评

暂无评论

Research on 2D Obstacle Avoidance Model Based on Field Rescue 24

Research on 2D Obstacle Avoidance Model Based on Field Rescu...

引用

Proceedings of the 2024 6th International Conference on Mechatronics Systems and Control engineering

作者： Ye Zhou Nengxin Mou College of Computer Science and Software Engineering Hohai University Nanjing Jiangsu China

ISBN: (纸本)9798400710346

This paper presents a novel 2D obstacle avoidance model tailored for field search and rescue robots. Leveraging advanced deep reinforcement learning techniques, the proposed model aims to enhance the robots’ navigation capabilities in dynamic and unpredictable environments. We evaluated four distinct reinforcement learning models—Deep Q-Learning (DQL), Proximal Policy Optimization (PPO), Advantage Actor-Critic (A2C), and Deep Q-Network (DQN)—to identify the most effective approach for real-time obstacle avoidance. Experimental results demon- strata that the DQL model outperforms the others in terms of success rate, average reward, and stability, making it particularly suitable for complex field operations. This research contributes to the development of robust and efficient navigation strategies for autonomous search and rescue missions, with potential implications for broader applications in robotics.

关键词： 2D Obstacle Avoidance

来源：评论

学校读者我要写书评

暂无评论

Multifaceted collaborative evolutionary constrained multimodal multiobjective optimization

引用

Swarm and Evolutionary Computation 2025年 96卷

作者： Wang, Zeyi Liu, Songbai Ma, Lijia Lin, Qiuzhen Chen, Jianyong College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China

In addressing constrained multimodal multiobjective optimization problems (CMMOPs), this paper proposes a multifaceted collaborative evolutionary algorithm (MCEA) designed to balance feasibility, convergence, and diversity in both the objective and decision spaces. Existing approaches often focus solely on maintaining population diversity or feasibility, neglecting the intricacies of CMMOPs, which require simultaneous consideration of multiple conflicting goals. Our MCEA framework features a local–global collaborative search strategy that employs dynamic clustering for effective exploration and exploitation of diverse decision space regions. Additionally, a parent–offspring collaborative transfer strategy facilitates knowledge sharing between populations, enhancing convergence early in the evolutionary process and preserving diversity in later stages. Furthermore, we customize an objective-search space collaborative selection strategy that filters solutions based on population diversity across both spaces. Extensive experiments on thirty-one benchmark CMMOPs demonstrate that MCEA significantly outperforms state-of-the-art algorithms on more than half of the test problems, as measured by IGD, IGDX, RPSP, and HV performance indicators. Furthermore, MCEA effectively locates multiple Pareto subsets, showcasing its ability to balance convergence, diversity, and feasibility in solving CMMOPs. This work underscores the importance of a comprehensive approach to tackling the complexities of CMMOPs and provides valuable insights for future research in this domain. © 2025

关键词： Multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

A Predictive Model for Early Effort Estimation of Mobile Games

A Predictive Model for Early Effort Estimation of Mobile Gam...

引用

International Conference on Advancements in Computational Sciences (ICACS)

作者： Ahrar Javed Ali Afzal Malik Department of Software Engineering National University of Computer and Emerging Sciences Lahore Pakistan

ISBN: (数字)9798331533038

ISBN: (纸本)9798331533045

Mobile games are a popular, cheap, and convenient source of entertainment. However, the increasing complexity of these games is making it more difficult to thoroughly plan, monitor, and control mobile game development projects. Possessing the ability to accurately estimate the development effort for these projects early in their life cycle is expected to make the jobs of their managers much easier. The objective of this study was to build a calibrated and validated early effort prediction model for mobile games developed using the Unity game engine. First, we identified factors that potentially influence development effort in these projects by conducting an industrial survey. Once these factors were identified, we gathered information about these factors and actual development effort for more than 100 real mobile games developed by different game studios. Then, we performed a simple linear regression (SLR) analysis to rank these factors with respect to their individual influence on mobile game development effort. Finally, we used the games data collected earlier to build, assess, and validate a forward step-wise multiple linear regression (FSMLR)-based model for early effort prediction of such mobile games. The MMRE value of this model is 0.136 and its PRED (25) value is $85.98 \%$. These values indicate that this model has reasonably good estimation accuracy. It can be a useful tool in the arsenal of managers interested in effectively planning their mobile game development projects early in the life cycle.

关键词： Surveys Accuracy Linear regression Estimation Entertainment industry Games Predictive models Planning Monitoring Land mobile radio

来源：评论

学校读者我要写书评

暂无评论

HAPG-SAQAM: Human Auditory Perception Guided Spatial Audio Quality Assessment Metric

HAPG-SAQAM: Human Auditory Perception Guided Spatial Audio Q...

引用

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

作者： Zheng, Yuanming Yao, Jiaxuan Deng, Xiangyu Yang, Yuhong Liao, Ruiqi Tu, Weiping Lin, Cedar National Engineering Research Center for Multimedia Software School of Computer Science Wuhan University Wuhan China Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan China Guangdong OPPO Mobile Telecommunications Corp. Guangdong China

ISBN: (纸本)9798350368741

Spatial audio quality evaluation is essential for applications like virtual and augmented reality, where accurate sound reproduction enhances user immersion. While subjective listening tests are the gold standard, they are costly and time-consuming. To address this, we propose HAPG-SAQAM, an objective metric for assessing timbre quality, spatial quality, and overall quality of binaural audio, guided by human auditory perception. Our contributions include: (1) the Multi-scale Auditory Guided Feature Extraction (MAGFE) module, incorporating gammatone frequency cepstral coefficients for better alignment with human perception;(2) Perceptual Weighted Loss (PWL), optimizing the weighting of timbre quality (TQ) and spatial quality (SQ) loss based on subjective test data;and (3) data augmentation techniques to enhance robustness by amplifying perceptual distortions. Experimental results show HAPG-SAQAM improves correlation with subjective scores by 10%, with ablation studies confirming the contributions of its components to enhanced spatial and overall audio quality. © 2025 IEEE.

关键词： auditory perception spatial audio Speech and audio quality assessment subjective experiment timbre quality

来源：评论

学校读者我要写书评

暂无评论

An Experimental Study on Dynamic Lifelong Learning with GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis

引用

IEEE Access 2025年 13卷 90316-90332页

作者： Huang, Huang Asemi, Adeleh Universiti Malaya Faculty of Computer Science Information Technology Department of Software Engineering Kuala Lumpur Malaysia

GPT is widely recognized as one of the most versatile and powerful large language models, excelling across diverse domains. However, its significant computational demands often render it economically unfeasible for individuals and small businesses, underscoring the need for efficient, domainspecific alternatives. In Aspect-Based Sentiment Analysis (ABSA), existing models are typically optimized for single domains, facing challenges in performing effectively across multiple domains. A key issue, known as "catastrophic forgetting," arises when models trained on one domain lose previously learned knowledge upon exposure to new domain data. This leads to two significant problems: limited cross-domain generalization and difficulty in retaining prior knowledge while learning domain-specific information. To address these challenges, we introduce the Dynamic Lifelong Learning Aspect-Based Sentiment Analysis GPT (DllaGPT), a model designed to handle multiple domains while mitigating catastrophic forgetting. Leveraging datasets from four ABSA domains—Laptops, Restaurants, Tweets, and Finance—this study fine-tunes a pretrained GPT model from HuggingFace sequentially across domains. DllaGPT employs a mechanism to retain real data from earlier domains during new domain training, effectively preserving prior knowledge. Experimental results highlight that DllaGPT achieves an average accuracy of 0.85 and a Backward Transfer (BWT) score of -0.09 across the four domains, showcasing its high accuracy and robust lifelong learning capabilities. © 2025 Institute of Electrical and Electronics Engineers Inc.. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

NumScout: Unveiling Numerical Defects in Smart Contracts Using LLM-Pruning Symbolic Execution

引用

IEEE Transactions on software engineering 2025年第5期51卷 1538-1553页

作者： Chen, Jiachi Shao, Zhenzhe Yang, Shuo Shen, Yiming Wang, Yanlin Chen, Ting Shan, Zhenyu Zheng, Zibin Hangzhou Normal University Intelligent Transportation and Information Security Laboratory Hangzhou311121 China Sun Yat-sen University School of Software Engineering Zhuhai519082 China University of Electronic Science and Technology of China School of Computer Science and Engineering School of Cyber Security Chengdu611731 China Kashi Institute of Electronics and Information Industry Kashi844000 China

In recent years, the Ethereum platform has witnessed a proliferation of smart contracts, accompanied by exponential growth in total value locked (TVL). High-TVL smart contracts often require complex numerical computations, particularly in mathematical financial models used by many decentralized applications (DApps). Improper calculations can introduce numerical defects, posing potential security risks. Existing research primarily focuses on traditional numerical defects like integer overflow, and there is currently a lack of systematic research and effective detection methods targeting new types of numerical defects. In this paper, we identify five new types of numerical defects through the analysis of 1,199 audit reports by utilizing the open card method. Each defect is defined and illustrated with a code example to highlight its features and potential consequences. We also propose NumScout, a symbolic execution-based tool designed to detect these five defects. Specifically, the tool combines information from source code and bytecode, analyzing key operations such as comparisons and transfers, to effectively locate defects and report them based on predefined detection patterns. Furthermore, NumScout uses a large language model (LLM) to prune functions which are unrelated to numerical operations. This step allows symbolic execution to quickly enter the target function and improve runtime speed by 28.4%. We run NumScout on 6,617 real-world contracts and evaluated its performance based on manually labeled results. We find that 1,774 contracts contained at least one of the five defects, and the tool achieved an overall precision of 89.7%. © 1976-2012 IEEE.

关键词： Smart contract

来源：评论

学校读者我要写书评

暂无评论

An LLM-Based Framework for Synthetic Data Generation

An LLM-Based Framework for Synthetic Data Generation

引用

IEEE Annual Computing and Communication Workshop and Conference (CCWC)

作者： Mandeep Goyal Qusay H. Mahmoud Department of Electrical Computer and Software Engineering Ontario Tech University Oshawa ON Canada

ISBN: (数字)9798331507695

ISBN: (纸本)9798331507701

The demand for high-quality datasets is rapidly increasing across sectors such as healthcare, finance, and cybersecurity, yet challenges like data scarcity and privacy concerns persist. To address this, we introduce a framework for synthetic data generation that empowers users to create realistic datasets while maintaining privacy. The framework leverages fine-tuned Large Language Models (LLMs) and differential privacy techniques, including IBM's diffprivlib, to generate synthetic data that replicates real-world patterns without exposing sensitive information. A proof-of-concept platform has been constructed to facilitate seamless data generation and augmentation, making it particularly useful in scenarios where original datasets are inaccessible, scarce, or privacy-restricted. The platform supports the creation of datasets across five key categories, employing advanced methods to preserve data integrity while ensuring compliance with stringent privacy standards. By combining cutting-edge AI technologies with robust privacy-preserving techniques, this framework offers a practical solution for researchers and professionals seeking reliable synthetic data to drive innovation in data-sensitive fields.

关键词： Technological innovation Differential privacy Large language models Data integrity Finance Medical services Data models Reliability Standards Synthetic data

来源：评论

学校读者我要写书评

暂无评论

Statistics-Fused Learning for Classification of Randomized Eeg Trials

SSRN

引用

SSRN 2025年

作者： Zhang, Wuxia Tian, Junchao Zhang, Xiaoyan Jiang, Jianmin College of Computer Science and Software Engineering Shenzhen University Shenzhen518000 China

Following the publication of EEG-ImageNet with 40 class labels by Spampinato et. al. [1], where natural images with multiple semantics are directly used to stimulate human brains in a deterministic order, extensive research has been attempted and widely reported in the literature over the topic of decoding human brains via EEG classifications, making people feel optimistic that mind reading via computational intelligence is emerging on the horizon [2, 3, 4, 5]. In 2020, however, Li et al. [6] published their investigation over EEG-ImageNet and reported that, when the order of brain stimulation is randomized, the classification rate in decoding brain perception of natural images drops to chances [7]. Ahmed et al. then updated EEG-ImageNet by increasing the number of samples to 10 times more and tested a range of classifiers, where the maximum classification rate achieved is 7% [8], presenting an extremely difficult challenge for this direction of *** this paper, we report our latest attempt to address this challenge with a new deep framework. Compared with the existing efforts, our proposed framework features in: (i) representational projection to transform decoding brain perception of natural images into a problem of image classification, paving the way for introducing ResNet as the primary backbone to exploit its success in image classification across the areas of computer vision and pattern recognition;(ii) replacing ReLU with ELU to increase its robustness to noises, and add a focused attention to improve its selectivity and intensiveness in learning;(iii) statistics-fused learning to coordinate four ResNet-based deep classifiers and maximize their collective efforts to produce the final decoding output. Extensive experiments and ablation studies are carried out, and the accuracy of classifying the randomized EEG trials has almost doubled the maximum rate reported so far in the literature. © 2025, The Authors. All rights reserved.

关键词： Image enhancement

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：