Procedia Computer Science最新文献_第4页

Influence of process parameters and wall thicknesses on the properties of tool steel 1.2709 (X3NiCoMoTi18-9-5) processed by Selective Laser Melting 工艺参数和壁厚对选择性激光熔化加工1.2709 （X3NiCoMoTi18-9-5）工具钢性能的影响

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.147

Sara Halilovic , Norbert Wild , Marko Orsolic , Aziz Huskic

Selective Laser Melting (SLM) has gained significant importance as a manufacturing process for complex geometries and high-precision components. For tool steels such as 1.2709, which are valued for their high strength and hardness, optimizing the process parameters is essential to achieve the desired mechanical properties and minimize defects. The ability to control factors like wall thickness and energy density is essential for producing thin-walled components with high structural integrity and reliability. In this study, the influence of the selected process parameter and wall thickness on the fabrication of tool steel 1.2709 by SLM was investigated. The tool steel was processed using two different process parameters and eleven wall thicknesses ranging from 0.2 to 1.0 mm. The two process parameters V1 (87 J/mm³) and V2 (48 J/mm³) differ significantly in their energy density. The specimens were examined for their final wall thickness. Furthermore, the influence of the process parameter and the wall thickness on the mechanical properties, the defects and the microstructure were determined. The investigations revealed a fine dendritic microstructure in the as-built condition. Depending on the process parameter, the mechanical properties are weakened by the porosity observed in thinner wall thicknesses.

选择性激光熔化（SLM）作为一种复杂几何形状和高精度部件的制造工艺，具有重要的意义。对于工具钢，如1.2709，其价值在于其高强度和硬度，优化工艺参数是必不可少的，以达到理想的机械性能和最小化缺陷。控制壁厚和能量密度等因素的能力对于生产具有高结构完整性和可靠性的薄壁部件至关重要。研究了工艺参数的选择和壁厚对工具钢1.2709的SLM成形的影响。采用两种不同的工艺参数和从0.2到1.0 mm的11种壁厚对工具钢进行加工。两个工艺参数V1 （87 J/mm3）和V2 （48 J/mm3）的能量密度差异显著。检验了这些标本的最终壁厚。进一步研究了工艺参数和壁厚对合金力学性能、缺陷和显微组织的影响。研究表明，在建造条件下具有良好的枝晶微观结构。根据工艺参数的不同，在较薄的壁厚中观察到的孔隙率会削弱机械性能。

{"title":"Influence of process parameters and wall thicknesses on the properties of tool steel 1.2709 (X3NiCoMoTi18-9-5) processed by Selective Laser Melting","authors":"Sara Halilovic , Norbert Wild , Marko Orsolic , Aziz Huskic","doi":"10.1016/j.procs.2025.01.147","DOIUrl":"10.1016/j.procs.2025.01.147","url":null,"abstract":"<div><div>Selective Laser Melting (SLM) has gained significant importance as a manufacturing process for complex geometries and high-precision components. For tool steels such as 1.2709, which are valued for their high strength and hardness, optimizing the process parameters is essential to achieve the desired mechanical properties and minimize defects. The ability to control factors like wall thickness and energy density is essential for producing thin-walled components with high structural integrity and reliability. In this study, the influence of the selected process parameter and wall thickness on the fabrication of tool steel 1.2709 by SLM was investigated. The tool steel was processed using two different process parameters and eleven wall thicknesses ranging from 0.2 to 1.0 mm. The two process parameters V1 (87 J/mm<sup>3</sup>) and V2 (48 J/mm<sup>3</sup>) differ significantly in their energy density. The specimens were examined for their final wall thickness. Furthermore, the influence of the process parameter and the wall thickness on the mechanical properties, the defects and the microstructure were determined. The investigations revealed a fine dendritic microstructure in the as-built condition. Depending on the process parameter, the mechanical properties are weakened by the porosity observed in thinner wall thicknesses.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"253 ","pages":"Pages 853-862"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143480507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Circular Data Service Cards: A Card-Based Ideation Tool For Data Services Supporting The Twin Transition 循环数据服务卡：支持双转换的数据服务的基于卡片的构思工具

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.150

Gert Breitfuss , Lilia Yang , Viktoria Pammer-Schindler , Leonie Disch

The Twin Transition encompasses the progression of both digital and green transformations. The integration of data science, data analytics, and data services with Industry 4.0 principles significantly enhances the operational efficiency, decision-making, and sustainability of manufacturing systems. In the context of green transformation, Circular Economy (CE) business models aim to optimize resource use and reduce environmental impact. The development of data services and the application of artificial intelligence (AI) are essential for CE, as these technologies improve efficiency, traceability, and resource optimization. This paper presents the development process of the Circular Data Service Cards (DSC), an extension card set (20 newly designed cards, one new category) to the existing Data Service Cards (50 cards, grouped into 5 categories) to assist the co-creation process in developing data services that support the circular economy. The cards address the challenges of interdisciplinary collaboration (user-centered service design, data science and circular economy) and varying expertise levels essential for building a circular data-driven business. Alongside the developed sub-categories (new cards), the outcomes of this study provide a valuable enhancement of the existing DSC. Initial evaluation results indicate that the Circular DSC are perceived as both useful and user-friendly. This research contributes to the twin transition by providing an actionable tool to support the digital transformation and the development of circular data-driven services.

双转型包括数字化和绿色转型的进程。数据科学、数据分析和数据服务与工业4.0原则的集成显著提高了制造系统的运营效率、决策和可持续性。在绿色转型的背景下，循环经济（CE）商业模式旨在优化资源利用和减少环境影响。数据服务的发展和人工智能（AI）的应用对CE至关重要，因为这些技术提高了效率、可追溯性和资源优化。本文介绍了循环数据服务卡（DSC）的开发过程，这是现有数据服务卡（50张卡，分为5类）的扩展卡组（20张新设计卡，一个新类别），以协助共同创造过程，开发支持循环经济的数据服务。这些卡解决了跨学科合作（以用户为中心的服务设计、数据科学和循环经济）的挑战，以及建立循环数据驱动业务所必需的不同专业水平。除了开发的子类别（新卡）外，本研究的结果对现有的DSC提供了有价值的增强。初步评估结果表明，循环DSC被认为既有用又用户友好。本研究通过提供一个可操作的工具来支持数字化转型和循环数据驱动服务的发展，为双重转型做出了贡献。

{"title":"Circular Data Service Cards: A Card-Based Ideation Tool For Data Services Supporting The Twin Transition","authors":"Gert Breitfuss , Lilia Yang , Viktoria Pammer-Schindler , Leonie Disch","doi":"10.1016/j.procs.2025.01.150","DOIUrl":"10.1016/j.procs.2025.01.150","url":null,"abstract":"<div><div>The Twin Transition encompasses the progression of both digital and green transformations. The integration of data science, data analytics, and data services with Industry 4.0 principles significantly enhances the operational efficiency, decision-making, and sustainability of manufacturing systems. In the context of green transformation, Circular Economy (CE) business models aim to optimize resource use and reduce environmental impact. The development of data services and the application of artificial intelligence (AI) are essential for CE, as these technologies improve efficiency, traceability, and resource optimization. This paper presents the development process of the Circular Data Service Cards (DSC), an extension card set (20 newly designed cards, one new category) to the existing Data Service Cards (50 cards, grouped into 5 categories) to assist the co-creation process in developing data services that support the circular economy. The cards address the challenges of interdisciplinary collaboration (user-centered service design, data science and circular economy) and varying expertise levels essential for building a circular data-driven business. Alongside the developed sub-categories (new cards), the outcomes of this study provide a valuable enhancement of the existing DSC. Initial evaluation results indicate that the Circular DSC are perceived as both useful and user-friendly. This research contributes to the twin transition by providing an actionable tool to support the digital transformation and the development of circular data-driven services.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"253 ","pages":"Pages 882-891"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143480510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Classification of Sarcoma Based on Genomic Data Using Machine Learning Models 基于基因组数据的肉瘤分类使用机器学习模型

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2024.12.034

Pratham Gala , Yash Pandloskar , Shubham Godbole , Fayed Hakim , Pratik Kanani , Lakshmi Kurup

The proposed work provides a new machine-learnt classification approach for the various types of soft tissue sarcoma based on genomics data which addresses a considerable gap in sarcoma diagnostics. The previous studies have investigated various aspects of sarcoma but this study is unique in that it targets the predicting sarcoma variant types using genetic information, which has not been done before. Random Forest was used as the meta-estimator and a stacking ensemble model comprising of Random Forest, Extreme Gradient Boosting and LightGBM were used for this study. The model which was trained and validated on a complete dataset of 206 adult soft tissue sarcoma samples containing genomic alterations, transcriptomic, epigenomic and proteomic data achieved an accuracy of 89.44% at a precision level as high as 91%. Stratified k-fold cross validation is employed to ensure that class imbalance is not a hindrance to performance. This innovative approach outmatches single classifiers and traditional single model methods at great length hence making it possible and effective to use machine learning on genomic data for predicting sarcoma variants. Thus, the findings from this research could change cancer diagnosis forever; they promise more accurate classification as well as personalized treatment modalities while also providing a framework for analogous applications in other rare complex cancers.

提出的工作为基于基因组学数据的各种类型的软组织肉瘤提供了一种新的机器学习分类方法，这解决了肉瘤诊断中相当大的空白。以往的研究已经研究了肉瘤的各个方面，但本研究的独特之处在于，它针对的是利用遗传信息预测肉瘤的变异类型，这是以前从未做过的。采用随机森林作为元估计量，采用随机森林、极端梯度增强和LightGBM组成的叠加集成模型进行研究。该模型在包含基因组改变、转录组学、表观基因组学和蛋白质组学数据的206个成人软组织肉瘤样本完整数据集上进行训练和验证，准确率达到89.44%，精度水平高达91%。采用分层k-fold交叉验证来确保类不平衡不会妨碍性能。这种创新的方法在很大程度上优于单一分类器和传统的单一模型方法，从而使机器学习在基因组数据上预测肉瘤变异成为可能和有效。因此，这项研究的发现可能永远改变癌症的诊断；他们承诺更准确的分类和个性化的治疗方式，同时也为其他罕见的复杂癌症的类似应用提供了一个框架。

{"title":"Classification of Sarcoma Based on Genomic Data Using Machine Learning Models","authors":"Pratham Gala , Yash Pandloskar , Shubham Godbole , Fayed Hakim , Pratik Kanani , Lakshmi Kurup","doi":"10.1016/j.procs.2024.12.034","DOIUrl":"10.1016/j.procs.2024.12.034","url":null,"abstract":"<div><div>The proposed work provides a new machine-learnt classification approach for the various types of soft tissue sarcoma based on genomics data which addresses a considerable gap in sarcoma diagnostics. The previous studies have investigated various aspects of sarcoma but this study is unique in that it targets the predicting sarcoma variant types using genetic information, which has not been done before. Random Forest was used as the meta-estimator and a stacking ensemble model comprising of Random Forest, Extreme Gradient Boosting and LightGBM were used for this study. The model which was trained and validated on a complete dataset of 206 adult soft tissue sarcoma samples containing genomic alterations, transcriptomic, epigenomic and proteomic data achieved an accuracy of 89.44% at a precision level as high as 91%. Stratified k-fold cross validation is employed to ensure that class imbalance is not a hindrance to performance. This innovative approach outmatches single classifiers and traditional single model methods at great length hence making it possible and effective to use machine learning on genomic data for predicting sarcoma variants. Thus, the findings from this research could change cancer diagnosis forever; they promise more accurate classification as well as personalized treatment modalities while also providing a framework for analogous applications in other rare complex cancers.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 317-330"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Implementation of Multi-Label Fuzzy Classification System using Topic Detection Data set 基于主题检测数据集的多标签模糊分类系统实现

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2024.12.003

R. Kanagaraj , N. Krishnaraj , J. Selvakumar , J. Ramprasath

Multiclass Classification can be implemented by using consequent approaches to translate the multiclass problem into binary class classification problems and fuzzy classification methods. This work proposes a predictive analysis of the multiclass fuzzy Classification integrated with time series historical data and topic detection. The fuzzy classification techniques can be successfully applied to Topic detection and sub-topic detection. Text databases’ manual topic detection method must be more feasible, uncontrollable and effective. Thus, initiating the huge amount of data implemented by manual methods is idealistic. Fuzzy historical data is more significant for data analysis in different models to make predictions. Innumerable fuzzy logic on time series methods has been implemented for data prediction. A Multiclass Fuzzy Time Series Classification Algorithm has been implemented to analyze and predict the topic detection database. The outcomes of the Fuzzy classification technique have been implemented for the need for an extensive pattern of topic detection. An enhanced Multiclass Fuzzy Time Series Classification Algorithm has been applied to achieve the efficient de-fuzzification operation of the topic detection data set. To illuminate the forecasting method, the historical data of multi-labeled has been used for the predictive model. The investigation result illustrates that the MHTSC algorithm generates mode fuzzy classification and irregular rules, efficiently reducing the error rate from multi-labeled data.

多类分类可以通过使用顺次方法将多类问题转化为二分类问题和模糊分类方法来实现。本文提出了一种结合时间序列历史数据和主题检测的多类模糊分类预测分析方法。模糊分类技术可以成功地应用于主题检测和子主题检测。文本数据库的人工主题检测方法必须更具可行性、不可控性和有效性。因此，初始化通过手工方法实现的大量数据是理想的。模糊历史数据对于不同模型下的数据分析进行预测更为重要。时间序列上的无数模糊逻辑方法已被用于数据预测。采用多类模糊时间序列分类算法对主题检测数据库进行分析和预测。由于需要广泛的主题检测模式，模糊分类技术的结果已经实现。采用增强型多类模糊时间序列分类算法对主题检测数据集进行高效的去模糊化处理。为了说明这种预测方法，我们将多标签的历史数据用于预测模型。研究结果表明，MHTSC算法生成模式模糊分类和不规则规则，有效地降低了多标签数据的错误率。

{"title":"Implementation of Multi-Label Fuzzy Classification System using Topic Detection Data set","authors":"R. Kanagaraj , N. Krishnaraj , J. Selvakumar , J. Ramprasath","doi":"10.1016/j.procs.2024.12.003","DOIUrl":"10.1016/j.procs.2024.12.003","url":null,"abstract":"<div><div>Multiclass Classification can be implemented by using consequent approaches to translate the multiclass problem into binary class classification problems and fuzzy classification methods. This work proposes a predictive analysis of the multiclass fuzzy Classification integrated with time series historical data and topic detection. The fuzzy classification techniques can be successfully applied to Topic detection and sub-topic detection. Text databases’ manual topic detection method must be more feasible, uncontrollable and effective. Thus, initiating the huge amount of data implemented by manual methods is idealistic. Fuzzy historical data is more significant for data analysis in different models to make predictions. Innumerable fuzzy logic on time series methods has been implemented for data prediction. A Multiclass Fuzzy Time Series Classification Algorithm has been implemented to analyze and predict the topic detection database. The outcomes of the Fuzzy classification technique have been implemented for the need for an extensive pattern of topic detection. An enhanced Multiclass Fuzzy Time Series Classification Algorithm has been applied to achieve the efficient de-fuzzification operation of the topic detection data set. To illuminate the forecasting method, the historical data of multi-labeled has been used for the predictive model. The investigation result illustrates that the MHTSC algorithm generates mode fuzzy classification and irregular rules, efficiently reducing the error rate from multi-labeled data.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 15-24"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A GAN-Enhanced Multimodal Diagnostic Framework Utilizing an Ensemble of BiLSTM, BiGRU, and RNN Models for Malaria and Dengue Detection 利用BiLSTM， BiGRU和RNN模型集成的gan增强多模态诊断框架用于疟疾和登革热检测

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2024.12.039

Rathnakar Achary, Chetan J Shelke, Alluru Lekhya

Quick detection of Malaria and Dengue is crucial for doctors to start treatment and manage patients effectively. As patient conditions become more complex with overlapping symptoms, traditional diagnostic tools become inefficient, slow, and less accurate. Modernizing diagnostics with AI-powered systems is essential. Inaccurate or delayed diagnoses lead to transmission and sustained spread of these diseases. Improving diagnostic tools with accuracy, precision, recall, and speed enhances patient outcomes, reduces infection spread, and streamlines health sector operations. Despite advances, current diagnostic algorithms have weaknesses, especially in applying machine learning to diverse datasets at granular levels. Continuous effort is needed to improve accuracy and recall. This research proposes a GAN-Based Synthesized Multimodal Diagnostic System, combining BiLSTM, BiGRU, and RNN approaches. Utilizing GANs for data augmentation and recurrent networks, this framework shows innovative infectious disease detection. It improves diagnostic precision by 4.9%, accuracy by 3.5%, recall by 3.5%, and AUC by 4.5%, while reducing the gap between disease progression and detection by 8.3%. These outcomes can reduce triage time, misdiagnoses, and lead to faster, quality healthcare. The GAN-Enhanced Multimodal Diagnostic Framework shows promise for diagnosing Malaria, Dengue, and other infectious diseases.

快速发现疟疾和登革热对于医生开始治疗和有效管理患者至关重要。随着患者病情变得更加复杂，症状重叠，传统的诊断工具变得低效、缓慢和不准确。利用人工智能驱动的系统实现诊断现代化至关重要。不准确或延误的诊断导致这些疾病的传播和持续传播。提高诊断工具的准确性、精确性、召回率和速度，可以改善患者的治疗效果，减少感染传播，并简化卫生部门的运作。尽管取得了进步，但目前的诊断算法仍存在弱点，特别是在将机器学习应用于粒度级别的各种数据集方面。需要持续的努力来提高准确性和召回率。本研究提出一种基于gan的综合多模态诊断系统，结合BiLSTM、BiGRU和RNN方法。利用gan进行数据增强和循环网络，该框架显示了创新的传染病检测。它将诊断精度提高了4.9%，准确度提高了3.5%，召回率提高了3.5%，AUC提高了4.5%，同时将疾病进展和检测之间的差距减少了8.3%。这些结果可以减少分诊时间和误诊，并带来更快、更高质量的医疗保健。gan增强型多模式诊断框架有望用于诊断疟疾、登革热和其他传染病。

{"title":"A GAN-Enhanced Multimodal Diagnostic Framework Utilizing an Ensemble of BiLSTM, BiGRU, and RNN Models for Malaria and Dengue Detection","authors":"Rathnakar Achary, Chetan J Shelke, Alluru Lekhya","doi":"10.1016/j.procs.2024.12.039","DOIUrl":"10.1016/j.procs.2024.12.039","url":null,"abstract":"<div><div>Quick detection of Malaria and Dengue is crucial for doctors to start treatment and manage patients effectively. As patient conditions become more complex with overlapping symptoms, traditional diagnostic tools become inefficient, slow, and less accurate. Modernizing diagnostics with AI-powered systems is essential. Inaccurate or delayed diagnoses lead to transmission and sustained spread of these diseases. Improving diagnostic tools with accuracy, precision, recall, and speed enhances patient outcomes, reduces infection spread, and streamlines health sector operations. Despite advances, current diagnostic algorithms have weaknesses, especially in applying machine learning to diverse datasets at granular levels. Continuous effort is needed to improve accuracy and recall. This research proposes a GAN-Based Synthesized Multimodal Diagnostic System, combining BiLSTM, BiGRU, and RNN approaches. Utilizing GANs for data augmentation and recurrent networks, this framework shows innovative infectious disease detection. It improves diagnostic precision by 4.9%, accuracy by 3.5%, recall by 3.5%, and AUC by 4.5%, while reducing the gap between disease progression and detection by 8.3%. These outcomes can reduce triage time, misdiagnoses, and lead to faster, quality healthcare. The GAN-Enhanced Multimodal Diagnostic Framework shows promise for diagnosing Malaria, Dengue, and other infectious diseases.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 381-393"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Stray Dog Detection System using YOLOv5 使用YOLOv5的流浪狗检测系统

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.041

Ashwini Bhosale , Pranav Shinde , Yash Firke , Shivprasad Patil , Pranav Mitake , Samruddhi Shinde

Stray dogs present significant public health and safety risks, particularly in developing countries like India, where the stray dog population is the largest globally. This paper details the implementation of a Stray Dog Detection System using the YOLOv5 object detection model to automatically detect and track stray dogs in real time via CCTV feeds. YOLOv5’s high accuracy and real-time processing capabilities make it well-suited for detecting stray dogs in complex, crowded urban environments. The system leverages a YOLOv5 model trained on custom datasets tailored to local conditions, including specific dog breeds and deployment environments. It integrates an alert mechanism that triggers when stray dog populations surpass predefined thresholds, allowing timely interventions. Additionally, the system incorporates geographic mapping to provide data-driven insights for municipal authorities to manage stray populations effectively and ethically. Experimental results demonstrate an F1 score of 0.97, validating the system’s robustness for practical deployment. This paper discusses system architecture, implementation, and performance, highlighting its scalability and cost-effectiveness for humane stray dog population control.

流浪狗对公共健康和安全构成重大风险，特别是在印度等发展中国家，那里的流浪狗数量是全球最多的。本文详细介绍了利用YOLOv5目标检测模型实现的流浪狗检测系统，通过CCTV视频实时自动检测和跟踪流浪狗。YOLOv5的高精度和实时处理能力使其非常适合在复杂拥挤的城市环境中检测流浪狗。该系统利用了YOLOv5模型，该模型是根据当地条件定制的数据集训练的，包括特定的犬种和部署环境。它集成了一个警报机制，当流浪狗数量超过预定义的阈值时触发，允许及时干预。此外，该系统还结合了地理地图，为市政当局有效和合乎道德地管理流浪人口提供数据驱动的见解。实验结果表明，该系统的F1得分为0.97，验证了系统对实际部署的鲁棒性。本文讨论了系统的架构、实现和性能，强调了其可扩展性和成本效益，用于人道流浪狗种群控制。

{"title":"Stray Dog Detection System using YOLOv5","authors":"Ashwini Bhosale , Pranav Shinde , Yash Firke , Shivprasad Patil , Pranav Mitake , Samruddhi Shinde","doi":"10.1016/j.procs.2025.01.041","DOIUrl":"10.1016/j.procs.2025.01.041","url":null,"abstract":"<div><div>Stray dogs present significant public health and safety risks, particularly in developing countries like India, where the stray dog population is the largest globally. This paper details the implementation of a Stray Dog Detection System using the YOLOv5 object detection model to automatically detect and track stray dogs in real time via CCTV feeds. YOLOv5’s high accuracy and real-time processing capabilities make it well-suited for detecting stray dogs in complex, crowded urban environments. The system leverages a YOLOv5 model trained on custom datasets tailored to local conditions, including specific dog breeds and deployment environments. It integrates an alert mechanism that triggers when stray dog populations surpass predefined thresholds, allowing timely interventions. Additionally, the system incorporates geographic mapping to provide data-driven insights for municipal authorities to manage stray populations effectively and ethically. Experimental results demonstrate an F1 score of 0.97, validating the system’s robustness for practical deployment. This paper discusses system architecture, implementation, and performance, highlighting its scalability and cost-effectiveness for humane stray dog population control.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 806-813"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Influencer Ranking Framework Using TH-DCNN for influence maximization 使用TH-DCNN实现影响力最大化的影响者排名框架

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.018

Vishakha Shelke , Ashish Jadhav Dr.

As the influencer gains more significance in social media marketing, companies raise their budgets for influencer campaigns. With business increasing day by day, finding efficient influencers is becoming the most prominent factor for success, but choosing the right influencer from these social media users is quite a challenge. This manuscript proposes a novel method to rank influencers by their effectiveness based on their posting behavior and social relations over time. Initially, the data from Twitter is collected from the Indian politics tweets and reactions dataset. This raw data undergoes preprocessing using various techniques including, tokenization, stemming, lemmatization, stop word removal, and data normalization using the Min-Max normalization approach to ensure the data is relevant and suitable format for analysis. Next, construct a heterogeneous network to represent the complex interactions between entities like users, tweets, hashtags, and mentions. Then Tree Hierarchical Deep Convolutional Neural Network (TH-DCNN) is applied to these networks to derive information representation for each influencer at each period. Finally, a Cosine similarity (CS) is used to learn from the network and predict the influencer rankings. The performance metrics such as accuracy, f1-score, mean average precision (MAP), Normalized Discounted Cumulative Gain (NDCG), Receiver Operating characteristic (ROC), Mean Reciprocal Rank (MRR), and Hit Rate are analyzed in experimental evaluations. The proposed method improved the accuracy compared with existing techniques.

随着网红在社交媒体营销中的重要性越来越大，公司也会增加网红活动的预算。随着业务的日益增长，找到有效的影响者成为成功的最重要因素，但从这些社交媒体用户中选择合适的影响者是一项相当大的挑战。本文提出了一种新颖的方法，根据他们的发布行为和社会关系的有效性对影响者进行排名。最初，Twitter上的数据是从印度政治推文和反应数据集中收集的。使用各种技术对原始数据进行预处理，包括标记化、词干提取、词序化、停止词删除和使用最小-最大规范化方法的数据规范化，以确保数据是相关的和适合分析的格式。接下来，构建一个异构网络来表示用户、tweet、hashtag和提及等实体之间的复杂交互。然后将树层次深度卷积神经网络（TH-DCNN）应用于这些网络，得到每个影响者在每个时期的信息表示。最后，使用余弦相似度（CS）从网络中学习并预测网红排名。在实验评估中分析了准确率、f1分数、平均平均精度（MAP）、归一化贴现累积增益（NDCG）、接收者工作特性（ROC）、平均倒数秩（MRR）和命中率等性能指标。与现有技术相比，该方法提高了精度。

{"title":"Influencer Ranking Framework Using TH-DCNN for influence maximization","authors":"Vishakha Shelke , Ashish Jadhav Dr.","doi":"10.1016/j.procs.2025.01.018","DOIUrl":"10.1016/j.procs.2025.01.018","url":null,"abstract":"<div><div>As the influencer gains more significance in social media marketing, companies raise their budgets for influencer campaigns. With business increasing day by day, finding efficient influencers is becoming the most prominent factor for success, but choosing the right influencer from these social media users is quite a challenge. This manuscript proposes a novel method to rank influencers by their effectiveness based on their posting behavior and social relations over time. Initially, the data from Twitter is collected from the Indian politics tweets and reactions dataset. This raw data undergoes preprocessing using various techniques including, tokenization, stemming, lemmatization, stop word removal, and data normalization using the Min-Max normalization approach to ensure the data is relevant and suitable format for analysis. Next, construct a heterogeneous network to represent the complex interactions between entities like users, tweets, hashtags, and mentions. Then Tree Hierarchical Deep Convolutional Neural Network (TH-DCNN) is applied to these networks to derive information representation for each influencer at each period. Finally, a Cosine similarity (CS) is used to learn from the network and predict the influencer rankings. The performance metrics such as accuracy, f1-score, mean average precision (MAP), Normalized Discounted Cumulative Gain (NDCG), Receiver Operating characteristic (ROC), Mean Reciprocal Rank (MRR), and Hit Rate are analyzed in experimental evaluations. The proposed method improved the accuracy compared with existing techniques.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 583-592"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Location Verification of Wireless Sensor Node using Integrated Trilateration in Outdoor WSN 户外WSN中基于集成三边测量的无线传感器节点位置验证

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.016

Arjun , Supreeth N M , Akhil K M , Sougandh Sunil

This paper presents a novel method for enhancing localization accuracy in Wireless Sensor Networks (WSNs) through an improved trilateration approach. Despite advancements in localization techniques, challenges remain in achieving reliable accuracy, particularly in complex environments. The proposed method enhances traditional trilateration by integrating angle of arrival (AoA) measurements, leading to better positioning of sensor nodes. In this study, the aim is to confirm whether the final coordinates of trilateration can be supported by adding residual analysis and AoA observations. Experiments were conducted to assess the localization system’s effectiveness. The results showed that residual validation outperformed AoA localization, particularly in noisy outdoor environments, providing more reliable distance estimates even under challenging conditions. This method enhances the trustworthiness of the localization system while also minimizing hardware needs and reducing computational complexity, making it a practical choice for resource-limited WSNs. The AoA verification method achieved average accuracies of 54% for x-coordinates and 55% for y-coordinates. In contrast, incorporating residual analysis improved these figures to 79% for x-coordinates and 80% for y-coordinates. These insights focus on the localization process and demonstrate the value of residual analysis in boosting system performance.

提出了一种利用改进的三边定位方法提高无线传感器网络定位精度的新方法。尽管定位技术取得了进步，但在实现可靠的精度方面仍然存在挑战，特别是在复杂的环境中。该方法通过积分到达角（AoA）测量值来改进传统的三边测量方法，从而更好地定位传感器节点。在本研究中，目的是确认是否可以通过添加残差分析和AoA观测来支持三边测量的最终坐标。通过实验验证了该定位系统的有效性。结果表明，残差验证优于AoA定位，特别是在嘈杂的室外环境中，即使在具有挑战性的条件下也能提供更可靠的距离估计。该方法提高了定位系统的可信度，同时最大限度地减少了硬件需求和计算复杂度，是资源有限的无线传感器网络的实用选择。AoA验证方法的x坐标平均精度为54%，y坐标平均精度为55%。相比之下，结合残差分析将这些数字提高到x坐标的79%和y坐标的80%。这些见解集中于定位过程，并展示了残差分析在提高系统性能方面的价值。

{"title":"Location Verification of Wireless Sensor Node using Integrated Trilateration in Outdoor WSN","authors":"Arjun , Supreeth N M , Akhil K M , Sougandh Sunil","doi":"10.1016/j.procs.2025.01.016","DOIUrl":"10.1016/j.procs.2025.01.016","url":null,"abstract":"<div><div>This paper presents a novel method for enhancing localization accuracy in Wireless Sensor Networks (WSNs) through an improved trilateration approach. Despite advancements in localization techniques, challenges remain in achieving reliable accuracy, particularly in complex environments. The proposed method enhances traditional trilateration by integrating angle of arrival (AoA) measurements, leading to better positioning of sensor nodes. In this study, the aim is to confirm whether the final coordinates of trilateration can be supported by adding residual analysis and AoA observations. Experiments were conducted to assess the localization system’s effectiveness. The results showed that residual validation outperformed AoA localization, particularly in noisy outdoor environments, providing more reliable distance estimates even under challenging conditions. This method enhances the trustworthiness of the localization system while also minimizing hardware needs and reducing computational complexity, making it a practical choice for resource-limited WSNs. The AoA verification method achieved average accuracies of 54% for x-coordinates and 55% for y-coordinates. In contrast, incorporating residual analysis improved these figures to 79% for x-coordinates and 80% for y-coordinates. These insights focus on the localization process and demonstrate the value of residual analysis in boosting system performance.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 567-575"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376955","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Hybrid Binary SGO-GA for solving MAX-SAT problem 求解MAX-SAT问题的混合二进制SGO-GA

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.055

Rhiddhi Prasad Das , Anuruddha Paul , Junali Jasmine Jena , Bibhuti Bhusan Dash , Utpal Chandra De , Mahendra Kumar Gourisaria

The Maximum Satisfiability Problem (MAX-SAT) is a crucial NP-hard optimization problem with applications in artificial intelligence, circuit design, scheduling, and combinatorial optimization. In this work, we provide a unique hybrid strategy that blends Genetic Algorithms (GA) with Social Group Optimization (SGO) algorithm to effectively solve the MAX-SAT problem. The SGO algorithm, inspired by the social behavior of groups, excels in exploring diverse regions of the search space. w used a binary variant of SGO i.e. Binary-SGO which is defined specifically for binary search spaces, while GA leverages evolutionary principles to exploit local optima through selection, crossover, and mutation. By integrating the exploration capabilities of SGO with the exploitation strengths of GA, the hybrid approach strikes an optimal balance between global and local search. Extensive experimental evaluations conducted on standard MAX-SAT benchmarks demonstrate that our hybrid algorithm outperforms several existing state-of-the-art meta-heuristic algorithms. Hybrid BSGO-GA achieved the highest average fitness values, with an average accuracy of 99.7% in Experiment 1, 99.61% in Experiment 2, and 99.21% in Experiment 3 and achieved complete satisfiability in 55 out of 75 cases in Experiment 1, 42 out of 75 cases in Experiment 2, and 7 out of 75 cases in Experiment 3. This approach demonstrates the potential of hybrid metaheuristics in addressing complex optimization problems and offers a robust framework for tackling other NP-hard problems.

最大可满足性问题（MAX-SAT）是一个重要的NP-hard优化问题，在人工智能、电路设计、调度和组合优化等领域都有应用。在这项工作中，我们提供了一种独特的混合策略，将遗传算法（GA）与社会群体优化（SGO）算法相结合，有效地解决了MAX-SAT问题。SGO算法受到群体社会行为的启发，擅长探索搜索空间的不同区域。w使用了SGO的二进制变体，即binary -SGO，它是专门为二进制搜索空间定义的，而GA利用进化原理通过选择、交叉和突变来利用局部最优。该混合方法将SGO的搜索能力与遗传算法的挖掘能力相结合，实现了全局搜索与局部搜索的最佳平衡。在标准MAX-SAT基准测试上进行的大量实验评估表明，我们的混合算法优于几种现有的最先进的元启发式算法。混合BSGO-GA获得了最高的平均适应度值，实验1的平均准确率为99.7%，实验2为99.61%，实验3为99.21%，实验1的75例中有55例完全满意，实验2的75例中有42例完全满意，实验3的75例中有7例完全满意。这种方法展示了混合元启发式在解决复杂优化问题方面的潜力，并为解决其他np困难问题提供了一个强大的框架。

{"title":"Hybrid Binary SGO-GA for solving MAX-SAT problem","authors":"Rhiddhi Prasad Das , Anuruddha Paul , Junali Jasmine Jena , Bibhuti Bhusan Dash , Utpal Chandra De , Mahendra Kumar Gourisaria","doi":"10.1016/j.procs.2025.01.055","DOIUrl":"10.1016/j.procs.2025.01.055","url":null,"abstract":"<div><div>The Maximum Satisfiability Problem (MAX-SAT) is a crucial NP-hard optimization problem with applications in artificial intelligence, circuit design, scheduling, and combinatorial optimization. In this work, we provide a unique hybrid strategy that blends Genetic Algorithms (GA) with Social Group Optimization (SGO) algorithm to effectively solve the MAX-SAT problem. The SGO algorithm, inspired by the social behavior of groups, excels in exploring diverse regions of the search space. w used a binary variant of SGO i.e. Binary-SGO which is defined specifically for binary search spaces, while GA leverages evolutionary principles to exploit local optima through selection, crossover, and mutation. By integrating the exploration capabilities of SGO with the exploitation strengths of GA, the hybrid approach strikes an optimal balance between global and local search. Extensive experimental evaluations conducted on standard MAX-SAT benchmarks demonstrate that our hybrid algorithm outperforms several existing state-of-the-art meta-heuristic algorithms. Hybrid BSGO-GA achieved the highest average fitness values, with an average accuracy of 99.7% in Experiment 1, 99.61% in Experiment 2, and 99.21% in Experiment 3 and achieved complete satisfiability in 55 out of 75 cases in Experiment 1, 42 out of 75 cases in Experiment 2, and 7 out of 75 cases in Experiment 3. This approach demonstrates the potential of hybrid metaheuristics in addressing complex optimization problems and offers a robust framework for tackling other NP-hard problems.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"252 ","pages":"Pages 944-953"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143377128","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A neural network approach for collaborative cells: an innovative online rescheduling strategy for maximizing productivity 协作细胞的神经网络方法：一种创新的在线重调度策略，以最大限度地提高生产率

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.103

Irene Granata , Matthias Bues , Martina Calzavara , Maurizio Faccio , Benjamin Wingert

Transitioning from Industry 4.0 to Industry 5.0 signifies a significant change in how technology integrates with workplace dynamics. While Industry 4.0 focused on streamlining production through automation, Industry 5.0 centers on human-centric approaches. This entails designing work environments that prioritize human comfort and efficiency by incorporating technology that complements human capabilities. Collaborative robots, known as cobots, play a pivotal role in this shift, aiding humans in tasks while fostering increased human involvement. However, maximizing the benefits of cobots necessitates workspace designs that optimize both human and robotic resources’ needs and preferences. A promising strategy involves implementing a dynamic task allocation system. This approach employs a neural network to adaptively reallocate tasks to prevent any loss in performance. Such advancements represent a significant stride towards establishing production settings that prioritize the effectiveness of human workers.

从工业4.0到工业5.0的过渡意味着技术如何与工作场所动态相结合的重大变化。工业4.0侧重于通过自动化简化生产，而工业5.0则侧重于以人为本的方法。这需要设计工作环境，通过结合技术来补充人类的能力，优先考虑人类的舒适和效率。协作机器人（cobots）在这一转变中发挥着关键作用，它们在帮助人类完成任务的同时，促进了人类的参与度。然而，最大化协作机器人的好处需要工作空间的设计，优化人力和机器人资源的需求和偏好。一个有前景的策略包括实现一个动态任务分配系统。该方法采用神经网络自适应重新分配任务，以防止性能损失。这些进步代表着朝着建立优先考虑人类工人效率的生产环境迈出了重要的一步。

引用次数: 0