Procedia Computer Science最新文献_第2页

HerbSimNet: Deep Learning -Based Classification of Indian Medicinal Plants with High Inter-Class Similarities HerbSimNet：基于深度学习的高类间相似性印度药用植物分类

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.309

N. Shobha Rani , Bhavya K R , Pushpa B.R. , Ragavendra M. Devadas

Medicinal plant species recognition is important across diverse sectors such as Ayurveda, agriculture, environment conservation and botanical research. Specific groups of plants in Indian medicinal plant ecosystem exhibit significant inter-class similarities due to varying abundance and ecological factors. To address the challenges involved in the process of classifying these species in this work a deep learning model Herb-SimNet is proposed. The Herb-SimNet analyzes similarity of plant species over other plant species using vision based deep learning and machine learning techniques. The proposed model works based on the combination of wavelet features and convolutional features extracted using three sequential convolution layers to extract the prominent features that distinguish variations among the inter class similarity plant species. To perform experiments, a dataset is created by capturing medicinal plant leaf images using box model in plain background and uniform lighting. A smart phone captured twelve Indian medicinal plant species comprising of about 1400+ samples that belongs different plant species but similar morphological structure is collected. Baseline experiments are carried out between Herb-SimNet and other state-of-the-art deep learning models for classification based on the proposed dataset. The outcomes demonstrate that Herb_SimNet provides clear interpretation one plant variety with others and achieves superior accuracy in prediction than that of state-of-the-art approaches. Furthermore, the model demonstrates better generalization towards the other inter-class similarity groups considered for testing. In conclusion, the proposed dataset and Herb-SimNet plays a a crucial role in advancement of research concerning Indian medicinal plant species classification resulting into enhancement of AI-based technology for biodiversity conservation and ethnobotanical studies.

药用植物物种识别在阿育吠陀、农业、环境保护和植物学研究等不同领域都很重要。在印度药用植物生态系统中，由于不同的丰度和生态因子，某些植物类群表现出显著的类间相似性。为了解决在这项工作中对这些物种进行分类的过程中所涉及的挑战，提出了一个深度学习模型Herb-SimNet。Herb-SimNet使用基于视觉的深度学习和机器学习技术分析植物物种与其他植物物种的相似性。该模型基于小波特征和卷积特征的结合，利用三个连续卷积层提取卷积特征，提取出区分类间相似植物物种差异的显著特征。为了进行实验，在普通背景和均匀光照条件下，使用盒模型捕获药用植物叶片图像，创建数据集。智能手机采集了12种印度药用植物，包括约1400+样本，属于不同的植物物种，但形态结构相似。在Herb-SimNet和其他最先进的深度学习模型之间进行基线实验，以基于所提出的数据集进行分类。结果表明，Herb_SimNet可以清晰地解释一个植物品种与其他植物品种，并且在预测方面取得了比现有方法更高的准确性。此外，该模型对用于测试的其他类间相似性组表现出更好的泛化。综上所述，该数据集和Herb-SimNet在印度药用植物物种分类研究中发挥了至关重要的作用，从而增强了基于人工智能的生物多样性保护和民族植物学研究技术。

{"title":"HerbSimNet: Deep Learning -Based Classification of Indian Medicinal Plants with High Inter-Class Similarities","authors":"N. Shobha Rani , Bhavya K R , Pushpa B.R. , Ragavendra M. Devadas","doi":"10.1016/j.procs.2025.04.309","DOIUrl":"10.1016/j.procs.2025.04.309","url":null,"abstract":"<div><div>Medicinal plant species recognition is important across diverse sectors such as Ayurveda, agriculture, environment conservation and botanical research. Specific groups of plants in Indian medicinal plant ecosystem exhibit significant inter-class similarities due to varying abundance and ecological factors. To address the challenges involved in the process of classifying these species in this work a deep learning model Herb-SimNet is proposed. The Herb-SimNet analyzes similarity of plant species over other plant species using vision based deep learning and machine learning techniques. The proposed model works based on the combination of wavelet features and convolutional features extracted using three sequential convolution layers to extract the prominent features that distinguish variations among the inter class similarity plant species. To perform experiments, a dataset is created by capturing medicinal plant leaf images using box model in plain background and uniform lighting. A smart phone captured twelve Indian medicinal plant species comprising of about 1400+ samples that belongs different plant species but similar morphological structure is collected. Baseline experiments are carried out between Herb-SimNet and other state-of-the-art deep learning models for classification based on the proposed dataset. The outcomes demonstrate that Herb_SimNet provides clear interpretation one plant variety with others and achieves superior accuracy in prediction than that of state-of-the-art approaches. Furthermore, the model demonstrates better generalization towards the other inter-class similarity groups considered for testing. In conclusion, the proposed dataset and Herb-SimNet plays a a crucial role in advancement of research concerning Indian medicinal plant species classification resulting into enhancement of AI-based technology for biodiversity conservation and ethnobotanical studies.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"258 ","pages":"Pages 765-774"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimal resource selection for Green Software Development using Machine Learning 利用机器学习进行绿色软件开发的最佳资源选择

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.298

Nisha Kumari , Tirthankar Gayen

Today software development consumes a lot of natural resources which are needed to be preserved for future needs. The resources that are used in developing software are huge in numbers casting a negative impact on the environment. Hence, one needs to utilize these resources in an efficient manner in order to conserve it. Since resources are limited, there is a need for more improved software as well as an efficient software development process which consumes less energy and resources. In order to fulfill this objective, Green Software Development (GSD) can be useful. But sometimes the cost incurred for the GSD may be too high and benefits obtained may be very less or negligible. This outcome may not be very beneficial to the developers. Therefore, this article proposes an effective approach using machine learning for cost-benefit analysis to provide optimal resource selection for GSD. This approach makes a trade-off between requirements and expenditures (cost incurred to achieve the objective based on the requirements) to provide optimal resource selection and aids in analyzing the economic feasibility for GSD.

今天的软件开发消耗了大量的自然资源，这些资源需要保存起来以备将来需要。用于开发软件的资源数量巨大，对环境产生了负面影响。因此，人们需要以有效的方式利用这些资源，以保护它。由于资源有限，因此需要更多的改进软件以及消耗更少能源和资源的高效软件开发过程。为了实现这一目标，绿色软件开发（GSD）是有用的。但有时，GSD的成本可能过高，而获得的利益可能非常少或微不足道。这种结果可能对开发人员不是很有利。因此，本文提出了一种利用机器学习进行成本效益分析的有效方法，为GSD提供最优的资源选择。这种方法在需求和支出（实现基于需求的目标所产生的成本）之间进行了权衡，以提供最佳的资源选择，并有助于分析GSD的经济可行性。

引用次数: 0

Named Entity Recognition in Assamese Language using two separate models: BiLSTM and BERT 阿萨姆语命名实体识别使用两个独立的模型：BiLSTM和BERT

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.262

Plabita Baruah, Bandana Dutta, Shikhar Kumar Sarma, Kuwali Talukdar

Named Entity Recognition (NER) is a tool based on principles of Artificial Intelligence (AI) and Natural Language Processing (NLP) for automatically tagging Named Entities from unstructured text. In the realm of Natural Language Processing (NLP) applications, Named Entity Recognition (NER) holds significance as it involves the crucial task of identifying and categorizing proper nouns into classes such as person, location, organization, and miscellaneous. While considerable progress has been made in widely spoken languages like English and other European languages, resulting in higher accuracy rates, the task of NER in Indian languages prove to be challenging due to limited resources. This study explores the implementation of NER in Assamese using two separate approaches: BiLSTM and BERT. The proposed methodology achieves an accuracy of 31%in the BiLSTM model. While using BERT, which is a pretrained model, fine-tuned for Assamese, we achieved a precision of 81.5% and F1- score of 0.383. Our comparative analysis shows that both models are effective for NER in a resource-scarce language like Assamese, but BERT performs better overall in recognizing entities. This suggests that BERT could play a key role in improving NER techniques for underrepresented languages.

命名实体识别（NER）是一种基于人工智能（AI）和自然语言处理（NLP）原理的工具，用于从非结构化文本中自动标记命名实体。在自然语言处理（NLP）应用领域，命名实体识别（NER）具有重要意义，因为它涉及识别和分类专有名词的关键任务，如人员、位置、组织和其他类。虽然在英语和其他欧洲语言等广泛使用的语言中取得了相当大的进步，从而提高了准确率，但由于资源有限，印度语言的NER任务被证明是具有挑战性的。本研究使用两种不同的方法：BiLSTM和BERT探讨了在阿萨姆邦实施NER。该方法在BiLSTM模型中达到了31%的准确率。而使用BERT，这是一个针对阿萨姆邦进行微调的预训练模型，我们实现了81.5%的精度和0.383的F1-分数。我们的比较分析表明，这两种模型对于资源稀缺语言（如阿萨姆语）中的NER都是有效的，但BERT在识别实体方面的总体表现更好。这表明BERT可以在改进未被充分代表的语言的NER技术方面发挥关键作用。

{"title":"Named Entity Recognition in Assamese Language using two separate models: BiLSTM and BERT","authors":"Plabita Baruah, Bandana Dutta, Shikhar Kumar Sarma, Kuwali Talukdar","doi":"10.1016/j.procs.2025.04.262","DOIUrl":"10.1016/j.procs.2025.04.262","url":null,"abstract":"<div><div>Named Entity Recognition (NER) is a tool based on principles of Artificial Intelligence (AI) and Natural Language Processing (NLP) for automatically tagging Named Entities from unstructured text. In the realm of Natural Language Processing (NLP) applications, Named Entity Recognition (NER) holds significance as it involves the crucial task of identifying and categorizing proper nouns into classes such as person, location, organization, and miscellaneous. While considerable progress has been made in widely spoken languages like English and other European languages, resulting in higher accuracy rates, the task of NER in Indian languages prove to be challenging due to limited resources. This study explores the implementation of NER in Assamese using two separate approaches: BiLSTM and BERT. The proposed methodology achieves an accuracy of 31%in the BiLSTM model. While using BERT, which is a pretrained model, fine-tuned for Assamese, we achieved a precision of 81.5% and F1- score of 0.383. Our comparative analysis shows that both models are effective for NER in a resource-scarce language like Assamese, but BERT performs better overall in recognizing entities. This suggests that BERT could play a key role in improving NER techniques for underrepresented languages.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"258 ","pages":"Pages 242-251"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Empirical Study on Efficiency of Different Language Modeling Techniques using Masking of Named Entities for Indic Languages 基于命名实体遮蔽的不同语言建模技术对印度语建模效率的实证研究

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.228

Sravan Kumar Reddy, Shailashree K Sheshadri, Krishna Likith Avatapalli, Deepa Gupta

Processing unstructured text in Natural Language Processing (NLP) poses significant challenges for Indic languages, which feature flexible word order, spelling variations, and complex sentence structures. Traditional models often struggle with these complexities, leading to issues such as out-of-vocabulary (OOV) words and increased perplexity. Neural Language Models (NLMs), particularly transformer-based models, address some of these challenges by employing word representations and self-attention mechanisms. However, OOV problems persist, especially with named entities, which are dynamic and vary across domains, making it difficult to create comprehensive lists of names for people, organizations, and locations. To address this, the Masked Entity-Based Language Model (ME-LM) has been introduced, focusing on masking named entities identified through Named Entity Recognition (NER) using pre-trained models like BERT-base-NER and IndicNER. Applied to Indic languages such as Hindi, Kannada, and Telugu for the first time, ME-LM has significantly reduced OOV occurrences by 18.60% to 94.70% and lowered perplexity. Since this is the first application of ME-LM to these languages, no standard benchmark exists for direct comparison, but the results show strong potential for improving named entity handling in these languages.

在自然语言处理（NLP）中处理非结构化文本对印度语提出了重大挑战，印度语具有灵活的词序，拼写变化和复杂的句子结构。传统模型经常与这些复杂性作斗争，导致诸如超出词汇表（OOV）的单词和增加的困惑等问题。神经语言模型（nlm），特别是基于变换的模型，通过使用词表示和自注意机制来解决这些挑战。然而，OOV问题仍然存在，特别是对于命名实体，它们是动态的，并且跨域变化，因此很难为人员、组织和位置创建全面的名称列表。为了解决这个问题，引入了基于屏蔽实体的语言模型（ME-LM），重点是使用BERT-base-NER和IndicNER等预训练模型，屏蔽通过命名实体识别（NER）识别的命名实体。首次应用于印地语、卡纳达语和泰卢固语等印度语言，ME-LM显著减少了OOV的出现，从18.60%减少到94.70%，降低了困惑度。由于这是ME-LM在这些语言中的第一个应用程序，因此没有标准基准可以进行直接比较，但是结果显示了在这些语言中改进命名实体处理的巨大潜力。

{"title":"Empirical Study on Efficiency of Different Language Modeling Techniques using Masking of Named Entities for Indic Languages","authors":"Sravan Kumar Reddy, Shailashree K Sheshadri, Krishna Likith Avatapalli, Deepa Gupta","doi":"10.1016/j.procs.2025.04.228","DOIUrl":"10.1016/j.procs.2025.04.228","url":null,"abstract":"<div><div>Processing unstructured text in Natural Language Processing (NLP) poses significant challenges for Indic languages, which feature flexible word order, spelling variations, and complex sentence structures. Traditional models often struggle with these complexities, leading to issues such as out-of-vocabulary (OOV) words and increased perplexity. Neural Language Models (NLMs), particularly transformer-based models, address some of these challenges by employing word representations and self-attention mechanisms. However, OOV problems persist, especially with named entities, which are dynamic and vary across domains, making it difficult to create comprehensive lists of names for people, organizations, and locations. To address this, the Masked Entity-Based Language Model (ME-LM) has been introduced, focusing on masking named entities identified through Named Entity Recognition (NER) using pre-trained models like BERT-base-NER and IndicNER. Applied to Indic languages such as Hindi, Kannada, and Telugu for the first time, ME-LM has significantly reduced OOV occurrences by 18.60% to 94.70% and lowered perplexity. Since this is the first application of ME-LM to these languages, no standard benchmark exists for direct comparison, but the results show strong potential for improving named entity handling in these languages.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"258 ","pages":"Pages 146-159"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931690","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimized Feature Engineering for Dentition based Cattle Age Estimation 基于牙列的牛龄估计的优化特征工程

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.334

D S Guru , Swaroop D , Anusha P , Keerthana N , Shivaprasad D L

Accurate age estimation of cattle is crucial for effective herd management, breeding, and health monitoring. In this novel study, a unique methodology for cattle age estimation is introduced using high-resolution images of the teeth and canal, captured at local farms and from cow breeders. This approach involves capturing these images, annotating them to distinguish between teeth and canal, and employing a tailored YOLO v9 deep learning model for detection and segmentation, achieving a mean Average Precision (mAP) of 98% with a confidence threshold of 0.5 to 0.95. The teeth and canal regions are prominent in age computation for experts. After segmenting these Regions of Interest (RoI), conventional feature descriptors were used to extract edge features from the segmented images such as Histogram of Oriented Gradients (HOG). Initial linear regression analysis of these features yielded a Root Mean Square Error (RMSE) close to 52. To enhance predictive performance, personalized feature engineering pipelines incorporating advanced feature engineering and selection techniques were developed. This refinement led to a substantial improvement, reducing RMSE to approximately 0.06 with an R² of 0.99 for HOG features. HOG was selected over Convolutional Neural Networks (CNNs) due to its computational efficiency and suitability for resource-constrained environments. HOG demonstrated strong performance with minimal computational requirements, making it well-suited for real-time applications on mobile devices. While CNNs offer potential for future enhancements, our current approach prioritizes practicality and performance for small-scale applications. Our research significantly advances machine-learning-based cattle age prediction, offering a reliable, scalable solution for agricultural practices and also paving the way for future research in this field.

牛的准确年龄估计对于有效的牛群管理、育种和健康监测至关重要。在这项新颖的研究中，引入了一种独特的方法来估计牛的年龄，使用在当地农场和奶牛饲养者那里捕获的牙齿和运河的高分辨率图像。该方法包括捕获这些图像，对它们进行注释以区分牙齿和牙根，并采用量身定制的YOLO v9深度学习模型进行检测和分割，平均平均精度（mAP）达到98%，置信阈值为0.5至0.95。牙齿和牙根管区域是专家计算年龄的重点。在对感兴趣区域（RoI）进行分割后，使用传统的特征描述符从分割后的图像中提取边缘特征，如定向梯度直方图（HOG）。这些特征的初始线性回归分析产生的均方根误差（RMSE）接近52。为了提高预测性能，开发了结合先进特征工程和选择技术的个性化特征工程管道。这种改进带来了实质性的改进，将HOG特征的RMSE降低到大约0.06，R²为0.99。HOG在卷积神经网络（cnn）中被选择是因为它的计算效率和对资源约束环境的适用性。HOG以最小的计算需求展示了强大的性能，使其非常适合移动设备上的实时应用程序。虽然cnn提供了未来增强的潜力，但我们目前的方法优先考虑小规模应用的实用性和性能。我们的研究显著推进了基于机器学习的牛龄预测，为农业实践提供了可靠的、可扩展的解决方案，也为该领域的未来研究铺平了道路。

{"title":"Optimized Feature Engineering for Dentition based Cattle Age Estimation","authors":"D S Guru , Swaroop D , Anusha P , Keerthana N , Shivaprasad D L","doi":"10.1016/j.procs.2025.04.334","DOIUrl":"10.1016/j.procs.2025.04.334","url":null,"abstract":"<div><div>Accurate age estimation of cattle is crucial for effective herd management, breeding, and health monitoring. In this novel study, a unique methodology for cattle age estimation is introduced using high-resolution images of the teeth and canal, captured at local farms and from cow breeders. This approach involves capturing these images, annotating them to distinguish between teeth and canal, and employing a tailored YOLO v9 deep learning model for detection and segmentation, achieving a mean Average Precision (mAP) of 98% with a confidence threshold of 0.5 to 0.95. The teeth and canal regions are prominent in age computation for experts. After segmenting these Regions of Interest (RoI), conventional feature descriptors were used to extract edge features from the segmented images such as Histogram of Oriented Gradients (HOG). Initial linear regression analysis of these features yielded a Root Mean Square Error (RMSE) close to 52. To enhance predictive performance, personalized feature engineering pipelines incorporating advanced feature engineering and selection techniques were developed. This refinement led to a substantial improvement, reducing RMSE to approximately 0.06 with an R² of 0.99 for HOG features. HOG was selected over Convolutional Neural Networks (CNNs) due to its computational efficiency and suitability for resource-constrained environments. HOG demonstrated strong performance with minimal computational requirements, making it well-suited for real-time applications on mobile devices. While CNNs offer potential for future enhancements, our current approach prioritizes practicality and performance for small-scale applications. Our research significantly advances machine-learning-based cattle age prediction, offering a reliable, scalable solution for agricultural practices and also paving the way for future research in this field.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"258 ","pages":"Pages 961-980"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Meta-Heuristic Optimization Algorithms for Resource Allocation in 5G New Radio Networks 5G新型无线网络资源分配的元启发式优化算法

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.277

Jyoti , Amandeep Noliya , Dharmender Kumar

The objective of this research paper is to evaluate effectiveness of various resource allocation algorithms currently used in 5G new radio networks. Due to these complications, the network is experiencing operational difficulties. Incorporating the development trend of 5G into efficient resource management is not only imperative but also requires hardware requirements and improvements to the current network architecture. In order to effectively tackle issue of resource allocation (RA) in a 5G network, primary purpose is to present a proposed scheme for RA that employs learning-based as well as optimization resource allocation methodologies. To ensure effective management of network traffic and operations, resource allocation has emerged as a problematic issue due to the concomitant increase in cellular service demand and the constrained resources at our disposal to provide it. In order to attain the desired level of quality of service (QoS), one of the most critical issues that must be resolved is the reduction of interference activity within the network. This study investigates the subject of resource allocation and optimization and the inspiration for the hunting behavior of meta-heuristic algorithms. This paper evaluates the current 5G NR network resource allocation technique. We formulate the issue of resource allocation as a stochastic optimization problem. Furthermore, throughput and path loss, SNR, and SINR are considered when performing this optimization. The comparison study shows that COA performs best in SNR optimization and FMNS in SINR optimization in resource allocation. Lower standard deviations suggest stability in algorithms like KOA. For effective wireless communication system resource management, the best method relies on network criteria such signal quality and consistency.

本研究论文的目的是评估目前在5G新无线网络中使用的各种资源分配算法的有效性。由于这些复杂因素，该网络正在经历运营困难。将5G的发展趋势融入到高效的资源管理中，不仅势在必行，而且需要硬件要求和对现有网络架构的改进。为了有效地解决5G网络中的资源分配（RA）问题，主要目的是提出一种基于学习和优化资源分配方法的RA方案。为了确保有效地管理网络流量和运营，由于蜂窝服务需求的增加和我们所能提供的资源有限，资源分配已经成为一个问题。为了达到期望的服务质量（QoS）水平，必须解决的最关键问题之一是减少网络内的干扰活动。本研究探讨了资源分配与优化的主题，以及对元启发式算法狩猎行为的启示。本文对当前5G NR网络资源分配技术进行了评估。我们将资源分配问题表述为一个随机优化问题。此外，在执行此优化时还考虑了吞吐量和路径损耗、信噪比和信噪比。对比研究表明，在资源分配中，COA在信噪比优化方面表现最好，FMNS在信噪比优化方面表现最好。较低的标准差表明像KOA这样的算法是稳定的。对于有效的无线通信系统资源管理，最好的方法依赖于信号质量和一致性等网络标准。

{"title":"Meta-Heuristic Optimization Algorithms for Resource Allocation in 5G New Radio Networks","authors":"Jyoti , Amandeep Noliya , Dharmender Kumar","doi":"10.1016/j.procs.2025.04.277","DOIUrl":"10.1016/j.procs.2025.04.277","url":null,"abstract":"<div><div>The objective of this research paper is to evaluate effectiveness of various resource allocation algorithms currently used in 5G new radio networks. Due to these complications, the network is experiencing operational difficulties. Incorporating the development trend of 5G into efficient resource management is not only imperative but also requires hardware requirements and improvements to the current network architecture. In order to effectively tackle issue of resource allocation (RA) in a 5G network, primary purpose is to present a proposed scheme for RA that employs learning-based as well as optimization resource allocation methodologies. To ensure effective management of network traffic and operations, resource allocation has emerged as a problematic issue due to the concomitant increase in cellular service demand and the constrained resources at our disposal to provide it. In order to attain the desired level of quality of service (QoS), one of the most critical issues that must be resolved is the reduction of interference activity within the network. This study investigates the subject of resource allocation and optimization and the inspiration for the hunting behavior of meta-heuristic algorithms. This paper evaluates the current 5G NR network resource allocation technique. We formulate the issue of resource allocation as a stochastic optimization problem. Furthermore, throughput and path loss, SNR, and SINR are considered when performing this optimization. The comparison study shows that COA performs best in SNR optimization and FMNS in SINR optimization in resource allocation. Lower standard deviations suggest stability in algorithms like KOA. For effective wireless communication system resource management, the best method relies on network criteria such signal quality and consistency.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"258 ","pages":"Pages 408-419"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931781","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Reinforcement Learning based Hybrid GR-DQN Model for Predicting Ichthyophthiriosis Disease in Aquaculture Through Water Quality Analysis 基于强化学习的混合GR-DQN模型在水产养殖鱼鳞病预测中的应用

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.04.274

Bhawna Kol , Khetavath Jairam Naik

Aquaculture is a fast-growing industry that provides nutritious food to a growing population and generates substantial revenue for countries. The high water quality is required to be maintained for aquatic animal’s survival and health; otherwise, it may cause many diseases like Furunculosis, Bacterial gill disease, and others. Traditionally available methods for water quality analysis are typically difficult to perform due to being time-consuming and lacking accuracy. In this study, a new approach has been developed using an optimal deep reinforcement learning technique, Hybrid Gated Recurrent Unit (GRU) network with Deep Q-Network (DQN), to analyze the state of the water quality of aquaculture by predicting Ichthyophthiriosis (white spot diseases) in an aquaculture environment. The GRU deep learning model with DQN helps in improving the prediction by approximating Q-values and produces a loss function to guide the learning process; rewards are provided due to correct predictions, thereby disease detection corrected accuracy was enhanced. The proposed hybrid GR-DQN model was implemented on the “Pondsdata” dataset and compared the results with the existing model M-DQN. The Hybrid GR-DQN achieved 94.69% accuracy in comparison to the existing model M-DQN’s 84.16% accuracy on the same dataset.

水产养殖是一个快速发展的产业，为不断增长的人口提供营养食品，并为各国带来可观的收入。水生动物的生存和健康需要保持良好的水质；否则，它可能引起许多疾病，如疖病、细菌性鳃病等。传统上可用的水质分析方法通常由于耗时和缺乏准确性而难以执行。本研究利用最优深度强化学习技术——混合门控循环单元（GRU）网络和深度q -网络（DQN），开发了一种新的方法，通过预测水产养殖环境中的鱼鳞病（白斑病）来分析水产养殖水质状况。带DQN的GRU深度学习模型通过逼近q值来帮助改进预测，并产生损失函数来指导学习过程；对预测正确的人给予奖励，从而提高疾病检测的正确率。在“Pondsdata”数据集上实现了混合GR-DQN模型，并与现有模型M-DQN进行了比较。在相同的数据集上，与现有模型M-DQN的84.16%的准确率相比，Hybrid GR-DQN的准确率达到了94.69%。

{"title":"A Reinforcement Learning based Hybrid GR-DQN Model for Predicting Ichthyophthiriosis Disease in Aquaculture Through Water Quality Analysis","authors":"Bhawna Kol , Khetavath Jairam Naik","doi":"10.1016/j.procs.2025.04.274","DOIUrl":"10.1016/j.procs.2025.04.274","url":null,"abstract":"<div><div>Aquaculture is a fast-growing industry that provides nutritious food to a growing population and generates substantial revenue for countries. The high water quality is required to be maintained for aquatic animal’s survival and health; otherwise, it may cause many diseases like Furunculosis, Bacterial gill disease, and others. Traditionally available methods for water quality analysis are typically difficult to perform due to being time-consuming and lacking accuracy. In this study, a new approach has been developed using an optimal deep reinforcement learning technique, Hybrid Gated Recurrent Unit (GRU) network with Deep Q-Network (DQN), to analyze the state of the water quality of aquaculture by predicting Ichthyophthiriosis (white spot diseases) in an aquaculture environment. The GRU deep learning model with DQN helps in improving the prediction by approximating Q-values and produces a loss function to guide the learning process; rewards are provided due to correct predictions, thereby disease detection corrected accuracy was enhanced. The proposed hybrid GR-DQN model was implemented on the “Pondsdata” dataset and compared the results with the existing model M-DQN. The Hybrid GR-DQN achieved 94.69% accuracy in comparison to the existing model M-DQN’s 84.16% accuracy on the same dataset.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"258 ","pages":"Pages 374-385"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931853","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Comparison of Educational Perspectives on VDI 2221 and Axiomatic Design VDI 2221与公理化设计的教育视角比较

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.073

Patrick Kröpfl , Christian Landschützer , Hannes Hick , Wajih Haider Awan , Christopher A. Brown

Engineering design methods play a crucial role in both academia and industry. These systematic approaches facilitate product and system development, allowing for innovative solutions and refinements. Specifically, this paper will compare two common engineering design methods Axiomatic Design (AD) and VDI 2221 in terms of their application in teaching and their transferability to industry, especially for small and medium-sized enterprises (SMEs). Firstly, a quantitative comparison of the two methods will be conducted. Comparative factors will include scope, accessibility, required prior knowledge, and the availability of tools for each method. Following this, insights from teaching experiences at the Technical University of Graz and Worcester Polytechnic Institute (WPI) will be discussed, focusing on the teachability of the methods. This will provide insights into the effectiveness and suitability of the methods for higher education. The transfer potential of the methods to SMEs will be derived from these. Finally, the findings and improvement potential will be summarized, and possibilities for the knowledge transfer of engineering design methods to SMEs will be formulated.

工程设计方法在学术界和工业界都起着至关重要的作用。这些系统的方法促进了产品和系统的开发，允许创新的解决方案和改进。具体而言，本文将比较两种常见的工程设计方法公理设计（AD）和VDI 2221在教学中的应用和对工业的可转移性，特别是对中小型企业（SMEs）。首先，对两种方法进行定量比较。比较因素将包括范围、可及性、所需的先验知识，以及每种方法的工具可用性。在此之后，将讨论格拉茨技术大学和伍斯特理工学院（WPI）的教学经验，重点是方法的可教性。这将为高等教育方法的有效性和适用性提供见解。这些方法对中小企业的转移潜力将由此产生。最后，总结研究结果和改进潜力，并制定工程设计方法向中小企业知识转移的可能性。

{"title":"A Comparison of Educational Perspectives on VDI 2221 and Axiomatic Design","authors":"Patrick Kröpfl , Christian Landschützer , Hannes Hick , Wajih Haider Awan , Christopher A. Brown","doi":"10.1016/j.procs.2025.01.073","DOIUrl":"10.1016/j.procs.2025.01.073","url":null,"abstract":"<div><div>Engineering design methods play a crucial role in both academia and industry. These systematic approaches facilitate product and system development, allowing for innovative solutions and refinements. Specifically, this paper will compare two common engineering design methods Axiomatic Design (AD) and VDI 2221 in terms of their application in teaching and their transferability to industry, especially for small and medium-sized enterprises (SMEs). Firstly, a quantitative comparison of the two methods will be conducted. Comparative factors will include scope, accessibility, required prior knowledge, and the availability of tools for each method. Following this, insights from teaching experiences at the Technical University of Graz and Worcester Polytechnic Institute (WPI) will be discussed, focusing on the teachability of the methods. This will provide insights into the effectiveness and suitability of the methods for higher education. The transfer potential of the methods to SMEs will be derived from these. Finally, the findings and improvement potential will be summarized, and possibilities for the knowledge transfer of engineering design methods to SMEs will be formulated.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"253 ","pages":"Pages 94-103"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143480244","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

Game-based learning for industrial maintenance: a Unity 3D educational game of compressed air system training 基于游戏的工业维护学习：压缩空气系统培训的Unity 3D教育游戏

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.140

Birkan Işık , Gülbahar Emir Işık , Miroslav Zilka

This research introduces an innovative approach to industrial maintenance training by developing an interactive game using an interactive game developed with the Unity 3D engine and extended reality technologies. The game simulates the compressed air system maintenance, aiming to improve technicians’ practical skills and safety awareness through immersive, realistic scenarios. Leveraging Unity 3D’s advanced graphical and physics capabilities, it creates an engaging environment where participants interact with dynamic modules, enhancing decision-making, problem-solving, and analytical thinking. Gameplay involves guiding participants through the compressed air systems maintenance process with realistic controls that respond dynamically to user inputs, thereby allowing technicians to refine technical skills with a strong emphasis on safety. Performance is evaluated based on safety compliance and technical accuracy, demonstrating the value of game-based learning in technical education. This study highlights the potential of game-based learning within Industry 5.0, promoting lifelong learning and preparing professionals for future industrial challenges.

本研究通过使用Unity 3D引擎和扩展现实技术开发的互动游戏，开发了一种创新的工业维修培训方法。游戏模拟了压缩空气系统的维护，旨在通过沉浸式、逼真的场景来提高技术人员的实用技能和安全意识。利用Unity 3D先进的图形和物理功能，它创建了一个引人入胜的环境，参与者与动态模块交互，增强决策，解决问题和分析思维。游戏玩法包括引导参与者完成压缩空气系统的维护过程，并对用户输入进行动态响应，从而使技术人员能够在强调安全的同时完善技术技能。性能评估基于安全合规和技术准确性，展示了基于游戏的学习在技术教育中的价值。这项研究强调了工业5.0中基于游戏的学习的潜力，促进终身学习，并为未来的工业挑战做好准备。

{"title":"Game-based learning for industrial maintenance: a Unity 3D educational game of compressed air system training","authors":"Birkan Işık , Gülbahar Emir Işık , Miroslav Zilka","doi":"10.1016/j.procs.2025.01.140","DOIUrl":"10.1016/j.procs.2025.01.140","url":null,"abstract":"<div><div>This research introduces an innovative approach to industrial maintenance training by developing an interactive game using an interactive game developed with the Unity 3D engine and extended reality technologies. The game simulates the compressed air system maintenance, aiming to improve technicians’ practical skills and safety awareness through immersive, realistic scenarios. Leveraging Unity 3D’s advanced graphical and physics capabilities, it creates an engaging environment where participants interact with dynamic modules, enhancing decision-making, problem-solving, and analytical thinking. Gameplay involves guiding participants through the compressed air systems maintenance process with realistic controls that respond dynamically to user inputs, thereby allowing technicians to refine technical skills with a strong emphasis on safety. Performance is evaluated based on safety compliance and technical accuracy, demonstrating the value of game-based learning in technical education. This study highlights the potential of game-based learning within Industry 5.0, promoting lifelong learning and preparing professionals for future industrial challenges.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"253 ","pages":"Pages 784-793"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143480648","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

Towards AI-enhanced process planning: assessing machine tool capability based on part design 面向人工智能增强的工艺规划：基于零件设计的机床能力评估

Procedia Computer Science

Pub Date : 2025-01-01 DOI: 10.1016/j.procs.2025.01.122

Sepideh Abolghasem , Matthew Youssef , Faruk Abedrabbo , Amman Pandde

The emergence of the fourth industrial revolution, or Industry 4.0, necessitates a more automated approach to manufacturing process planning. This process begins with evaluating machine tool capabilities to handle specific part geometries and microstructures. Once a match is established, the focus shifts to developing an efficient method for converting design elements into physical components. This work aims to create and validate a framework that assesses the manufacturability of design features based on the available machinery and materials. Specifically, it involves classifying manufacturing processes, such as turning and milling, for a given part design geometry. To achieve this, feature attributes like rotational symmetry and D2 distribution are calculated for a dataset used to train a decision tree. This model then suggests the appropriate manufacturing process for a given CAD model. The decision tree is validated with a separate dataset, showing reasonable accuracy. Ultimately, the goal is to enhance process planning, ensuring the seamless translation of designs into physical products, with a particular emphasis on geometry, microstructure, and cost.

第四次工业革命（即工业 4.0）的出现，要求采用更加自动化的方法进行制造工艺规划。这一过程首先要评估机床处理特定零件几何形状和微观结构的能力。一旦确定了匹配，重点就转移到开发一种高效的方法，将设计元素转换为物理组件。这项工作旨在创建和验证一个框架，根据现有的机械和材料评估设计特征的可制造性。具体来说，它涉及对给定零件设计几何形状的车削和铣削等制造工艺进行分类。为此，需要计算数据集的特征属性，如旋转对称性和 D2 分布，以训练决策树。然后，该模型会为给定的 CAD 模型建议合适的制造工艺。决策树通过一个单独的数据集进行验证，显示出合理的准确性。最终的目标是加强工艺规划，确保将设计无缝转化为物理产品，特别强调几何形状、微观结构和成本。

{"title":"Towards AI-enhanced process planning: assessing machine tool capability based on part design","authors":"Sepideh Abolghasem , Matthew Youssef , Faruk Abedrabbo , Amman Pandde","doi":"10.1016/j.procs.2025.01.122","DOIUrl":"10.1016/j.procs.2025.01.122","url":null,"abstract":"<div><div>The emergence of the fourth industrial revolution, or Industry 4.0, necessitates a more automated approach to manufacturing process planning. This process begins with evaluating machine tool capabilities to handle specific part geometries and microstructures. Once a match is established, the focus shifts to developing an efficient method for converting design elements into physical components. This work aims to create and validate a framework that assesses the manufacturability of design features based on the available machinery and materials. Specifically, it involves classifying manufacturing processes, such as turning and milling, for a given part design geometry. To achieve this, feature attributes like rotational symmetry and D2 distribution are calculated for a dataset used to train a decision tree. This model then suggests the appropriate manufacturing process for a given CAD model. The decision tree is validated with a separate dataset, showing reasonable accuracy. Ultimately, the goal is to enhance process planning, ensuring the seamless translation of designs into physical products, with a particular emphasis on geometry, microstructure, and cost.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"253 ","pages":"Pages 603-611"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143480710","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