Systems and Soft Computing最新文献_第10页

Application of CNN-based financial risk identification and management convolutional neural networks in financial risk 基于cnn的金融风险识别与管理卷积神经网络在金融风险中的应用

Systems and Soft Computing

Pub Date : 2025-03-04 DOI: 10.1016/j.sasc.2025.200215

Zhen Wang

The application of intelligent financial analysis model to the research of enterprise financial risk prediction can improve the adaptability of the model, effectively capture complex patterns and adapt to large-scale data, but there are some problems such as insufficient accuracy and low recall rate. In order to improve the effect of enterprise financial risk management, this research applies convolutional neural network to enterprise financial risk management, and proposes a binary classification prediction model of financial risk dilemma based on one-dimensional convolution and sparse attention mechanism. Then, combined with experimental research, this research verifies that the synergy of multiple modules enables the model proposed in this research to understand and classify the input data more comprehensively and accurately, and then achieve significant improvements in various indicators. Moreover, compared with the comparison model of a single module, it shows superior performance. After training, the accuracy of the model is 75.98 % in the training set and 82.34 % in the test set, which shows the ideal training results, and proves that the model has good generalization ability The model has the best performance in precision, recall and F1, which is due to the comprehensive use of CNN module, LSTM module, encoder module and AR module. After training, the accuracy of the model is 75.98 % in the training set and 82.34 % in the test set, which shows the ideal training results, and proves that the model has good generalization ability. The model has the best performance in precision, recall and F1, which is due to the comprehensive use of CNN module, LSTM module, encoder module and AR module. The experimental results show that the model proposed in this research can realize the accurate classification of binary classification prediction of financial risk dilemma, help enterprises to rationally allocate resources, control the government's unnecessary financial support to enterprises that are on the verge of bankruptcy and have no prospect, and prevent the loss of enterprises' assets.

将智能财务分析模型应用于企业财务风险预测研究，可以提高模型的适应性，有效捕获复杂模式，适应大规模数据，但存在准确率不足、召回率低等问题。为了提高企业财务风险管理的效果，本研究将卷积神经网络应用于企业财务风险管理，提出了一种基于一维卷积和稀疏注意机制的财务风险困境二分类预测模型。然后，结合实验研究，验证了多个模块的协同作用，使本研究提出的模型能够更全面、更准确地理解和分类输入数据，从而在各项指标上取得显著提升。并且，与单模块的比较模型相比，该模型表现出了优越的性能。经过训练，模型在训练集中的准确率为75.98%，在测试集中的准确率为82.34%，显示出理想的训练结果，证明模型具有良好的泛化能力，模型在精度、召回率和F1方面表现最好，这是由于综合使用了CNN模块、LSTM模块、编码器模块和AR模块。经过训练，该模型在训练集中的准确率为75.98%，在测试集中的准确率为82.34%，显示出理想的训练结果，证明该模型具有良好的泛化能力。该模型在准确率、召回率和F1方面表现最好，这是由于综合使用了CNN模块、LSTM模块、编码器模块和AR模块。实验结果表明，本研究提出的模型能够实现财务风险困境二元分类预测的准确分类，帮助企业合理配置资源，控制政府对濒临破产、没有前景的企业进行不必要的资金支持，防止企业资产的流失。

{"title":"Application of CNN-based financial risk identification and management convolutional neural networks in financial risk","authors":"Zhen Wang","doi":"10.1016/j.sasc.2025.200215","DOIUrl":"10.1016/j.sasc.2025.200215","url":null,"abstract":"<div><div>The application of intelligent financial analysis model to the research of enterprise financial risk prediction can improve the adaptability of the model, effectively capture complex patterns and adapt to large-scale data, but there are some problems such as insufficient accuracy and low recall rate. In order to improve the effect of enterprise financial risk management, this research applies convolutional neural network to enterprise financial risk management, and proposes a binary classification prediction model of financial risk dilemma based on one-dimensional convolution and sparse attention mechanism. Then, combined with experimental research, this research verifies that the synergy of multiple modules enables the model proposed in this research to understand and classify the input data more comprehensively and accurately, and then achieve significant improvements in various indicators. Moreover, compared with the comparison model of a single module, it shows superior performance. After training, the accuracy of the model is 75.98 % in the training set and 82.34 % in the test set, which shows the ideal training results, and proves that the model has good generalization ability The model has the best performance in precision, recall and F1, which is due to the comprehensive use of CNN module, LSTM module, encoder module and AR module. After training, the accuracy of the model is 75.98 % in the training set and 82.34 % in the test set, which shows the ideal training results, and proves that the model has good generalization ability. The model has the best performance in precision, recall and F1, which is due to the comprehensive use of CNN module, LSTM module, encoder module and AR module. The experimental results show that the model proposed in this research can realize the accurate classification of binary classification prediction of financial risk dilemma, help enterprises to rationally allocate resources, control the government's unnecessary financial support to enterprises that are on the verge of bankruptcy and have no prospect, and prevent the loss of enterprises' assets.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200215"},"PeriodicalIF":0.0,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143579441","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

Forecasting the Bitcoin price using the various Machine Learning: A systematic review in data-driven marketing 使用各种机器学习预测比特币价格：数据驱动营销的系统回顾

Systems and Soft Computing

Pub Date : 2025-02-25 DOI: 10.1016/j.sasc.2025.200209

Payam Boozary, Sogand Sheykhan, Hamed GhorbanTanhaei

The emergence of Bitcoin as a pioneering cryptocurrency has transformed financial markets, garnering widespread interest from academicians, policymakers, and investors. The market's inherent volatility and the rapid integration of public information into price movements continue to present a formidable challenge in accurately forecasting Bitcoin prices despite its potential. The limitations of conventional financial models, which frequently need to consider the distinctive attributes of cryptocurrencies, further exacerbate this challenge. Despite the proliferation of ML in various fields, existing models have not fully harnessed these techniques, performing only marginally better than random guesses due to the unique challenges posed by the high volatility and complex dynamics of cryptocurrency markets. This study introduces a systematic review of ML methods specifically tailored for Bitcoin price prediction, with a focus on evaluating the robustness, accuracy, and appropriateness of advanced ML techniques like Long Short-Term Memory (LSTM) networks. The novelty lies in its comprehensive assessment of these methods in the context of data-driven marketing, aiming to enhance both academic understanding and practical applications in financial technology. The previous studies haven't Machine Learning (ML) has become a formidable instrument that has the potential to improve the accuracy of forecasting; however, there still needs to be more comprehension regarding the most effective ML models in this field. The study's importance is derived from its systematic examination of various machine learning (ML) techniques employed to predict the price of Bitcoin, with a particular emphasis on their integration into data-driven marketing strategies. The results will substantially contribute to both academic research and practical applications, providing valuable insights that can be used to develop more dependable forecasting tools, thereby benefiting investors, marketers, and policymakers.

比特币作为一种开创性的加密货币的出现改变了金融市场，引起了学者、政策制定者和投资者的广泛兴趣。尽管比特币潜力巨大，但市场固有的波动性和将公共信息快速整合到价格变动中，仍然给准确预测比特币价格带来了巨大挑战。传统金融模型经常需要考虑加密货币的独特属性，其局限性进一步加剧了这一挑战。尽管机器学习在各个领域的扩散，现有的模型并没有完全利用这些技术，由于加密货币市场的高波动性和复杂动态带来的独特挑战，它们的表现只比随机猜测好一点点。本研究对专门为比特币价格预测量身定制的机器学习方法进行了系统回顾，重点是评估长短期记忆（LSTM）网络等高级机器学习技术的鲁棒性、准确性和适当性。新颖之处在于它在数据驱动营销的背景下对这些方法进行了全面评估，旨在加强对金融技术的学术理解和实际应用。机器学习（ML）已经成为一种强大的工具，有可能提高预测的准确性；然而，对于这个领域中最有效的机器学习模型，仍然需要更多的理解。这项研究的重要性在于，它系统地研究了用于预测比特币价格的各种机器学习（ML）技术，特别强调了它们与数据驱动的营销策略的整合。研究结果将极大地促进学术研究和实际应用，提供有价值的见解，可用于开发更可靠的预测工具，从而使投资者、营销人员和决策者受益。

{"title":"Forecasting the Bitcoin price using the various Machine Learning: A systematic review in data-driven marketing","authors":"Payam Boozary, Sogand Sheykhan, Hamed GhorbanTanhaei","doi":"10.1016/j.sasc.2025.200209","DOIUrl":"10.1016/j.sasc.2025.200209","url":null,"abstract":"<div><div>The emergence of Bitcoin as a pioneering cryptocurrency has transformed financial markets, garnering widespread interest from academicians, policymakers, and investors. The market's inherent volatility and the rapid integration of public information into price movements continue to present a formidable challenge in accurately forecasting Bitcoin prices despite its potential. The limitations of conventional financial models, which frequently need to consider the distinctive attributes of cryptocurrencies, further exacerbate this challenge. Despite the proliferation of ML in various fields, existing models have not fully harnessed these techniques, performing only marginally better than random guesses due to the unique challenges posed by the high volatility and complex dynamics of cryptocurrency markets. This study introduces a systematic review of ML methods specifically tailored for Bitcoin price prediction, with a focus on evaluating the robustness, accuracy, and appropriateness of advanced ML techniques like Long Short-Term Memory (LSTM) networks. The novelty lies in its comprehensive assessment of these methods in the context of data-driven marketing, aiming to enhance both academic understanding and practical applications in financial technology. The previous studies haven't Machine Learning (ML) has become a formidable instrument that has the potential to improve the accuracy of forecasting; however, there still needs to be more comprehension regarding the most effective ML models in this field. The study's importance is derived from its systematic examination of various machine learning (ML) techniques employed to predict the price of Bitcoin, with a particular emphasis on their integration into data-driven marketing strategies. The results will substantially contribute to both academic research and practical applications, providing valuable insights that can be used to develop more dependable forecasting tools, thereby benefiting investors, marketers, and policymakers.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200209"},"PeriodicalIF":0.0,"publicationDate":"2025-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143552307","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

Optimizing multilevel image segmentation with a modified new Caledonian crow learning algorithm 基于改进的Caledonian乌鸦学习算法的多级图像分割优化

Systems and Soft Computing

Pub Date : 2025-02-25 DOI: 10.1016/j.sasc.2025.200206

Osama Moh'd Alia

Image segmentation is a fundamental component of image analysis. While numerous algorithms exist for this task, thresholding is one of the most widely used methods. Multilevel thresholding, which involves dividing an image into multiple segments, is computationally intensive due to its need to search for optimal thresholds. This paper presents a solution to this optimization problem by exploring the New Caledonian crow learning algorithm (NCCLA). Inspired by nature, the NCCLA algorithm draws from the behaviors of New Caledonian crows, which use tools from Pandanus trees to obtain food. To improve the algorithm's capacity to discover optimal thresholds while balancing the exploitation and exploration processes, this paper introduces a modification inspired by the pitch adjustment rate portion of the harmony search algorithm. The performance of this modified NCCLA algorithm was evaluated on benchmark images, and a comparative analysis was conducted against other metaheuristic-based algorithms including particle swarm optimization, harmony search, bacterial foraging, and genetic algorithms; the experimental results demonstrate the effectiveness of the proposed algorithm, which was further statistically validated using a t-test.

图像分割是图像分析的一个基本组成部分。虽然存在许多算法用于此任务，阈值分割是最广泛使用的方法之一。多级阈值分割涉及到将图像分割成多个片段，由于需要搜索最优阈值，因此计算量很大。本文通过探索新喀里多尼亚乌鸦学习算法（New Caledonian crow learning algorithm, ncla）来解决这一优化问题。受大自然的启发，ncla算法借鉴了新喀里多尼亚乌鸦的行为，它们使用Pandanus树上的工具来获取食物。为了提高算法在平衡挖掘和探索过程的同时发现最优阈值的能力，本文引入了一种受和声搜索算法中音调调整率部分启发的改进。在基准图像上对改进的ncla算法进行了性能评价，并与粒子群优化、和谐搜索、细菌觅食和遗传算法等其他基于元启发式算法进行了比较分析；实验结果证明了该算法的有效性，并通过t检验进一步进行了统计验证。

{"title":"Optimizing multilevel image segmentation with a modified new Caledonian crow learning algorithm","authors":"Osama Moh'd Alia","doi":"10.1016/j.sasc.2025.200206","DOIUrl":"10.1016/j.sasc.2025.200206","url":null,"abstract":"<div><div>Image segmentation is a fundamental component of image analysis. While numerous algorithms exist for this task, thresholding is one of the most widely used methods. Multilevel thresholding, which involves dividing an image into multiple segments, is computationally intensive due to its need to search for optimal thresholds. This paper presents a solution to this optimization problem by exploring the New Caledonian crow learning algorithm (NCCLA). Inspired by nature, the NCCLA algorithm draws from the behaviors of New Caledonian crows, which use tools from Pandanus trees to obtain food. To improve the algorithm's capacity to discover optimal thresholds while balancing the exploitation and exploration processes, this paper introduces a modification inspired by the pitch adjustment rate portion of the harmony search algorithm. The performance of this modified NCCLA algorithm was evaluated on benchmark images, and a comparative analysis was conducted against other metaheuristic-based algorithms including particle swarm optimization, harmony search, bacterial foraging, and genetic algorithms; the experimental results demonstrate the effectiveness of the proposed algorithm, which was further statistically validated using a <em>t</em>-test.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200206"},"PeriodicalIF":0.0,"publicationDate":"2025-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143552308","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

Design of an intelligent grading system for college English translation based on big data technology 基于大数据技术的大学英语翻译智能评分系统设计

Systems and Soft Computing

Pub Date : 2025-02-19 DOI: 10.1016/j.sasc.2025.200205

Xiaoyan Li , Chengzhou Huang

Background

The Intelligent Grading System (IGS), now in use for college English translation, has the following problems in actual use: the running time is excessive, and the final score result differs significantly from the actual one. Given this scenario, it is essential to substantially increase grading efficiency and result correctness to reduce manual participation in the enhancing grading efficiency.

Objective

This research investigates the use of big data technology in designing an IGS for college English translation. The study focuses on the intersection of literature and English language teaching, aiming to enhance the accuracy and efficiency of grading translation assignments.

Methods

The deep learning methodology is the core approach for developing the intelligent grading system. By leveraging the power of a Hybrid gradient-boosting decision tree with an ensemble Back Propagation Neural Network (HGBDT-EBPNN), the system learns from large volumes of labeled translation data to identify patterns and extract meaningful features that contribute to accurate grading.

Results

The findings of this research contribute to the growing body of knowledge on the use of big data technology and deep learning in the field of translation assessment. The proposed study has provided 98 % of accuracy in the performance metrics.

Conclusion

The IGS offers a promising solution for enhancing the efficiency and objectivity of grading college English translation assignments. It could improve the quality of feedback provided to students as well as streamline the assessment process for instructors.

目前正在使用的大学英语翻译智能评分系统（IGS）在实际使用中存在以下问题：运行时间过长，最终评分结果与实际评分结果相差较大。在这种情况下，有必要大幅度提高评分效率和结果的正确性，以减少人工参与提高评分效率。目的探讨大数据技术在大学英语翻译系统设计中的应用。本研究着眼于文学与英语教学的交叉，旨在提高翻译作业评分的准确性和效率。方法深度学习方法是开发智能评分系统的核心方法。通过利用混合梯度增强决策树和集成反向传播神经网络（HGBDT-EBPNN）的功能，系统从大量标记的翻译数据中学习，以识别模式并提取有助于准确评分的有意义的特征。结果本研究的发现有助于在翻译评估领域使用大数据技术和深度学习的知识体系的发展。所提出的研究在性能指标上提供了98%的准确性。结论IGS为提高大学英语翻译作业评分的效率和客观性提供了一种有前景的解决方案。它可以提高提供给学生的反馈的质量，并简化教师的评估过程。

{"title":"Design of an intelligent grading system for college English translation based on big data technology","authors":"Xiaoyan Li , Chengzhou Huang","doi":"10.1016/j.sasc.2025.200205","DOIUrl":"10.1016/j.sasc.2025.200205","url":null,"abstract":"<div><h3>Background</h3><div>The Intelligent Grading System (IGS), now in use for college English translation, has the following problems in actual use: the running time is excessive, and the final score result differs significantly from the actual one. Given this scenario, it is essential to substantially increase grading efficiency and result correctness to reduce manual participation in the enhancing grading efficiency.</div></div><div><h3>Objective</h3><div>This research investigates the use of big data technology in designing an IGS for college English translation. The study focuses on the intersection of literature and English language teaching, aiming to enhance the accuracy and efficiency of grading translation assignments.</div></div><div><h3>Methods</h3><div>The deep learning methodology is the core approach for developing the intelligent grading system. By leveraging the power of a Hybrid gradient-boosting decision tree with an ensemble Back Propagation Neural Network (HGBDT-EBPNN), the system learns from large volumes of labeled translation data to identify patterns and extract meaningful features that contribute to accurate grading.</div></div><div><h3>Results</h3><div>The findings of this research contribute to the growing body of knowledge on the use of big data technology and deep learning in the field of translation assessment. The proposed study has provided 98 % of accuracy in the performance metrics.</div></div><div><h3>Conclusion</h3><div>The IGS offers a promising solution for enhancing the efficiency and objectivity of grading college English translation assignments. It could improve the quality of feedback provided to students as well as streamline the assessment process for instructors.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200205"},"PeriodicalIF":0.0,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143578332","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

Application of interactive AI system based on image recognition in rural landscape design 基于图像识别的交互式AI系统在乡村景观设计中的应用

Systems and Soft Computing

Pub Date : 2025-02-19 DOI: 10.1016/j.sasc.2025.200204

Da He

Rural landscape design plays a critical role in improving the quality of rural environment and residents' quality of life. However, the relevant material library is not sufficient and some original styles need to be preserved, making the design work difficult. The research aims to construct an intelligent and interactive rural landscape design system to improve design efficiency and optimize design outcomes. The study uses attention generative adversarial networks to enrich elements in rural landscape design, improving the problem of inaccurate mapping. In addition, a stable diffusion model is introduced to optimize the quality of its landscape mapping. The findings denoted that the intelligent landscape design system designed in the study had an 8-fold reduction in drawing time compared to designer drawings. The generated rural landscape plan was evaluated in detail, including image quality, element diversity, similarity, and referenceability, with a total score of 99.3 points. The overall evaluation score for the overall image performance was 93.9 points, both of which were superior to other intelligent landscape design systems. From this, it can be seen that the research system not only has efficient design efficiency but also meets the requirements for the image quality of the landscape plan drawn and adaptability to the environment. The study proposes a new landscape design tool that contributes to the environmental improvement of rural landscapes.

乡村景观设计对于改善乡村环境质量、提高居民生活质量有着至关重要的作用。但是，相关的资料库并不充足，一些原有的风格需要保留，给设计工作带来了困难。本研究旨在构建一个智能互动的乡村景观设计系统，以提高设计效率，优化设计成果。本研究利用注意力生成对抗网络来丰富乡村景观设计的元素，改善地图绘制不准确的问题。此外，还引入了一个稳定的扩散模型来优化其景观映射的质量。研究结果表明，研究中设计的智能景观设计系统与设计师绘制的图纸相比，绘制时间减少了8倍。对生成的乡村景观平面图进行详细评价，包括图像质量、元素多样性、相似性和可参考性，总分为99.3分。整体形象表现的综合评价得分为93.9分，均优于其他智能景观设计系统。由此可见，研究系统不仅具有高效的设计效率，而且满足了所绘制的景观平面图图像质量和对环境适应性的要求。该研究提出了一种新的景观设计工具，有助于改善乡村景观的环境。

{"title":"Application of interactive AI system based on image recognition in rural landscape design","authors":"Da He","doi":"10.1016/j.sasc.2025.200204","DOIUrl":"10.1016/j.sasc.2025.200204","url":null,"abstract":"<div><div>Rural landscape design plays a critical role in improving the quality of rural environment and residents' quality of life. However, the relevant material library is not sufficient and some original styles need to be preserved, making the design work difficult. The research aims to construct an intelligent and interactive rural landscape design system to improve design efficiency and optimize design outcomes. The study uses attention generative adversarial networks to enrich elements in rural landscape design, improving the problem of inaccurate mapping. In addition, a stable diffusion model is introduced to optimize the quality of its landscape mapping. The findings denoted that the intelligent landscape design system designed in the study had an 8-fold reduction in drawing time compared to designer drawings. The generated rural landscape plan was evaluated in detail, including image quality, element diversity, similarity, and referenceability, with a total score of 99.3 points. The overall evaluation score for the overall image performance was 93.9 points, both of which were superior to other intelligent landscape design systems. From this, it can be seen that the research system not only has efficient design efficiency but also meets the requirements for the image quality of the landscape plan drawn and adaptability to the environment. The study proposes a new landscape design tool that contributes to the environmental improvement of rural landscapes.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200204"},"PeriodicalIF":0.0,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143508691","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

Neural networks optimization via Gauss–Newton based QR factorization on SARS-CoV-2 variant classification 基于高斯牛顿QR分解的神经网络优化SARS-CoV-2变异分类

Systems and Soft Computing

Pub Date : 2025-02-14 DOI: 10.1016/j.sasc.2025.200195

Mohammad Jamhuri , Mohammad Isa Irawan , Imam Mukhlash , Mohammad Iqbal , Ni Nyoman Tri Puspaningsih

Studies on the COVID-19 pandemic continue due to the potential mutation creating new variants. One response to be aware of the situation is by classifying SARS-CoV-2 variants. Neural networks (NNs)-based classifiers showed good accuracies but are known very costly in the learning process. Second-order optimization approaches are alternatives for NNs to work faster instead of the first-order ones. Still, it needs a huge memory usage. Therefore, we propose a new second-order optimization method for NNs, called QR-GN, to efficiently classify SARS-CoV-2 variants. The proposed method is derived from NNs and Gauss–Newton with QR factorization. The goal of this study is to classify SARS-CoV-2 variants given their spike protein sequences efficiently with high accuracy. In this study, the proposed method was demonstrated on a public dataset for the protein SARS-CoV-2. In the demonstrations, the proposed method outperformed other optimization methods in terms of memory usage and run time. Moreover, the proposed method can significantly elevate the accuracy classification for various NNs, such as: single layer perceptron, multilayer perceptron, and convolutional neural networks.

由于可能产生新的变异，对COVID-19大流行的研究仍在继续。了解这种情况的一种应对措施是对SARS-CoV-2变体进行分类。基于神经网络的分类器具有良好的准确率，但在学习过程中代价高昂。二阶优化方法是神经网络的替代方案，可以比一阶优化方法更快地工作。但是，它需要大量的内存使用。因此，我们提出了一种新的神经网络二阶优化方法，称为QR-GN，以有效地分类SARS-CoV-2变体。该方法是基于神经网络和高斯-牛顿的QR分解。本研究的目的是根据其刺突蛋白序列高效、高精度地对SARS-CoV-2变体进行分类。在这项研究中，提出的方法在SARS-CoV-2蛋白的公共数据集上得到了验证。在演示中，所提出的方法在内存使用和运行时间方面优于其他优化方法。此外，该方法可以显著提高各种神经网络的分类精度，如单层感知器、多层感知器和卷积神经网络。

{"title":"Neural networks optimization via Gauss–Newton based QR factorization on SARS-CoV-2 variant classification","authors":"Mohammad Jamhuri , Mohammad Isa Irawan , Imam Mukhlash , Mohammad Iqbal , Ni Nyoman Tri Puspaningsih","doi":"10.1016/j.sasc.2025.200195","DOIUrl":"10.1016/j.sasc.2025.200195","url":null,"abstract":"<div><div>Studies on the COVID-19 pandemic continue due to the potential mutation creating new variants. One response to be aware of the situation is by classifying SARS-CoV-2 variants. Neural networks (NNs)-based classifiers showed good accuracies but are known very costly in the learning process. Second-order optimization approaches are alternatives for NNs to work faster instead of the first-order ones. Still, it needs a huge memory usage. Therefore, we propose a new second-order optimization method for NNs, called <em>QR-GN</em>, to efficiently classify SARS-CoV-2 variants. The proposed method is derived from NNs and Gauss–Newton with QR factorization. The goal of this study is to classify SARS-CoV-2 variants given their spike protein sequences efficiently with high accuracy. In this study, the proposed method was demonstrated on a public dataset for the protein SARS-CoV-2. In the demonstrations, the proposed method outperformed other optimization methods in terms of memory usage and run time. Moreover, the proposed method can significantly elevate the accuracy classification for various NNs, such as: single layer perceptron, multilayer perceptron, and convolutional neural networks.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200195"},"PeriodicalIF":0.0,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143463876","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

Research and application of optimization of physical education training model based on multi-objective differential evolutionary algorithm 基于多目标差分进化算法的体育训练模型优化研究与应用

Systems and Soft Computing

Pub Date : 2025-02-13 DOI: 10.1016/j.sasc.2025.200200

Man Wu

With the development of computer science, various algorithm models are gradually applied in various fields of life. In order to study the application of the multi-objective differential evolution algorithm in the field of sports transportation. Based on the improvement of multi-objective differential evolution algorithm, this paper proposes the training model of PE education, and compares the prediction results and the actual results. The specific conclusions are as follows: (1) MODE algorithm is better to other algorithms in convergence speed and accuracy; MODE algorithm can not only reach the optimal particle position quickly, but also fluctuate around the best point.(2) AMODE-MPS has great potential for dealing with complex and multiple objectives.(3) There are significant differences between the prediction performance of the proposed algorithm model and the statistical performance, in which the statistical performance is significantly higher than the predicted performance.(4) The proposed model can basically meet the prediction requirements. Although there are some differences between the prediction results and the actual results, this is because the statistical process is affected by the weather, physical condition and other factors. The results show that the PE training model has good results in practice, so this paper can provide reference for the improvement of PE teaching model.

随着计算机科学的发展，各种算法模型逐渐应用于生活的各个领域。为了研究多目标差分进化算法在体育交通领域的应用。本文在改进多目标差分进化算法的基础上，提出了体育教育人才培养模型，并将预测结果与实际结果进行了比较。具体结论如下：(1)MODE算法在收敛速度和精度上优于其他算法；MODE算法不仅能快速到达最优粒子位置，而且能在最佳点附近波动。(2)MODE- mps在处理复杂多目标方面具有很大的潜力。(3)本文算法模型的预测性能与统计性能存在显著差异，其中统计性能显著高于预测性能。(4)本文模型基本能满足预测要求。虽然预测结果与实际结果存在一定差异，但这是因为统计过程受到天气、身体状况等因素的影响。结果表明，该体育教学模式在实践中取得了良好的效果，可以为体育教学模式的改进提供参考。

{"title":"Research and application of optimization of physical education training model based on multi-objective differential evolutionary algorithm","authors":"Man Wu","doi":"10.1016/j.sasc.2025.200200","DOIUrl":"10.1016/j.sasc.2025.200200","url":null,"abstract":"<div><div>With the development of computer science, various algorithm models are gradually applied in various fields of life. In order to study the application of the multi-objective differential evolution algorithm in the field of sports transportation. Based on the improvement of multi-objective differential evolution algorithm, this paper proposes the training model of PE education, and compares the prediction results and the actual results. The specific conclusions are as follows: (1) MODE algorithm is better to other algorithms in convergence speed and accuracy; MODE algorithm can not only reach the optimal particle position quickly, but also fluctuate around the best point.(2) AMODE-MPS has great potential for dealing with complex and multiple objectives.(3) There are significant differences between the prediction performance of the proposed algorithm model and the statistical performance, in which the statistical performance is significantly higher than the predicted performance.(4) The proposed model can basically meet the prediction requirements. Although there are some differences between the prediction results and the actual results, this is because the statistical process is affected by the weather, physical condition and other factors. The results show that the PE training model has good results in practice, so this paper can provide reference for the improvement of PE teaching model.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200200"},"PeriodicalIF":0.0,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143528733","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

Controllability results for multi-order impulsive neutral fuzzy functional integro-differential equations with finite delay 有限时滞多阶脉冲中立型模糊泛函积分微分方程的可控性结果

Systems and Soft Computing

Pub Date : 2025-02-10 DOI: 10.1016/j.sasc.2025.200202

T. Gunasekar , J. Thiravidarani , P. Raghavendran , B.N. Hanumagowda , Jagadish V. Tawade , Farrukh Yuldashev , Manish Gupta , M. Ijaz Khan

This manuscript focuses on examining the controllability of fuzzy mild solutions for nonlocal impulsive neutral functional integro-differential equations of the first and second order, including systems with finite delay. Furthermore, it explores the characteristics of fuzzy set-valued mappings over real variables, emphasizing important features such upper semi-continuity, convexity, normalcy, and compact support. The key conclusions are obtained by applying the Banach fixed-point theorem. The study makes extensive use of fundamental ideas from functional analysis, fuzzy set theory, and the Hausdorff metric. To demonstrate the practical application of the proposed method, a detailed example is provided.

本文主要研究一、二阶非局部脉冲中立型泛函积分微分方程模糊温和解的可控性，包括有限时滞系统。进一步探讨了实变量上模糊集值映射的特征，强调了上半连续性、凸性、正态性和紧支持等重要特征。应用Banach不动点定理得到了关键结论。该研究广泛使用了泛函分析、模糊集理论和豪斯多夫度量的基本思想。为了说明该方法的实际应用，给出了一个详细的算例。

引用次数: 0

Real-time data-driven estimation of production for point bottom sealing and cutting machines using machine learning 使用机器学习对点底封切机的生产进行实时数据驱动估计

Systems and Soft Computing

Pub Date : 2025-02-08 DOI: 10.1016/j.sasc.2025.200194

Subha R , Diana F.R.I. M , Selvadass M

Demand for sophisticated, data-driven techniques to improve production efficiency has increased due to the expansion of the packaging sector, especially in the packaging of polypropylene (PP) flexible materials. PP-based materials offer a variety of packaging applications due to their robust and versatile qualities, but they also require careful production planning to maximize time and resources. By examining important factors that affect output rates and manufacturing costs, such as material dimensions, thickness, and machine cutting speed, this study investigates how predictive modeling may transform production forecasting. This study attempts to build reliable models with optimized hyperpameters to forecast production yield by integrating simple Machine Learning (ML) approaches, such as Support Vector Regression (SVR), Artificial Neural Networks (ANN), Gaussian Process Regression (GPR), and ensemble based approaches such as Random Forest Regression (RFR), Gradient Boosting Regression (GBR), AdaBoost Regression (ABR), Bagging Regression (BR) and Extra Trees Regression (ETR). A performance based comparison of these models, revealed that the ensemble based models using GBR and BR outperformed the others. Further, the prediction performance was improvised by incorporating them as base models and training a voting regressor model. The usefulness of the prediction model has been further demonstrated in creation of reference charts for effective estimation of cost and runtime.

由于包装部门的扩张，特别是聚丙烯（PP）柔性材料的包装，对复杂的、数据驱动的技术的需求增加了，以提高生产效率。pp基材料由于其坚固和通用的品质而提供了各种包装应用，但它们也需要仔细的生产计划，以最大限度地利用时间和资源。通过检查影响产量和制造成本的重要因素，如材料尺寸、厚度和机器切割速度，本研究探讨了预测建模如何改变生产预测。本研究试图通过整合简单的机器学习（ML）方法，如支持向量回归（SVR）、人工神经网络（ANN）、高斯过程回归（GPR），以及基于集成的方法，如随机森林回归（RFR）、梯度增强回归（GBR）、AdaBoost回归（ABR）、Bagging回归（BR）和额外树回归（ETR），建立具有优化超参数的可靠模型来预测产量。对这些模型进行性能比较，发现使用GBR和BR的基于集成的模型优于其他模型。此外，通过将它们作为基本模型并训练投票回归模型来临时提高预测性能。预测模型的有用性在创建有效估计成本和运行时间的参考图表中得到了进一步证明。

{"title":"Real-time data-driven estimation of production for point bottom sealing and cutting machines using machine learning","authors":"Subha R , Diana F.R.I. M , Selvadass M","doi":"10.1016/j.sasc.2025.200194","DOIUrl":"10.1016/j.sasc.2025.200194","url":null,"abstract":"<div><div>Demand for sophisticated, data-driven techniques to improve production efficiency has increased due to the expansion of the packaging sector, especially in the packaging of polypropylene (PP) flexible materials. PP-based materials offer a variety of packaging applications due to their robust and versatile qualities, but they also require careful production planning to maximize time and resources. By examining important factors that affect output rates and manufacturing costs, such as material dimensions, thickness, and machine cutting speed, this study investigates how predictive modeling may transform production forecasting. This study attempts to build reliable models with optimized hyperpameters to forecast production yield by integrating simple Machine Learning (ML) approaches, such as Support Vector Regression (SVR), Artificial Neural Networks (ANN), Gaussian Process Regression (GPR), and ensemble based approaches such as Random Forest Regression (RFR), Gradient Boosting Regression (GBR), AdaBoost Regression (ABR), Bagging Regression (BR) and Extra Trees Regression (ETR). A performance based comparison of these models, revealed that the ensemble based models using GBR and BR outperformed the others. Further, the prediction performance was improvised by incorporating them as base models and training a voting regressor model. The usefulness of the prediction model has been further demonstrated in creation of reference charts for effective estimation of cost and runtime.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200194"},"PeriodicalIF":0.0,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143402858","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

An Enhancement of Bilateral Closed-Loop Functions in Stability Standards for Networked Control Systems with Transmission Delay 具有传输延迟的网络控制系统稳定性标准中双边闭环函数的增强

Systems and Soft Computing

Pub Date : 2025-02-07 DOI: 10.1016/j.sasc.2025.200201

Shenping Xiao , Jiajun Peng , Honghai Lian , Jianglin Huang

This research looks into a new method for analyzing systems of sampled data that have time delay and stability. The system's enhanced stability condition is suggested by constructing a new Lyapunov-Krasovskii (L-K) function containing both continuous and loop functions with full consideration of the information of the transmission interval and the transmission time delay. This new L-K function better captures the dynamics of the system, leading to more reasonable and effective stability analysis results. The sample control system's stability conditions are written as linear matrix inequalities using the integral inequality method and Lyapunov stability theory. Finally, by building matlab simulation experiments, the proposed Theorem 1 and Theorem 2 allow the maximum sampling bound to be significantly improved in two different numerical examples, and the proposed results are more effective and superior when compared with the existing literature.

本研究探索了一种分析具有时滞和稳定性的采样数据系统的新方法。在充分考虑传输间隔和传输时延信息的情况下，构造了一个包含连续函数和环路函数的新的Lyapunov-Krasovskii （L-K）函数，提出了系统增强稳定的条件。新的L-K函数更好地捕捉了系统的动态，从而得到更合理有效的稳定性分析结果。利用积分不等式法和李雅普诺夫稳定性理论，将样本控制系统的稳定性条件写成线性矩阵不等式。最后，通过搭建matlab仿真实验，所提出的定理1和定理2在两个不同的数值算例中都能显著提高最大采样界，与已有文献相比，所提出的结果更加有效和优越。

引用次数: 0