Pub Date : 2024-07-25DOI: 10.54254/2755-2721/79/20241661
Shunqi Qin
With the continuous development of science and technology, automobile autonomous driving technology has gradually become a research hotspot. Among them, path planning and optimization technology is the key link to realizing automatic driving. This paper aims to discuss the path planning and optimization technology in automotive autonomous driving, analyze its current situation and development trend, and verify its effect through experiments. The role of path planning in our lives is very much needed and very important. Excellent path planning and optimization techniques can effectively improve the autonomous driving performance of vehicles. Reduce traffic accidents and improve safety and comfort. Through the ongoing route optimization research, autonomous driving technology will also be widely promoted and applied.
{"title":"A review of path planning algorithms in automobile autonomous driving","authors":"Shunqi Qin","doi":"10.54254/2755-2721/79/20241661","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241661","url":null,"abstract":"With the continuous development of science and technology, automobile autonomous driving technology has gradually become a research hotspot. Among them, path planning and optimization technology is the key link to realizing automatic driving. This paper aims to discuss the path planning and optimization technology in automotive autonomous driving, analyze its current situation and development trend, and verify its effect through experiments. The role of path planning in our lives is very much needed and very important. Excellent path planning and optimization techniques can effectively improve the autonomous driving performance of vehicles. Reduce traffic accidents and improve safety and comfort. Through the ongoing route optimization research, autonomous driving technology will also be widely promoted and applied.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"51 17","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805171","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/79/20241576
Jingxuan Wu, Jiahao Yang, Guanlin Peng
With the development of computers, machine learning continues to be widely used in various fields. And there are many application scenarios in the field of medicine. Among these, the broadest one is the field of medical image analysis. Medical image has the characteristics of huge data, excessive noise, and recognition difficulty. And the most difficult one is the analysis of lung medical images. Lung cancer has a higher incidence rate and mortality rate than other cancers. According to the National Cancer Center, about 127,070 people died from lung cancer in 2023, making it the highest death rate in the United States. Therefore, early detection of malignant pulmonary nodules has become crucial in the field of medical imaging. The medical imaging's inadequacies are most noticeable in the pictures of malignant pulmonary nodules, which are difficult for a doctor to identify with their naked eyes. However, pre-processing, segmentation difficulties, and poor fitting impact are the drawbacks of classical machine learning. As a result, we must create fresh approaches to these issues.
{"title":"The application progress of Convolutional Neural Networks (CNN) in lung nodule diagnosis","authors":"Jingxuan Wu, Jiahao Yang, Guanlin Peng","doi":"10.54254/2755-2721/79/20241576","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241576","url":null,"abstract":"With the development of computers, machine learning continues to be widely used in various fields. And there are many application scenarios in the field of medicine. Among these, the broadest one is the field of medical image analysis. Medical image has the characteristics of huge data, excessive noise, and recognition difficulty. And the most difficult one is the analysis of lung medical images. Lung cancer has a higher incidence rate and mortality rate than other cancers. According to the National Cancer Center, about 127,070 people died from lung cancer in 2023, making it the highest death rate in the United States. Therefore, early detection of malignant pulmonary nodules has become crucial in the field of medical imaging. The medical imaging's inadequacies are most noticeable in the pictures of malignant pulmonary nodules, which are difficult for a doctor to identify with their naked eyes. However, pre-processing, segmentation difficulties, and poor fitting impact are the drawbacks of classical machine learning. As a result, we must create fresh approaches to these issues.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"30 8","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805413","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/55/20241444
Qian Yang
The k-Nearest Neighbors (k-NN) algorithm is a fundamental and widely-used classification technique that has found applications in various domains, including text classification. In this paper, we present a comprehensive analysis of the k-NN classification algorithm applied to the Reuters news article dataset. Our study includes the data, implementation k-NN classification with different parameters, performance evaluation, and statistical analysis to draw meaningful conclusions. In a comprehensive analysis of the k-NN classification algorithm used for the Reuters news article data-set. A variety of metrics is used to evaluate the performance of the k-NN algorithm, such as accuracy, precision, recall, and F1 scores. These metrics provide a comprehensive view of how well the algorithm classifies news articles. Our statistical analysis reveals significant performance differences between various k-NN configurations. This can help researchers and practitioners make informed decisions when choosing the best parameters for their specific text classification tasks. In conclusion, our study provides valuable insights into the application of k-NN classification algorithms to textual data, highlighting the importance of parameter tuning and rigorous evaluation. These findings can guide practitioners to effectively use k-NN for text classification tasks and inspire further research in the field.
{"title":"Performance analysis of k-Nearest Neighbors classification on Reuters news article datasets","authors":"Qian Yang","doi":"10.54254/2755-2721/55/20241444","DOIUrl":"https://doi.org/10.54254/2755-2721/55/20241444","url":null,"abstract":"The k-Nearest Neighbors (k-NN) algorithm is a fundamental and widely-used classification technique that has found applications in various domains, including text classification. In this paper, we present a comprehensive analysis of the k-NN classification algorithm applied to the Reuters news article dataset. Our study includes the data, implementation k-NN classification with different parameters, performance evaluation, and statistical analysis to draw meaningful conclusions. In a comprehensive analysis of the k-NN classification algorithm used for the Reuters news article data-set. A variety of metrics is used to evaluate the performance of the k-NN algorithm, such as accuracy, precision, recall, and F1 scores. These metrics provide a comprehensive view of how well the algorithm classifies news articles. Our statistical analysis reveals significant performance differences between various k-NN configurations. This can help researchers and practitioners make informed decisions when choosing the best parameters for their specific text classification tasks. In conclusion, our study provides valuable insights into the application of k-NN classification algorithms to textual data, highlighting the importance of parameter tuning and rigorous evaluation. These findings can guide practitioners to effectively use k-NN for text classification tasks and inspire further research in the field.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"11 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141803184","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/79/20241639
Ziyan Zhao, Zixiao Xie, Jiaze Fu, Xintao Tian
This paper explores music genre classification, aiming to enhance existing methodologies. As a crucial aspect of music information retrieval, genre classification facilitates organization and recommendation in music databases and streaming services. Our research, inspired by a Kaggle project, examines the background of music genre classification and introduces improvements. The study focuses on data preparation techniques and a novel methodology using Support Vector Machines (SVM). Utilizing the GTZAN dataset, we applied data segmentation and feature extraction, employing machine learning algorithms like Logistic Regression, Random Forest, and SVM. A significant innovation is our segmentation technique based on music's beats per minute (BPM), designed to preserve rhythmic structure, believed to be essential for accurate classification. We explored various feature extraction methods to boost classifier performance. Experimental results showed the 3-second segmented dataset performed better with SVM's linear kernel. Additionally, a 4-beat segmentation experiment suggested that finer segmentation captures richer audio features, potentially improving classification accuracy. The paper concludes with findings and future research directions, including dataset expansion, advanced segmentation based on musical theory, deep learning applications, and developing real-time classification systems.
{"title":"Music genre classification: Machine Learning on GTZAN","authors":"Ziyan Zhao, Zixiao Xie, Jiaze Fu, Xintao Tian","doi":"10.54254/2755-2721/79/20241639","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241639","url":null,"abstract":"This paper explores music genre classification, aiming to enhance existing methodologies. As a crucial aspect of music information retrieval, genre classification facilitates organization and recommendation in music databases and streaming services. Our research, inspired by a Kaggle project, examines the background of music genre classification and introduces improvements. The study focuses on data preparation techniques and a novel methodology using Support Vector Machines (SVM). Utilizing the GTZAN dataset, we applied data segmentation and feature extraction, employing machine learning algorithms like Logistic Regression, Random Forest, and SVM. A significant innovation is our segmentation technique based on music's beats per minute (BPM), designed to preserve rhythmic structure, believed to be essential for accurate classification. We explored various feature extraction methods to boost classifier performance. Experimental results showed the 3-second segmented dataset performed better with SVM's linear kernel. Additionally, a 4-beat segmentation experiment suggested that finer segmentation captures richer audio features, potentially improving classification accuracy. The paper concludes with findings and future research directions, including dataset expansion, advanced segmentation based on musical theory, deep learning applications, and developing real-time classification systems.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"24 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141803322","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/79/20241622
Jiyue Zhao, Tony Yuxiang Pan, Weibo Yao, Hongwei Lu, Zihan Liu
Various classification algorithms applied to sophisticated datasets have seen significant development over the years, which involves dealing with the growing complexities of real-world data and providing efficient solutions for numerous domains like healthcare and data analysis. There is a critical need to identify the most effective algorithms to deliver high precision and generalizability. This study intends to assess diverse models, including Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), DTs (DT), and Random Forests (RF), on Modified National Institute of Standards and Technology (MNIST) and Wisconsin Diagnostic Breast Cancer (WDBC) datasets, utilizing metrics like Overall Accuracy (OA), Average Accuracy (AA), and Cohens kappa. The study has shown that the performance of the algorithms is mainly determined by the dataset's features. Additionally, insights into the strengths and limitations of each model are provided.
{"title":"Analysis of classification algorithms: Insights from MNIST and WDBC datasets","authors":"Jiyue Zhao, Tony Yuxiang Pan, Weibo Yao, Hongwei Lu, Zihan Liu","doi":"10.54254/2755-2721/79/20241622","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241622","url":null,"abstract":"Various classification algorithms applied to sophisticated datasets have seen significant development over the years, which involves dealing with the growing complexities of real-world data and providing efficient solutions for numerous domains like healthcare and data analysis. There is a critical need to identify the most effective algorithms to deliver high precision and generalizability. This study intends to assess diverse models, including Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), DTs (DT), and Random Forests (RF), on Modified National Institute of Standards and Technology (MNIST) and Wisconsin Diagnostic Breast Cancer (WDBC) datasets, utilizing metrics like Overall Accuracy (OA), Average Accuracy (AA), and Cohens kappa. The study has shown that the performance of the algorithms is mainly determined by the dataset's features. Additionally, insights into the strengths and limitations of each model are provided.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"37 12","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805318","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/79/20241055
Bojun Wang, Jingyi Li, Tianqi Wu, Changjia Qu
With the acceleration of industry reform, new communication technologies are required to have larger bandwidth, faster transmission speed and more comprehensive applications. As a mature and effective high-frequency technology, RF band technology can meet many communication requirements in practical applications and plays an irreplaceable role in current production and life. At the same time, RF technology is also an important part of modern communication system, which has a good development prospect and has been widely used in many fields. In order to let more people understand the importance of radio frequency technology and promote the further development of radio frequency technology, this paper will introduce the important applications of radio frequency technology in practice from four aspects: the application of radio frequency in satellite, radio frequency identification, radio frequency ablation technology and radio frequency integrated circuit, and treat the application of radio frequency technology in practice from different angles, and illustrate the significance and development prospects of radio frequency technology.
{"title":"A review of the development and application of RF technology and its sub-technologies","authors":"Bojun Wang, Jingyi Li, Tianqi Wu, Changjia Qu","doi":"10.54254/2755-2721/79/20241055","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241055","url":null,"abstract":"With the acceleration of industry reform, new communication technologies are required to have larger bandwidth, faster transmission speed and more comprehensive applications. As a mature and effective high-frequency technology, RF band technology can meet many communication requirements in practical applications and plays an irreplaceable role in current production and life. At the same time, RF technology is also an important part of modern communication system, which has a good development prospect and has been widely used in many fields. In order to let more people understand the importance of radio frequency technology and promote the further development of radio frequency technology, this paper will introduce the important applications of radio frequency technology in practice from four aspects: the application of radio frequency in satellite, radio frequency identification, radio frequency ablation technology and radio frequency integrated circuit, and treat the application of radio frequency technology in practice from different angles, and illustrate the significance and development prospects of radio frequency technology.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"47 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805691","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/55/20241535
Yihang Chen, Shuoyu Chen, Yicheng Yang, Siming Lu
This paper presents an in-depth exploration of the Adaboost algorithm in the context of machine learning, focusing on its application in classification tasks. Adaboost, known for its adaptive boosting approach, is examined for its ability to enhance weak learners, particularly decision tree classifiers. The study delves into the theoretical underpinnings of Adaboost, emphasizing its iterative process for minimizing the exponential loss function. The role of decision trees, as integral components of this algorithm, is analyzed in detail. These trees, with their hierarchical query structure, are pivotal in categorizing items based on relevant features. The paper further compares Adaboost with random forests, another prominent machine learning algorithm, highlighting the nuances in their methodologies and applications. Significantly, the research introduces improved methods for selecting and fine-tuning these algorithms to optimize performance in various data classification scenarios. Practical applications of Adaboost and decision trees in real-world data classification tasks are demonstrated, providing insights into their operational effectiveness. This study not only elucidates the strengths of these machine learning techniques but also offers a comparative analysis, guiding practitioners in choosing the most suitable algorithm for specific classification challenges. The findings contribute to the broader understanding of machine learning algorithms, particularly in the context of data classification, and propose innovative approaches for enhancing algorithmic efficiency and accuracy. This research serves as a valuable resource for both academic and practical applications in the field of machine learning.
{"title":"Comparison of decision tree and ensemble algorithms","authors":"Yihang Chen, Shuoyu Chen, Yicheng Yang, Siming Lu","doi":"10.54254/2755-2721/55/20241535","DOIUrl":"https://doi.org/10.54254/2755-2721/55/20241535","url":null,"abstract":"This paper presents an in-depth exploration of the Adaboost algorithm in the context of machine learning, focusing on its application in classification tasks. Adaboost, known for its adaptive boosting approach, is examined for its ability to enhance weak learners, particularly decision tree classifiers. The study delves into the theoretical underpinnings of Adaboost, emphasizing its iterative process for minimizing the exponential loss function. The role of decision trees, as integral components of this algorithm, is analyzed in detail. These trees, with their hierarchical query structure, are pivotal in categorizing items based on relevant features. The paper further compares Adaboost with random forests, another prominent machine learning algorithm, highlighting the nuances in their methodologies and applications. Significantly, the research introduces improved methods for selecting and fine-tuning these algorithms to optimize performance in various data classification scenarios. Practical applications of Adaboost and decision trees in real-world data classification tasks are demonstrated, providing insights into their operational effectiveness. This study not only elucidates the strengths of these machine learning techniques but also offers a comparative analysis, guiding practitioners in choosing the most suitable algorithm for specific classification challenges. The findings contribute to the broader understanding of machine learning algorithms, particularly in the context of data classification, and propose innovative approaches for enhancing algorithmic efficiency and accuracy. This research serves as a valuable resource for both academic and practical applications in the field of machine learning.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"7 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141804224","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/79/20241669
Shilin Zhang
This study addresses the critical issue of optimizing outpatient services in high-capacity hospitals, focusing on developing cost-effective management strategies. Utilizing a simulated model of outpatient services, this research incorporates real data from the National Health Service (NHS) to tackle practical challenges in hospital management. The methodology encompasses the application of backward induction, Q-learning, and Deep Q-Network (DQN) algorithms to formulate solutions. The findings indicate that backward induction effectively resolves simpler scenarios within the assumed conditions. In contrast, Q-learning offers a viable approach, with DQN demonstrating superior performance in addressing more complex, realistic problems. The conclusion drawn from this study is that each algorithm exhibits unique strengths in its respective operational environment. While direct comparison between the models based on output analysis is not feasible due to the variation in environmental settings, it is evident that all three algorithms significantly contribute to resolving the targeted issues in outpatient service management. This research not only provides valuable insights into hospital outpatient service optimization but also opens avenues for further exploration in the application of advanced computational techniques in healthcare management.
{"title":"Optimizing hospital outpatient services: A comparative study of backward induction and Q-learning techniques","authors":"Shilin Zhang","doi":"10.54254/2755-2721/79/20241669","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241669","url":null,"abstract":"This study addresses the critical issue of optimizing outpatient services in high-capacity hospitals, focusing on developing cost-effective management strategies. Utilizing a simulated model of outpatient services, this research incorporates real data from the National Health Service (NHS) to tackle practical challenges in hospital management. The methodology encompasses the application of backward induction, Q-learning, and Deep Q-Network (DQN) algorithms to formulate solutions. The findings indicate that backward induction effectively resolves simpler scenarios within the assumed conditions. In contrast, Q-learning offers a viable approach, with DQN demonstrating superior performance in addressing more complex, realistic problems. The conclusion drawn from this study is that each algorithm exhibits unique strengths in its respective operational environment. While direct comparison between the models based on output analysis is not feasible due to the variation in environmental settings, it is evident that all three algorithms significantly contribute to resolving the targeted issues in outpatient service management. This research not only provides valuable insights into hospital outpatient service optimization but also opens avenues for further exploration in the application of advanced computational techniques in healthcare management.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"36 12","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805141","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}
Since Deep Neural Networks easily overfit label errors, which will degenerate the performance of Deep Learning algorithms, recent research gives a lot of methodology for this problem. A recent model, causalNL, uses a structural causalNL model for instance-dependent label-noise learning and obtained excellent experimental results. The implementation of the algorithm is based on the VAE model, which encodes latent variables Y and Z with the observable variables X and Y. This in turn generates the transfer matrix. But it relies on some unreasonable assumptions. In this paper, we introduce CGAN to the causalNL model, which avoids setting P(Y) and P(Z) for a specific distribution. GANs ability of processing data do not need to set a specific distribution. ICC was validated on several authoritative datasets and compared to a variety of proven algorithms including causalNL. The paper presents notable findings on the ICC model (Introduce CGAN to causalNL) shows excellent training ability on most datasets. Surprisingly, ICC shows totally higher accuracy than causalNL in CIFAR10.
由于深度神经网络很容易出现标签过拟合错误，这会使深度学习算法的性能下降，因此最近的研究给出了很多解决这一问题的方法。最近的一个模型，causalNL，使用了一个结构化的 causalNL 模型，用于实例依赖的标签噪声学习，取得了很好的实验结果。该算法的实现基于 VAE 模型，即用可观测变量 X 和 Y 对潜变量 Y 和 Z 进行编码，进而生成转移矩阵。但它依赖于一些不合理的假设。在本文中，我们将 CGAN 引入因果 NL 模型，从而避免了为特定分布设置 P(Y) 和 P(Z)。GAN 处理数据的能力不需要设置特定的分布。ICC 在多个权威数据集上进行了验证，并与包括 causalNL 在内的多种成熟算法进行了比较。论文介绍了 ICC 模型（将 CGAN 引入 causalNL）在大多数数据集上表现出的出色训练能力。令人惊讶的是，在 CIFAR10 中，ICC 的准确率完全高于 causalNL。
{"title":"Label noise learning with the combination of CausalNL and CGAN models","authors":"Zixing Gou, Yifan Sun, Zhebin Jin, Hanqiu Hu, Weiyi Xia","doi":"10.54254/2755-2721/79/20241399","DOIUrl":"https://doi.org/10.54254/2755-2721/79/20241399","url":null,"abstract":"Since Deep Neural Networks easily overfit label errors, which will degenerate the performance of Deep Learning algorithms, recent research gives a lot of methodology for this problem. A recent model, causalNL, uses a structural causalNL model for instance-dependent label-noise learning and obtained excellent experimental results. The implementation of the algorithm is based on the VAE model, which encodes latent variables Y and Z with the observable variables X and Y. This in turn generates the transfer matrix. But it relies on some unreasonable assumptions. In this paper, we introduce CGAN to the causalNL model, which avoids setting P(Y) and P(Z) for a specific distribution. GANs ability of processing data do not need to set a specific distribution. ICC was validated on several authoritative datasets and compared to a variety of proven algorithms including causalNL. The paper presents notable findings on the ICC model (Introduce CGAN to causalNL) shows excellent training ability on most datasets. Surprisingly, ICC shows totally higher accuracy than causalNL in CIFAR10.","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"48 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805568","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}
Pub Date : 2024-07-25DOI: 10.54254/2755-2721/69/20241503
Xu Li, Qi Shen, Tiancheng Yang
This paper explores the design and optimization of multidimensional data models to enhance the query performance and data analysis capabilities of OLAP (Online Analytical Processing) systems. It delves into three prominent dimensional modeling techniques: Star Schema, Snowflake Schema, and Galaxy Schema, analyzing their impact on query complexity, data redundancy, storage requirements, and ease of maintenance. Additionally, it examines three aggregation strategiesPre-Aggregation, Dynamic Aggregation, and Hybrid Aggregationfocusing on their effectiveness in balancing query response time, storage efficiency, flexibility, and computational cost. The study further investigates performance optimization techniques, including query optimization, partitioning, and materialized views, providing case studies and experimental data to illustrate their benefits and challenges. The findings underscore the importance of tailored optimization strategies in OLAP systems to meet varying business needs and query patterns, highlighting the trade-offs between performance gains, storage requirements, and implementation complexity
{"title":"Design and optimization of multidimensional data models for enhanced OLAP query performance and data analysis","authors":"Xu Li, Qi Shen, Tiancheng Yang","doi":"10.54254/2755-2721/69/20241503","DOIUrl":"https://doi.org/10.54254/2755-2721/69/20241503","url":null,"abstract":"This paper explores the design and optimization of multidimensional data models to enhance the query performance and data analysis capabilities of OLAP (Online Analytical Processing) systems. It delves into three prominent dimensional modeling techniques: Star Schema, Snowflake Schema, and Galaxy Schema, analyzing their impact on query complexity, data redundancy, storage requirements, and ease of maintenance. Additionally, it examines three aggregation strategiesPre-Aggregation, Dynamic Aggregation, and Hybrid Aggregationfocusing on their effectiveness in balancing query response time, storage efficiency, flexibility, and computational cost. The study further investigates performance optimization techniques, including query optimization, partitioning, and materialized views, providing case studies and experimental data to illustrate their benefits and challenges. The findings underscore the importance of tailored optimization strategies in OLAP systems to meet varying business needs and query patterns, highlighting the trade-offs between performance gains, storage requirements, and implementation complexity","PeriodicalId":502253,"journal":{"name":"Applied and Computational Engineering","volume":"45 19","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141805605","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：