Pub Date : 2023-12-01DOI: 10.53106/160792642023122407011
Tse-Chuan Hsu Tse-Chuan Hsu, Chih-Hung Chang Tse-Chuan Hsu, William Cheng-Chung Chu Chih-Hung Chang, Shou-Yu Lee William Cheng-Chung Chu, Shih-Yun Huang Shou-Yu Lee
Computer vision technology allows the computer to interact with a human operator to quickly complete the interpretation of events and improve the operational workflow of processing events. As computer vision technology continues to evolve, the cost of the equipment continues to increase. Therefore, the stability of the system can be ensured through the design of the middleware and the calculation of the auxiliary functions of the software agents. At present, the recognition of characters for image processing is based on the technology of image recognition, which can provide a more flexible user experience. However, the dilemma of contactless design lies in the processing and calculation of images, which should reduce the inconvenience caused by delays.This article uses a Raspberry Pi as an example of a computing proxy application. After the visual inspection, the verification operation of the software is carried out. Our system is the detection of hand position and movement, and the detection of hand mark position in reconnaissance. In addition, we simultaneously developed management and remote control events and connected to the remote edge computer. After that, we successfully completed the automatic control and serial application of two different edge computing recognition jobs, and verified the image vision computing based on the Raspberry Pi software agent, which can be used for image vision analysis and control applications.
计算机视觉技术允许计算机与人类操作员进行交互，从而快速完成对事件的解释，并改进处理事件的操作流程。随着计算机视觉技术的不断发展，设备的成本也在不断增加。因此，可以通过中间件的设计和软件代理辅助功能的计算来保证系统的稳定性。目前，图像处理的字符识别是基于图像识别的技术，可以提供更加灵活的用户体验。然而，非接触式设计的困境在于图像的处理和计算，应减少延迟带来的不便。本文以 Raspberry Pi 为例，介绍计算代理应用。在视觉检测之后，进行软件的验证操作。我们的系统是手部位置和移动的检测，以及侦察中手部标记位置的检测。此外，我们还同时开发了管理和远程控制事件，并连接到远程边缘计算机。之后，我们成功完成了两种不同边缘计算识别作业的自动控制和串行应用，并验证了基于树莓派软件代理的图像视觉计算，可用于图像视觉分析和控制应用。
{"title":"A Dynamic Model for the Computation of Gesture Types for Image-based Software Agents","authors":"Tse-Chuan Hsu Tse-Chuan Hsu, Chih-Hung Chang Tse-Chuan Hsu, William Cheng-Chung Chu Chih-Hung Chang, Shou-Yu Lee William Cheng-Chung Chu, Shih-Yun Huang Shou-Yu Lee","doi":"10.53106/160792642023122407011","DOIUrl":"https://doi.org/10.53106/160792642023122407011","url":null,"abstract":"Computer vision technology allows the computer to interact with a human operator to quickly complete the interpretation of events and improve the operational workflow of processing events. As computer vision technology continues to evolve, the cost of the equipment continues to increase. Therefore, the stability of the system can be ensured through the design of the middleware and the calculation of the auxiliary functions of the software agents. At present, the recognition of characters for image processing is based on the technology of image recognition, which can provide a more flexible user experience. However, the dilemma of contactless design lies in the processing and calculation of images, which should reduce the inconvenience caused by delays.This article uses a Raspberry Pi as an example of a computing proxy application. After the visual inspection, the verification operation of the software is carried out. Our system is the detection of hand position and movement, and the detection of hand mark position in reconnaissance. In addition, we simultaneously developed management and remote control events and connected to the remote edge computer. After that, we successfully completed the automatic control and serial application of two different edge computing recognition jobs, and verified the image vision computing based on the Raspberry Pi software agent, which can be used for image vision analysis and control applications.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"54 2","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139187806","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 : 2023-12-01DOI: 10.53106/160792642023122407008
Minjuan Zhong Minjuan Zhong, Han Yang Minjuan Zhong, Keyang Zhong Han Yang, Xilong Qu Keyang Zhong, Zhenjin Li Xilong Qu
Consumers depend on online reviews to influences their purchase decisions. On account of that,many vendors and retailers try to manipulate online reviews to mislead potential consumers to take risky purchase decisions. Many scholars have conducted a lot of research on the impact of online product reviews on consumer behavior and sales. However, the existing work are mainly based on the premise of real product reviews, but few attentions have been paid of fake ones. Based on the recognition results of deceptive reviews, this article explores whether consumers be aware or perceive it when deceptive reviews are flooding the online review system, and further analyze what influence will be imposed on final purchase decision with different perception. The empirical analysis of the questionnaire survey show that in the context of two different perceptions of consumers, deceptive reviews have significant differences in the results of purchase decisions. In addition, research also shows that consumers’ persuasive knowledge plays a moderating role between perceived deception and purchase decision.
{"title":"The Impact of Online Reviews Manipulation on Consumer Purchase Decision Based on The Perspective of Consumers’ Perception","authors":"Minjuan Zhong Minjuan Zhong, Han Yang Minjuan Zhong, Keyang Zhong Han Yang, Xilong Qu Keyang Zhong, Zhenjin Li Xilong Qu","doi":"10.53106/160792642023122407008","DOIUrl":"https://doi.org/10.53106/160792642023122407008","url":null,"abstract":"Consumers depend on online reviews to influences their purchase decisions. On account of that,many vendors and retailers try to manipulate online reviews to mislead potential consumers to take risky purchase decisions. Many scholars have conducted a lot of research on the impact of online product reviews on consumer behavior and sales. However, the existing work are mainly based on the premise of real product reviews, but few attentions have been paid of fake ones. Based on the recognition results of deceptive reviews, this article explores whether consumers be aware or perceive it when deceptive reviews are flooding the online review system, and further analyze what influence will be imposed on final purchase decision with different perception. The empirical analysis of the questionnaire survey show that in the context of two different perceptions of consumers, deceptive reviews have significant differences in the results of purchase decisions. In addition, research also shows that consumers’ persuasive knowledge plays a moderating role between perceived deception and purchase decision.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"87 3-4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139189769","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 : 2023-12-01DOI: 10.53106/160792642023122407006
Wen-Li Lee Wen-Li Lee, Koyin Chang Wen-Li Lee, Ying-Chen Chi Koyin Chang, Wen-Shou Chou Ying-Chen Chi, Chen-Long Wu Wen-Shou Chou
Cardiovascular diseases (CVDs) are the leading cause of mortality globally. To effectively prevent CVDs, a variety of techniques have been employed to evaluate the mechanical properties of arteries, among which, aortic stiffness measured by aortic pulse wave velocity (PWV) has been proven to be an independent predictor of CVDs. However, the traditional way to measure PWV is complex and time consuming. Recent studies suggest the digital volume pulse (DVP) waveform to be an effective non-invasive method to obtain PWV. In this study, we present a cloud computing system that analyzes and calculates the relevant indices of arterial stiffness after receiving the measured DVP signals. The result of the analysis can be retrieved online for the user to view or download for further analysis. With this technique, arterial Stiffness Index (SI) for population can be obtained easily and inexpensively. This will help health authorities to do mass screening at population level and, hence, establish references of arterial SI for different cohorts by age, gender, ethnicity, and diseases.
{"title":"A Cloud-Based Assessment of Arterial Stiffness Through Contour Analysis of A Photoplethysmography","authors":"Wen-Li Lee Wen-Li Lee, Koyin Chang Wen-Li Lee, Ying-Chen Chi Koyin Chang, Wen-Shou Chou Ying-Chen Chi, Chen-Long Wu Wen-Shou Chou","doi":"10.53106/160792642023122407006","DOIUrl":"https://doi.org/10.53106/160792642023122407006","url":null,"abstract":"Cardiovascular diseases (CVDs) are the leading cause of mortality globally. To effectively prevent CVDs, a variety of techniques have been employed to evaluate the mechanical properties of arteries, among which, aortic stiffness measured by aortic pulse wave velocity (PWV) has been proven to be an independent predictor of CVDs. However, the traditional way to measure PWV is complex and time consuming. Recent studies suggest the digital volume pulse (DVP) waveform to be an effective non-invasive method to obtain PWV. In this study, we present a cloud computing system that analyzes and calculates the relevant indices of arterial stiffness after receiving the measured DVP signals. The result of the analysis can be retrieved online for the user to view or download for further analysis. With this technique, arterial Stiffness Index (SI) for population can be obtained easily and inexpensively. This will help health authorities to do mass screening at population level and, hence, establish references of arterial SI for different cohorts by age, gender, ethnicity, and diseases.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"11 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139192071","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}
This paper established a Convolutional Neural Network (CNN) with the concept of transfer learning, and combined the main feature analysis calculation and clustering algorithm to further demonstrate the superiority of the proposed small CNN trainer in the identification of traditional Kinmen desserts. Food dessert identification methods directly use skin texture, color, shape, and other features as the basis. This paper effectively extracted image features of an object by the small CNN trainer and classified the featured dataset into the right food categories. It was able to complete classification quickly and also achieved high-precision classification results. Different types of Kinmen desserts were identified through a multi-layer training cycle. A total of 100 training images for each of the 10 food categories and each image size is converted into a smaller training data set by capturing the important features through the CNN trainer. Then, the main features were generated and the dimensions of each food image data were reduced again by using the t-Distributed Stochastic Neighbor Embedding (t-SNE) or Principal Component Analysis (PCA) methods. An individually K-means or k-nearest neighbors (KNN) algorithms efficiently completed the grouping results and in the classified image restoration. The experimental results compared the classifications after the learning cycle of different trainers and showed that the highest accuracy that the appropriated CNN trainer of the proposed methology obtained was up to 99% with a minimum executing time of about 99.37 seconds.
{"title":"Small Convolutional Neural Network Trainer Designed through Transfer Learning in Dessert Classification","authors":"Hua-Ching Chen Hua-Ching Chen, Hsuan-Ming Feng Hua-Ching Chen","doi":"10.53106/160792642023122407009","DOIUrl":"https://doi.org/10.53106/160792642023122407009","url":null,"abstract":"This paper established a Convolutional Neural Network (CNN) with the concept of transfer learning, and combined the main feature analysis calculation and clustering algorithm to further demonstrate the superiority of the proposed small CNN trainer in the identification of traditional Kinmen desserts. Food dessert identification methods directly use skin texture, color, shape, and other features as the basis. This paper effectively extracted image features of an object by the small CNN trainer and classified the featured dataset into the right food categories. It was able to complete classification quickly and also achieved high-precision classification results. Different types of Kinmen desserts were identified through a multi-layer training cycle. A total of 100 training images for each of the 10 food categories and each image size is converted into a smaller training data set by capturing the important features through the CNN trainer. Then, the main features were generated and the dimensions of each food image data were reduced again by using the t-Distributed Stochastic Neighbor Embedding (t-SNE) or Principal Component Analysis (PCA) methods. An individually K-means or k-nearest neighbors (KNN) algorithms efficiently completed the grouping results and in the classified image restoration. The experimental results compared the classifications after the learning cycle of different trainers and showed that the highest accuracy that the appropriated CNN trainer of the proposed methology obtained was up to 99% with a minimum executing time of about 99.37 seconds.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"11 11","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139189720","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}
Immutability enables the blockchain to store data permanently, which is considered to be the most essential property to protect the security of blockchain technology. However, illegal content stored on the blockchain cannot be modified either. "The right to be forgotten" stipulates that the data subject has the right to request the data controller to delete personal data about him, and the controller is obliged to delete the personal data in a timely manner in such circumstances. Therefore, redactable blockchains are proposed to solve the aforementioned problems. The redactable blockchain relaxes the immutability of blockchains in a controlled manner. However, a participant who is granted the privilege to modify the blockchain, he/she may add inappropriate or malicious modifications to the content of historical blocks. In this article, we introduce a redactable blockchain where transaction owners restrict what editors can rewrite in transactions. When posting an editable transaction, the transaction owner indicates what can be edited and what cannot be edited in order to restrict a transaction modifier from making malicious changes to the content of the transaction. The modifier’s changes to the transaction are then validated by the validator and any malicious changes that do not meet the requirements will fail to be validated. To counteract collusive attacks between modifiers and validators, they will be held accountable and penalized.
{"title":"Constraints-based and One-time Modification Redactable Blockchain","authors":"Chunli Wang Chunli Wang, Yuling Chen Chunli Wang, Wensheng Jia Yuling Chen","doi":"10.53106/160792642023122407005","DOIUrl":"https://doi.org/10.53106/160792642023122407005","url":null,"abstract":"Immutability enables the blockchain to store data permanently, which is considered to be the most essential property to protect the security of blockchain technology. However, illegal content stored on the blockchain cannot be modified either. \"The right to be forgotten\" stipulates that the data subject has the right to request the data controller to delete personal data about him, and the controller is obliged to delete the personal data in a timely manner in such circumstances. Therefore, redactable blockchains are proposed to solve the aforementioned problems. The redactable blockchain relaxes the immutability of blockchains in a controlled manner. However, a participant who is granted the privilege to modify the blockchain, he/she may add inappropriate or malicious modifications to the content of historical blocks. In this article, we introduce a redactable blockchain where transaction owners restrict what editors can rewrite in transactions. When posting an editable transaction, the transaction owner indicates what can be edited and what cannot be edited in order to restrict a transaction modifier from making malicious changes to the content of the transaction. The modifier’s changes to the transaction are then validated by the validator and any malicious changes that do not meet the requirements will fail to be validated. To counteract collusive attacks between modifiers and validators, they will be held accountable and penalized.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"37 22","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139195022","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 : 2023-12-01DOI: 10.53106/160792642023122407002
Xianqi Deng Xianqi Deng, Jianping Liu Xianqi Deng, Cheng Peng Jianping Liu, Yingfei Wang Cheng Peng
In this paper, we propose a new computer vision model called SE-YOLOv5-SPD for counting the number of log ends in large wood piles in log farms. This task traditionally requires a lot of manpower and previous computer vision methods struggle to detect logs in low pixels and small objects in images. Our model is based on the YOLOv5 model and incorporates the Squeeze-and-Excitation Networks (SENet) attention module and SPD-Conv (Space-to-Depth Convolution) module to improve accuracy. We also compare the performance of the SE attention module and SPD-Conv module to the CBAM attention module and Focus module using the SE-YOLOv5-SPD model. Results show that the SE-YOLOv5-SPD model can achieve excellent results of 0.652 in mAP50:95, 0.912 in mAP50, 0.968 in Precision, and 0.864 in Recall in a low-resolution environment with interference, which is significantly better than other models. Our findings indicate that the SE-YOLOv5-SPD model is a promising solution for counting the number of log ends in wood piles.
{"title":"Using Improved YOLOv5 Model to Detect Volume for Logs in Log Farms","authors":"Xianqi Deng Xianqi Deng, Jianping Liu Xianqi Deng, Cheng Peng Jianping Liu, Yingfei Wang Cheng Peng","doi":"10.53106/160792642023122407002","DOIUrl":"https://doi.org/10.53106/160792642023122407002","url":null,"abstract":"In this paper, we propose a new computer vision model called SE-YOLOv5-SPD for counting the number of log ends in large wood piles in log farms. This task traditionally requires a lot of manpower and previous computer vision methods struggle to detect logs in low pixels and small objects in images. Our model is based on the YOLOv5 model and incorporates the Squeeze-and-Excitation Networks (SENet) attention module and SPD-Conv (Space-to-Depth Convolution) module to improve accuracy. We also compare the performance of the SE attention module and SPD-Conv module to the CBAM attention module and Focus module using the SE-YOLOv5-SPD model. Results show that the SE-YOLOv5-SPD model can achieve excellent results of 0.652 in mAP50:95, 0.912 in mAP50, 0.968 in Precision, and 0.864 in Recall in a low-resolution environment with interference, which is significantly better than other models. Our findings indicate that the SE-YOLOv5-SPD model is a promising solution for counting the number of log ends in wood piles.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"42 2","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139188659","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 : 2023-12-01DOI: 10.53106/160792642023122407003
Jeng-Shyang Pan Jeng-Shyang Pan, Ling Li Jeng-Shyang Pan, Shu-Chuan Chu Ling Li, Kuo-Kun Tseng Shu-Chuan Chu, Hisham A. Shehadeh Kuo-Kun Tseng
This paper proposes a human behavior-based optimization algorithm, Martial Arts Learning Optimization (MALO), for optimization problems in continuous spaces. The algorithm simulates the process of characters in martial arts learning so as to apply it to optimization problems. Characters in martial arts stories usually go through multiple stages of learning martial arts, such as self-study and leader teaching. Multiple learning stages of characters are modeled in this paper, utilizing the wisdom of the characters learning martial arts in the novel, enabling the optimization process. To verify and analyze the performance of the proposed algorithm, the algorithm is numerically tested on 30 benchmark functions, and it is found that its performance was better than the state-of-the-art nine algorithms. In addition, the algorithm is also used to solve the problem of nighttime image brightness enhancement. Compared with other image enhancement methods, the proposed MALO algorithm has superior results in both visual effects and quantitative image quality assessments.
本文针对连续空间中的优化问题，提出了一种基于人类行为的优化算法--武术学习优化（MALO）。该算法模拟武侠小说中人物的习武过程，从而将其应用于优化问题。武侠小说中的人物通常会经历自学和领教等多个学武阶段。本文利用小说中人物学习武功的智慧，对人物的多个学习阶段进行建模，从而实现优化过程。为了验证和分析所提算法的性能，该算法在 30 个基准函数上进行了数值测试，发现其性能优于最先进的 9 种算法。此外，该算法还被用于解决夜间图像亮度增强问题。与其他图像增强方法相比，所提出的 MALO 算法在视觉效果和定量图像质量评估方面都具有更优越的效果。
{"title":"Martial Art Learning Optimization: A Novel Metaheuristic Algorithm for Night Image Enhancement","authors":"Jeng-Shyang Pan Jeng-Shyang Pan, Ling Li Jeng-Shyang Pan, Shu-Chuan Chu Ling Li, Kuo-Kun Tseng Shu-Chuan Chu, Hisham A. Shehadeh Kuo-Kun Tseng","doi":"10.53106/160792642023122407003","DOIUrl":"https://doi.org/10.53106/160792642023122407003","url":null,"abstract":"This paper proposes a human behavior-based optimization algorithm, Martial Arts Learning Optimization (MALO), for optimization problems in continuous spaces. The algorithm simulates the process of characters in martial arts learning so as to apply it to optimization problems. Characters in martial arts stories usually go through multiple stages of learning martial arts, such as self-study and leader teaching. Multiple learning stages of characters are modeled in this paper, utilizing the wisdom of the characters learning martial arts in the novel, enabling the optimization process. To verify and analyze the performance of the proposed algorithm, the algorithm is numerically tested on 30 benchmark functions, and it is found that its performance was better than the state-of-the-art nine algorithms. In addition, the algorithm is also used to solve the problem of nighttime image brightness enhancement. Compared with other image enhancement methods, the proposed MALO algorithm has superior results in both visual effects and quantitative image quality assessments.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"245 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139193052","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 : 2023-12-01DOI: 10.53106/160792642023122407010
Fang Xie Fang Xie, Hao Li Fang Xie, Beiye Zhang Hao Li, Jianan He Beiye Zhang, Xincong Zhong Jianan He
Text summarization is divided into extractive summarization and abstractive summarization. The extractive summarization technology aims to extract some main phrases and sentences from the original text to form a short summary for people to read quickly. However, extractive summarization is faced with problems such as poor sentence coherence and incomplete information, which makes it difficult to screen out important sentences from the source text. DNN (Deep Neural Network) is widely used for text summarization task. This paper proposes a TA-Sum model based on the neural topic model. Introducing the topic information can help people understand the relevant main content of source text quickly. We obtain the topic information using the neural topic model and implement the attention mechanism to fuse the topic information with the text representation, which improves the semantic integrity and completeness of the summary. The experimental results on the large-scale English data sets CNN/Daily mail are improved by 0.37%, 0.11%, and 0.17% respectively compared with BertSum, which demonstrates the effectiveness of our method.
{"title":"TA-Sum: The Extractive Summarization Research Based on Topic Information","authors":"Fang Xie Fang Xie, Hao Li Fang Xie, Beiye Zhang Hao Li, Jianan He Beiye Zhang, Xincong Zhong Jianan He","doi":"10.53106/160792642023122407010","DOIUrl":"https://doi.org/10.53106/160792642023122407010","url":null,"abstract":"Text summarization is divided into extractive summarization and abstractive summarization. The extractive summarization technology aims to extract some main phrases and sentences from the original text to form a short summary for people to read quickly. However, extractive summarization is faced with problems such as poor sentence coherence and incomplete information, which makes it difficult to screen out important sentences from the source text. DNN (Deep Neural Network) is widely used for text summarization task. This paper proposes a TA-Sum model based on the neural topic model. Introducing the topic information can help people understand the relevant main content of source text quickly. We obtain the topic information using the neural topic model and implement the attention mechanism to fuse the topic information with the text representation, which improves the semantic integrity and completeness of the summary. The experimental results on the large-scale English data sets CNN/Daily mail are improved by 0.37%, 0.11%, and 0.17% respectively compared with BertSum, which demonstrates the effectiveness of our method.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"21 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139194706","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 : 2023-11-01DOI: 10.53106/160792642023112406005
Jingliang Chen Jingliang Chen, Chenchen Wu Jingliang Chen, Shuisheng Chen Chenchen Wu, Yi Zhu Shuisheng Chen, Bin Li Yi Zhu
In the case of traditional methods such as network models and algorithms are highly open source and highly bound to hardware, data processing has become an important method to optimize the performance of neural networks. In this paper, we combine traditional data processing methods and propose a method based on the mini dataset which is strictly randomly divided within the training process; and takes the calculation results of the cross-entropy loss function as the measurement standard, by comparing the mini dataset, screening, and processing to optimize the deep neural network. Using this method, each iteration training can obtain a relatively optimal result, and the optimization effects of each time are integrated to optimize the results of each epoch. Finally, in order to verify the effectiveness and applicability of this data processing method, experiments are carried out on MNIST, HAGRID, and CIFAR-10 datasets to compare the effects of using this method and not using this method under different hyper-parameters, and finally, the effectiveness of this data processing method is verified. Finally, we summarize the advantages and limitations of this method and look forward to the future improvement direction of this method.
{"title":"Optimize the Performance of the Neural Network by using a Mini Dataset Processing Method","authors":"Jingliang Chen Jingliang Chen, Chenchen Wu Jingliang Chen, Shuisheng Chen Chenchen Wu, Yi Zhu Shuisheng Chen, Bin Li Yi Zhu","doi":"10.53106/160792642023112406005","DOIUrl":"https://doi.org/10.53106/160792642023112406005","url":null,"abstract":"In the case of traditional methods such as network models and algorithms are highly open source and highly bound to hardware, data processing has become an important method to optimize the performance of neural networks. In this paper, we combine traditional data processing methods and propose a method based on the mini dataset which is strictly randomly divided within the training process; and takes the calculation results of the cross-entropy loss function as the measurement standard, by comparing the mini dataset, screening, and processing to optimize the deep neural network. Using this method, each iteration training can obtain a relatively optimal result, and the optimization effects of each time are integrated to optimize the results of each epoch. Finally, in order to verify the effectiveness and applicability of this data processing method, experiments are carried out on MNIST, HAGRID, and CIFAR-10 datasets to compare the effects of using this method and not using this method under different hyper-parameters, and finally, the effectiveness of this data processing method is verified. Finally, we summarize the advantages and limitations of this method and look forward to the future improvement direction of this method.","PeriodicalId":442331,"journal":{"name":"網際網路技術學刊","volume":"17 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139295794","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：