Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137779
Xingyu Ling
To accurately detect the movements of break dance, a movement detection strategy based on improved SSD is proposed. Among them, in order to reduce the calculation amount of traditional SSD, MobileNet_V2 network is used to replace the traditional VGG backbone network, and then the mutex loss function is introduced to weaken the interference of overlapping movements on detection. Finally, the test is carried out in the data set. The results show that after optimization by Loss function, the detection of the model is more accurate in the case of overlapping targets. The accuracy of the model on the test set is 93.4%, and the recall rate is 91.6%, which indicates that the proposed detection network model has a good effect on movement tracking capture, and it can be used in the movement tracking detection of break dance.
{"title":"Movement Tracking Detection of Break Dance Based on Deep Learning","authors":"Xingyu Ling","doi":"10.1109/ACAIT56212.2022.10137779","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137779","url":null,"abstract":"To accurately detect the movements of break dance, a movement detection strategy based on improved SSD is proposed. Among them, in order to reduce the calculation amount of traditional SSD, MobileNet_V2 network is used to replace the traditional VGG backbone network, and then the mutex loss function is introduced to weaken the interference of overlapping movements on detection. Finally, the test is carried out in the data set. The results show that after optimization by Loss function, the detection of the model is more accurate in the case of overlapping targets. The accuracy of the model on the test set is 93.4%, and the recall rate is 91.6%, which indicates that the proposed detection network model has a good effect on movement tracking capture, and it can be used in the movement tracking detection of break dance.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131362943","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137822
Zhixun Liang, Peng Chen, Yunfei Yi, Yuanyuan Fan
In the coming smart city, the explosive growth of data makes the amount of data contained in news texts more and more, which leads to the decrease in the accuracy of traditional machine learning or deep learning models in the news text classification. Therefore, in this paper, we propose a news text classification model based on BiLSTM-Attention. The data set is selected as 30,000 news texts, and the word segmentation is carried out in turn. The stop words are removed, and the word vector is quantified. Then, the data set is cross-validated according to the ratio of training set to validation set of 8:1. Finally, the experiments with the bilstm model, lstm model and bilstm-short text model show that the BiLSTM-Attention model has the highest accuracy and the lowest loss value. In order to further verify the classification performance of BiLSTMAttention model, the experiment is designed again and Bayes and SVM are added to compare. The experimental results show that the accuracy, recall and F1 value of BiLSTM-Attention model are the highest, which proves that BiLSTM-Attention is more suitable for news text classification.
{"title":"News Text Classification Method for Edge Computing Based on BiLSTM-Attention","authors":"Zhixun Liang, Peng Chen, Yunfei Yi, Yuanyuan Fan","doi":"10.1109/ACAIT56212.2022.10137822","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137822","url":null,"abstract":"In the coming smart city, the explosive growth of data makes the amount of data contained in news texts more and more, which leads to the decrease in the accuracy of traditional machine learning or deep learning models in the news text classification. Therefore, in this paper, we propose a news text classification model based on BiLSTM-Attention. The data set is selected as 30,000 news texts, and the word segmentation is carried out in turn. The stop words are removed, and the word vector is quantified. Then, the data set is cross-validated according to the ratio of training set to validation set of 8:1. Finally, the experiments with the bilstm model, lstm model and bilstm-short text model show that the BiLSTM-Attention model has the highest accuracy and the lowest loss value. In order to further verify the classification performance of BiLSTMAttention model, the experiment is designed again and Bayes and SVM are added to compare. The experimental results show that the accuracy, recall and F1 value of BiLSTM-Attention model are the highest, which proves that BiLSTM-Attention is more suitable for news text classification.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131735474","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137800
Huang Guo, Rui Wang, Xiandi Jiang
In recent years, text recommendation has been widely used in various APPs as a key technology for users to quickly and accurately obtain relevant information. Traditional text recommendation cannot obtain the internal relationship between users and articles, and ignores the information generated by users. Therefore, this paper proposes a matching recommendation mechanism based on articles and comments. First introduce the word2vec word vector model, use the vector to measure the relative meaning between words, and construct the document vector and user distribution vector based on the word vector. Then, under the framework of the topic model, a joint deep learning method—long and short-term memory network LSTM, makes full use of the new model before and after the sentence to learn the document to update the word vector expression of the sentence and document vector. Among them, the conditional random field (CRF) model is added to train the tags to solve the problem of insufficient attention to key words. Finally, in the matching mechanism, the similar relationship among the topic distributions, the constructed document vector and the user vector are used for training. Compared with the current popular topic model TopicRNN method, topic word vector model LF-LDA method, topic vector-based text representation method and four methods of LF-LDA combined with Word2vec text representation, the experimental results show that the matching recommendation classification is obtained Improved and very robust, training time is greatly shortened, the algorithm in this paper is effective.
{"title":"A Matching Recommendation Mechanism Based on Deep Learning and Topic Model","authors":"Huang Guo, Rui Wang, Xiandi Jiang","doi":"10.1109/ACAIT56212.2022.10137800","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137800","url":null,"abstract":"In recent years, text recommendation has been widely used in various APPs as a key technology for users to quickly and accurately obtain relevant information. Traditional text recommendation cannot obtain the internal relationship between users and articles, and ignores the information generated by users. Therefore, this paper proposes a matching recommendation mechanism based on articles and comments. First introduce the word2vec word vector model, use the vector to measure the relative meaning between words, and construct the document vector and user distribution vector based on the word vector. Then, under the framework of the topic model, a joint deep learning method—long and short-term memory network LSTM, makes full use of the new model before and after the sentence to learn the document to update the word vector expression of the sentence and document vector. Among them, the conditional random field (CRF) model is added to train the tags to solve the problem of insufficient attention to key words. Finally, in the matching mechanism, the similar relationship among the topic distributions, the constructed document vector and the user vector are used for training. Compared with the current popular topic model TopicRNN method, topic word vector model LF-LDA method, topic vector-based text representation method and four methods of LF-LDA combined with Word2vec text representation, the experimental results show that the matching recommendation classification is obtained Improved and very robust, training time is greatly shortened, the algorithm in this paper is effective.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130409042","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137982
Shaobo Deng, Min Li, Xuegang Li, Lei Wang, Sujie Guan
The k-means clustering algorithm is a very classical clustering algorithm that is widely used because of its excellent efficiency and performance. The algorithm uses Euclidean distance to calculate the similarity between samples and iteratively updates the membership matrix to obtain clustering results. However, when k-means algorithm clusters datasets containing samples with intra-cluster distances greater than inter-cluster distances, errors often occur when partitioning the boundary samples, which eventually leads to unsatisfactory results. Moreover, although k-means algorithm makes the intra-cluster distance as small as possible, it neglects to maximize the inter-cluster distance, and eventually only finds the local optimal solution. Different from the existing k-means type algorithm, this paper proposes a similarity measure based on the impact factor, which determines the partitioning result by comparing the impact of samples on each cluster. And on the basis of the objective function of k-means algorithm, we combine the inter-cluster distance to solve the defects of local optimality that exist in k-means algorithm. In the paper, we theoretically analyze and prove the proposed method, and compare and analyze the clustering results of the algorithm with the class k-means algorithm on real datasets, and confirm that the proposed algorithm in this paper can effectively avoid the defects of the class k-means algorithm.
{"title":"An Improved K-Means Algorithm Based on Impact Index","authors":"Shaobo Deng, Min Li, Xuegang Li, Lei Wang, Sujie Guan","doi":"10.1109/ACAIT56212.2022.10137982","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137982","url":null,"abstract":"The k-means clustering algorithm is a very classical clustering algorithm that is widely used because of its excellent efficiency and performance. The algorithm uses Euclidean distance to calculate the similarity between samples and iteratively updates the membership matrix to obtain clustering results. However, when k-means algorithm clusters datasets containing samples with intra-cluster distances greater than inter-cluster distances, errors often occur when partitioning the boundary samples, which eventually leads to unsatisfactory results. Moreover, although k-means algorithm makes the intra-cluster distance as small as possible, it neglects to maximize the inter-cluster distance, and eventually only finds the local optimal solution. Different from the existing k-means type algorithm, this paper proposes a similarity measure based on the impact factor, which determines the partitioning result by comparing the impact of samples on each cluster. And on the basis of the objective function of k-means algorithm, we combine the inter-cluster distance to solve the defects of local optimality that exist in k-means algorithm. In the paper, we theoretically analyze and prove the proposed method, and compare and analyze the clustering results of the algorithm with the class k-means algorithm on real datasets, and confirm that the proposed algorithm in this paper can effectively avoid the defects of the class k-means algorithm.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"59 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130607107","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137977
Weibiao Huang, Xueqin Zheng
Aiming at the low accuracy of PID control in D-STATCOM (Distribution Static Synchronous Compensator), the design of Adaptive genetic algorithm-traditional PID controller of D-STATCOM control is studied when voltage swell and voltage sag occur in power grid. The q-axis actual current and reference current waveform, active power and reactive power waveform in D-STATCOM are compared and analyzed. The peak value of the DC side voltage waveform, grid voltage waveform and grid current waveform are effectively suppressed. The tracking error of q-axis reference current to the actual current waveform of D-STATCOM is reduced to 0.12pu. In the case of voltage swell and dip, the power grid is accurately compensated for reactive power and the voltage is stabilized.
{"title":"Design of PID Controllers in D-STATCOM Based on Adaptive Genetic Algorithm","authors":"Weibiao Huang, Xueqin Zheng","doi":"10.1109/ACAIT56212.2022.10137977","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137977","url":null,"abstract":"Aiming at the low accuracy of PID control in D-STATCOM (Distribution Static Synchronous Compensator), the design of Adaptive genetic algorithm-traditional PID controller of D-STATCOM control is studied when voltage swell and voltage sag occur in power grid. The q-axis actual current and reference current waveform, active power and reactive power waveform in D-STATCOM are compared and analyzed. The peak value of the DC side voltage waveform, grid voltage waveform and grid current waveform are effectively suppressed. The tracking error of q-axis reference current to the actual current waveform of D-STATCOM is reduced to 0.12pu. In the case of voltage swell and dip, the power grid is accurately compensated for reactive power and the voltage is stabilized.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116634093","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137890
Xiao Chang, Jianguang Sun
Aiming at the problem of poor regularity of hierarchical phrases distribution in Chinese-English machine translation software, this paper constructs a lexical ordering model of hierarchical phrases in Chinese-English machine translation software to improve the accuracy of Chinese-English machine translation software translation. This paper proposes a lexical ordering method of hierarchical phrases in Chinese-English machine translation software based on dynamic reusability and structured partition fusion. This paper constructs the rule type distribution set of Chinese-English machine translation software hierarchical phrases, adopts the dynamic compilation method of rules to realize semantic feature detection and sparse parameter identification of Chinese-English machine translation software hierarchical phrases, and adopts the multi-dimensional semantic network grouping feature sorting and dynamic detection method to cluster the target language monolingual corpus of Chinese-English machine translation software hierarchical phrases. This paper establishes the entity structure model of bilingual corpus that takes into account diversity, realizes semantic feature enhancement and information fusion after the combination of the translated text and the original bilingual corpus through tensor expression of data clustering, realizes grouping and filtering of interference data through multi-dimensional scale extended clustering processing of hierarchical phrases of Chinese-English machine translation software, and rearranges structured data of hierarchical phrases of Chinese-English machine translation software by link-based clustering method, thus realizing lexicalization and reordering of hierarchical phrases of Chinese-English machine translation software. The simulation results show that this method has good anti-interference performance and high accuracy of lexicalization, which improves the ability of extracting and identifying the lexical information of hierarchical phrases in Chinese-English machine translation software, thus improving the accuracy of Chinese-English machine translation software.
{"title":"Research on Lexicalization and Ordering Methods of Hierarchical Phrases in Chinese-English Machine Translation Software","authors":"Xiao Chang, Jianguang Sun","doi":"10.1109/ACAIT56212.2022.10137890","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137890","url":null,"abstract":"Aiming at the problem of poor regularity of hierarchical phrases distribution in Chinese-English machine translation software, this paper constructs a lexical ordering model of hierarchical phrases in Chinese-English machine translation software to improve the accuracy of Chinese-English machine translation software translation. This paper proposes a lexical ordering method of hierarchical phrases in Chinese-English machine translation software based on dynamic reusability and structured partition fusion. This paper constructs the rule type distribution set of Chinese-English machine translation software hierarchical phrases, adopts the dynamic compilation method of rules to realize semantic feature detection and sparse parameter identification of Chinese-English machine translation software hierarchical phrases, and adopts the multi-dimensional semantic network grouping feature sorting and dynamic detection method to cluster the target language monolingual corpus of Chinese-English machine translation software hierarchical phrases. This paper establishes the entity structure model of bilingual corpus that takes into account diversity, realizes semantic feature enhancement and information fusion after the combination of the translated text and the original bilingual corpus through tensor expression of data clustering, realizes grouping and filtering of interference data through multi-dimensional scale extended clustering processing of hierarchical phrases of Chinese-English machine translation software, and rearranges structured data of hierarchical phrases of Chinese-English machine translation software by link-based clustering method, thus realizing lexicalization and reordering of hierarchical phrases of Chinese-English machine translation software. The simulation results show that this method has good anti-interference performance and high accuracy of lexicalization, which improves the ability of extracting and identifying the lexical information of hierarchical phrases in Chinese-English machine translation software, thus improving the accuracy of Chinese-English machine translation software.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125729671","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137917
Sijie Shen, Qianqian Qiu, Sujie Guan, Min Li, Shaobo Deng
With the rapid and vigorous development of fuzzy clustering theory and methods, more fuzzy clustering algorithms have been proposed to establish the uncertainty description of the samples. However, when clustering is performed, existing fuzzy clustering algorithms mostly iterate feature weights or deal with noise.The objective function is mostly based on minimizing the Euclidean distance within the clusters. However, increasing the Euclidean distance between cluster centroids may also lead to an improvement in clustering performance.In this paper, a new fuzzy c-mean clustering algorithm (JCFCM) combining inter-cluster distances is proposed. Not only is an affiliation assigned within the original cluster, but it is also reflected in the form of affiliation between clusters.In this paper, clustering is performed by increasing the process of iterative selection of cluster centers between clusters. With this formalization an objective function is designed and the iterative formulas for the parameters in the function are obtained by solving the objective function optimally. Finally, experiments are conducted on five real data sets and compared with other fuzzy clustering algorithms. Overall, the JCFCM algorithm has better clustering results than the fuzzy C-mean algorithm and has some advantages over the existing improved fuzzy C-mean algorithm for different data sets.
{"title":"Fuzzy C-Mean Clustering Algorithm Combining Inter-Cluster Distance","authors":"Sijie Shen, Qianqian Qiu, Sujie Guan, Min Li, Shaobo Deng","doi":"10.1109/ACAIT56212.2022.10137917","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137917","url":null,"abstract":"With the rapid and vigorous development of fuzzy clustering theory and methods, more fuzzy clustering algorithms have been proposed to establish the uncertainty description of the samples. However, when clustering is performed, existing fuzzy clustering algorithms mostly iterate feature weights or deal with noise.The objective function is mostly based on minimizing the Euclidean distance within the clusters. However, increasing the Euclidean distance between cluster centroids may also lead to an improvement in clustering performance.In this paper, a new fuzzy c-mean clustering algorithm (JCFCM) combining inter-cluster distances is proposed. Not only is an affiliation assigned within the original cluster, but it is also reflected in the form of affiliation between clusters.In this paper, clustering is performed by increasing the process of iterative selection of cluster centers between clusters. With this formalization an objective function is designed and the iterative formulas for the parameters in the function are obtained by solving the objective function optimally. Finally, experiments are conducted on five real data sets and compared with other fuzzy clustering algorithms. Overall, the JCFCM algorithm has better clustering results than the fuzzy C-mean algorithm and has some advantages over the existing improved fuzzy C-mean algorithm for different data sets.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133224344","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137981
Bailing Xu
Aiming at the problem of poor analysis performance of traditional patent data in the biomedical field, a parallel strategy based on the combination of Spark framework and K-means clustering algorithm was proposed. Firstly, Spark tool was used to initially process the big data. Then, K-means clustering algorithm was used to cluster and analyze the patent data, and obtain the optimal solution, so as to realize the intelligent analysis of patent data. Experimental results showed that in the same test sample data and sample classification results, compared with a single K-means clustering algorithm, the proposed parallel clustering analysis algorithm has a better classification effect on the quantity and category of patent data, which can prove that the analysis effect of parallel clustering algorithm is better. At the same time, the parallel strategy greatly improves the accuracy and speed of patent data analysis, thereby effectively improving the ability of clustering and analysis of massive data.
{"title":"Intelligent Analysis of Patent Data in the Biomedical Field Based on Spark Parallel Clustering Algorithm","authors":"Bailing Xu","doi":"10.1109/ACAIT56212.2022.10137981","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137981","url":null,"abstract":"Aiming at the problem of poor analysis performance of traditional patent data in the biomedical field, a parallel strategy based on the combination of Spark framework and K-means clustering algorithm was proposed. Firstly, Spark tool was used to initially process the big data. Then, K-means clustering algorithm was used to cluster and analyze the patent data, and obtain the optimal solution, so as to realize the intelligent analysis of patent data. Experimental results showed that in the same test sample data and sample classification results, compared with a single K-means clustering algorithm, the proposed parallel clustering analysis algorithm has a better classification effect on the quantity and category of patent data, which can prove that the analysis effect of parallel clustering algorithm is better. At the same time, the parallel strategy greatly improves the accuracy and speed of patent data analysis, thereby effectively improving the ability of clustering and analysis of massive data.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130954844","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}
Cancer is one of the main diseases that threaten human death, and nasopharyngeal cancer also shows a high mortality rate. The early diagnosis is particularly important in the proper treatment of cancers. Computer-aided diagnosis has been widely used in the medical field. To harness the artificial intelligence in medical imaging, we implement two types of attention mechanism in the popular convolutional neural network ResNet50 to aid classification and diagnosis of the medical images of nasopharyngeal cancer. Compared with basic ResNet50 architecture, both “Convolutional Block Attention Module (CBAM)” and “Dual Attention Network (DANet)” gain the improved classification performance. Our results show that the implementing location affects the results. We compare six types of implementing ways, named as CBAM-A, CBAM-B, DANet-A, DANet-B, Fusion-A and Fusion-B. Among six models, DANet-B implemented network achieves the 96.5% accuracy, 96.8% precision, 96.5 % recall and 96.4 % F1-score, showing significant improvement compared with the basic ResNet50 with values of 54.4% accuracy, 60.5% precision, 54.4% recall and 50.6% F1-score, respectively. The results show that proper implementing attention mechanism can improve the classification performance and may be developed as an auxiliary diagnosis approach for the Nasopharyngeal Cancer.
{"title":"Implementing Attention Mechanism in Convolutional Neural Network to Improve Performance of MRI Image Classification of Nasopharyngeal Cancer","authors":"Rongzhi Mao, Wei Song, Cheng Ge, Xiaojun Xu, Liangxu Xie","doi":"10.1109/ACAIT56212.2022.10137993","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137993","url":null,"abstract":"Cancer is one of the main diseases that threaten human death, and nasopharyngeal cancer also shows a high mortality rate. The early diagnosis is particularly important in the proper treatment of cancers. Computer-aided diagnosis has been widely used in the medical field. To harness the artificial intelligence in medical imaging, we implement two types of attention mechanism in the popular convolutional neural network ResNet50 to aid classification and diagnosis of the medical images of nasopharyngeal cancer. Compared with basic ResNet50 architecture, both “Convolutional Block Attention Module (CBAM)” and “Dual Attention Network (DANet)” gain the improved classification performance. Our results show that the implementing location affects the results. We compare six types of implementing ways, named as CBAM-A, CBAM-B, DANet-A, DANet-B, Fusion-A and Fusion-B. Among six models, DANet-B implemented network achieves the 96.5% accuracy, 96.8% precision, 96.5 % recall and 96.4 % F1-score, showing significant improvement compared with the basic ResNet50 with values of 54.4% accuracy, 60.5% precision, 54.4% recall and 50.6% F1-score, respectively. The results show that proper implementing attention mechanism can improve the classification performance and may be developed as an auxiliary diagnosis approach for the Nasopharyngeal Cancer.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133077843","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137829
Shijiao Liu
To further strengthen the epidemic prevention and control management in schools, an improved stay point recognition algorithm based on the density-based spatial clustering of applications with noise (DBSCAN) is proposed to achieve accurate recognition of student activity trajectories. The experimental results show that the improved stay point recognition algorithm based on DBSCAN can realize the accurate recognition of student activity trajectories. When the time threshold MinPts is set to 10min and the radius threshold $varepsilon$ is set to 20m, the recall rate of trajectory stay point recognition reaches 97% and the precision rate reaches 90%. Compared with other algorithms, the recognition algorithm proposed in this paper has a higher recognition accuracy, reaching 0.9873. The above experimental results verify the feasibility of the trajectory analysis method proposed in this paper, which has certain practical value.
{"title":"Analysis of College Students’ Trajectories Utilizing Data Mining Under Epidemic Prevention and Control","authors":"Shijiao Liu","doi":"10.1109/ACAIT56212.2022.10137829","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137829","url":null,"abstract":"To further strengthen the epidemic prevention and control management in schools, an improved stay point recognition algorithm based on the density-based spatial clustering of applications with noise (DBSCAN) is proposed to achieve accurate recognition of student activity trajectories. The experimental results show that the improved stay point recognition algorithm based on DBSCAN can realize the accurate recognition of student activity trajectories. When the time threshold MinPts is set to 10min and the radius threshold $varepsilon$ is set to 20m, the recall rate of trajectory stay point recognition reaches 97% and the precision rate reaches 90%. Compared with other algorithms, the recognition algorithm proposed in this paper has a higher recognition accuracy, reaching 0.9873. The above experimental results verify the feasibility of the trajectory analysis method proposed in this paper, which has certain practical value.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133112451","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：