Pub Date : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543217
Yaxuan Fang, Junhan Wang, Lei Jia, Fung Wai Kin
Shopee has been a popular online shopping website in the Southeast Asia. Customers appreciate its easy, secure, and fast online shopping experience tailored to their region. At the same time, it allows customers to choose the one with the lower price of the same product. It relies on the product matching, that is the same product with the same description image must be removed. The base technology to achieve this function is multimodal learning, in which we focus on the images and text. In our article, we proposed a new multimodal learning model mainly based on transformer and BERT. For image matching, we use NFNet, Swin_Transformer and Efficientnet to get image embeddings. For text matching, we use Distil-Bert, Albert, Multilingual Bert and TF-IDF to get text embeddings. After we get the embedding vector, we choose KNN to classify. We use cosine and distance to measure the similarity of the different models. It is worth mentioning that the loss function is Arcface, not the traditional Softmax, which improve the difficulty of training to ensure the final performance in the test periods. In addition, 7 models vote for the final results ensuring the effect of prediction. To avoid the bad matching result, we add some postprocessing process.
{"title":"Shopee Price Match Guarantee Algorithm based on multimodal learning","authors":"Yaxuan Fang, Junhan Wang, Lei Jia, Fung Wai Kin","doi":"10.1109/CSAIEE54046.2021.9543217","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543217","url":null,"abstract":"Shopee has been a popular online shopping website in the Southeast Asia. Customers appreciate its easy, secure, and fast online shopping experience tailored to their region. At the same time, it allows customers to choose the one with the lower price of the same product. It relies on the product matching, that is the same product with the same description image must be removed. The base technology to achieve this function is multimodal learning, in which we focus on the images and text. In our article, we proposed a new multimodal learning model mainly based on transformer and BERT. For image matching, we use NFNet, Swin_Transformer and Efficientnet to get image embeddings. For text matching, we use Distil-Bert, Albert, Multilingual Bert and TF-IDF to get text embeddings. After we get the embedding vector, we choose KNN to classify. We use cosine and distance to measure the similarity of the different models. It is worth mentioning that the loss function is Arcface, not the traditional Softmax, which improve the difficulty of training to ensure the final performance in the test periods. In addition, 7 models vote for the final results ensuring the effect of prediction. To avoid the bad matching result, we add some postprocessing process.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126389223","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543231
Kuihua Zhang, Min Hu, Fuji Ren, Pengyuan Hu
Sentiment analysis data on Chinese shopping comments has gained much attention in recent years. Many previous studies focus on the relationship between words in a single sentence but ignore the context relationship between sentences. To better serve this problem, we propose a method based on Bidirectional Encoder Representations from Transformers (BERT) pre-training language model, Hierarchical Bi-directional Long Short-Term Memory (Hierarchical Bi-LSTM) and attention mechanism for Chinese sentiment analysis. We first use BERT pretraining language model to obtained word vector, then applies Hierarchical Bi-LSTM model to extract contextual feature from sentences and words. Finally, we inj ect attention mechanism to highlight key information. Base on the experimental results, our method achieves more idealistic performance.
{"title":"Sentiment Analysis of Chinese Product Reviews Based on BERT Word Vector and Hierarchical Bidirectional LSTM","authors":"Kuihua Zhang, Min Hu, Fuji Ren, Pengyuan Hu","doi":"10.1109/CSAIEE54046.2021.9543231","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543231","url":null,"abstract":"Sentiment analysis data on Chinese shopping comments has gained much attention in recent years. Many previous studies focus on the relationship between words in a single sentence but ignore the context relationship between sentences. To better serve this problem, we propose a method based on Bidirectional Encoder Representations from Transformers (BERT) pre-training language model, Hierarchical Bi-directional Long Short-Term Memory (Hierarchical Bi-LSTM) and attention mechanism for Chinese sentiment analysis. We first use BERT pretraining language model to obtained word vector, then applies Hierarchical Bi-LSTM model to extract contextual feature from sentences and words. Finally, we inj ect attention mechanism to highlight key information. Base on the experimental results, our method achieves more idealistic performance.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131144059","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543227
Runlin Li, H. Zou, Shitian He, Xu Cao, Fei Cheng, Li Sun
With the development of remote sensing technology and deep neural network, high-resolution optical remote sensing image bridge target detection based on deep learning has become a research hotspot. Bridge target detection is a great challenge because of its arbitrary direction, diverse scale and complex background. In view of the characteristics of bridge targets in remote sensing image, we propose a bridge target detection algorithm based on rotation and scale invariance. Our method is improved based on the DetectoRS network. Aiming at the difficulties of bridge with different scales and multi-directions, we use Recursive Feature Pyramid (RFP) to extract the scale invariant feature and add orientation-invariant model (OIM) to extract rotation invariant feature. In addition, most of the bridge dataset are labeled with horizontal rectangle, it is difficult for network to extract the rotation invariant feature, and the scale feature of bridge will also be blurred. In this paper, a rotated box regression algorithm based on Boxinst, a weakly supervised learning method, is proposed to transform the annotation. A cloud and negative sample data enhancement strategy is proposed since the background of remote images is complicated and there are a lot of false alarms with similar shapes as bridges. The algorithm we proposed in this paper has greatly improved the accuracy of bridge target detection in remote images with complex scenes, and achieved the second place in the preliminary competition in the bridge detection track of the 2020 Gaofen Challenge on the Automated High-Resolution Earth Observation Image Interpretation, with the map of 84.48%.
{"title":"Bridge Detection Algorithm Based on Rotation and Scale Invariance","authors":"Runlin Li, H. Zou, Shitian He, Xu Cao, Fei Cheng, Li Sun","doi":"10.1109/CSAIEE54046.2021.9543227","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543227","url":null,"abstract":"With the development of remote sensing technology and deep neural network, high-resolution optical remote sensing image bridge target detection based on deep learning has become a research hotspot. Bridge target detection is a great challenge because of its arbitrary direction, diverse scale and complex background. In view of the characteristics of bridge targets in remote sensing image, we propose a bridge target detection algorithm based on rotation and scale invariance. Our method is improved based on the DetectoRS network. Aiming at the difficulties of bridge with different scales and multi-directions, we use Recursive Feature Pyramid (RFP) to extract the scale invariant feature and add orientation-invariant model (OIM) to extract rotation invariant feature. In addition, most of the bridge dataset are labeled with horizontal rectangle, it is difficult for network to extract the rotation invariant feature, and the scale feature of bridge will also be blurred. In this paper, a rotated box regression algorithm based on Boxinst, a weakly supervised learning method, is proposed to transform the annotation. A cloud and negative sample data enhancement strategy is proposed since the background of remote images is complicated and there are a lot of false alarms with similar shapes as bridges. The algorithm we proposed in this paper has greatly improved the accuracy of bridge target detection in remote images with complex scenes, and achieved the second place in the preliminary competition in the bridge detection track of the 2020 Gaofen Challenge on the Automated High-Resolution Earth Observation Image Interpretation, with the map of 84.48%.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"273 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131994468","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543264
Yunes Al-Dhabi, Shuang Zhang
Nowadays, people are facing an emerging problem called deepfake videos. These videos were created using deep learning technology. Some are created just for fun, while others are trying to manipulate your opinions, cause threats to your privacy, reputation, and so on. Sometimes, deepfake videos created using the latest algorithms can be hard to distinguish with the naked eye. That's why we need better algorithms to detect deepfake. The system we are going to present is based on a combination of CNN and RNN, as research shows that using CNN and RNN combined achieve better results. We are going to use a pre-trained CNN model called Resnext50. Using this, we save the time of training the model from scratch. The proposed system uses Resnext pretrained model for Feature Extraction and these extracted features are used to train the Long short-term memory (LSTM). Using CNN and RNN combined, we capture the inter frames as well as intra frames features which will be used to detect if the video is real or fake. We evaluated our method using a large collection of deepfake videos gathered from a variety of distribution sources. We demonstrate how our system may obtain competitive results while utilizing a simplistic architecture.
{"title":"Deepfake Video Detection by Combining Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN)","authors":"Yunes Al-Dhabi, Shuang Zhang","doi":"10.1109/CSAIEE54046.2021.9543264","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543264","url":null,"abstract":"Nowadays, people are facing an emerging problem called deepfake videos. These videos were created using deep learning technology. Some are created just for fun, while others are trying to manipulate your opinions, cause threats to your privacy, reputation, and so on. Sometimes, deepfake videos created using the latest algorithms can be hard to distinguish with the naked eye. That's why we need better algorithms to detect deepfake. The system we are going to present is based on a combination of CNN and RNN, as research shows that using CNN and RNN combined achieve better results. We are going to use a pre-trained CNN model called Resnext50. Using this, we save the time of training the model from scratch. The proposed system uses Resnext pretrained model for Feature Extraction and these extracted features are used to train the Long short-term memory (LSTM). Using CNN and RNN combined, we capture the inter frames as well as intra frames features which will be used to detect if the video is real or fake. We evaluated our method using a large collection of deepfake videos gathered from a variety of distribution sources. We demonstrate how our system may obtain competitive results while utilizing a simplistic architecture.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116224727","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543356
Jin Xiao, Bo Liu, Sihan Li, Ke Liao, Jing Huang
In the era of internet and big data, traditional method of user preferences mining has been difficult to keep up with the update speed of enterprise product or service decision adjustment, so it is a new idea to apply recommendation algorithm to user preferences mining. Most of the recommendation algorithms based on review emotion analysis are carried out at a single level of fine-granularity or coarse-granularity, which is difficult to ensure the accuracy and comprehensiveness of user preferences mining. This paper proposes a new recommendation algorithm EAFM, which is based on cross-grained emotion analysis. Based on the latent dirichlet allocation, dependency syntactic analysis and convolutional neural network model, the algorithm synchronously performs fine-grained and coarse-grained emotion analysis with online review data as corpus, and then proposes the emotion score correction mechanism, which solves the problems of data sparsity and algorithm time complexity in user preference mining. In the experimental design section, we use Amazon product data for verification, and regard root mean square error as the performance evaluation index. Experimental results show that the EAFM approach has better user preference mining performance than the compared algorithm.
{"title":"Research on Recommendation Algorithm Based on Cross-grained Emotion Analysis","authors":"Jin Xiao, Bo Liu, Sihan Li, Ke Liao, Jing Huang","doi":"10.1109/CSAIEE54046.2021.9543356","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543356","url":null,"abstract":"In the era of internet and big data, traditional method of user preferences mining has been difficult to keep up with the update speed of enterprise product or service decision adjustment, so it is a new idea to apply recommendation algorithm to user preferences mining. Most of the recommendation algorithms based on review emotion analysis are carried out at a single level of fine-granularity or coarse-granularity, which is difficult to ensure the accuracy and comprehensiveness of user preferences mining. This paper proposes a new recommendation algorithm EAFM, which is based on cross-grained emotion analysis. Based on the latent dirichlet allocation, dependency syntactic analysis and convolutional neural network model, the algorithm synchronously performs fine-grained and coarse-grained emotion analysis with online review data as corpus, and then proposes the emotion score correction mechanism, which solves the problems of data sparsity and algorithm time complexity in user preference mining. In the experimental design section, we use Amazon product data for verification, and regard root mean square error as the performance evaluation index. Experimental results show that the EAFM approach has better user preference mining performance than the compared algorithm.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123697572","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543457
Zicheng Wang
Serverless computing - including “Function as a service (FaaS)”, gives a flexible computing model for users. Today, cloud providers use container to create isolated computing environment for FaaS users. However, containers share a same kernel for all instances run on top of that, which cannot guarantee an ABI-level security as virtual machine does. Therefore, a new kind of virtual machine with container-level low overhead, named as “micro VM” or “light weight virtual machine” comes. But using virtual machines means trade off. Comparing to the high performance and lightweight containers, virtual machines usually have unavoidable problems like I/O (input and output), and some existing problems of containers like the cold start latency may become more severe. But how much it takes and if it is deserving? This paper provides a comparison between traditional virtual machine, container, and the new light weight virtual machine (named micro VM) in terms of scalability and performance, aiming to determine whether the micro VM can be the suitable computing platform for FaaS.
{"title":"Can “micro VM” become the next generation computing platform?: Performance comparison between light weight Virtual Machine, container, and traditional Virtual Machine","authors":"Zicheng Wang","doi":"10.1109/CSAIEE54046.2021.9543457","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543457","url":null,"abstract":"Serverless computing - including “Function as a service (FaaS)”, gives a flexible computing model for users. Today, cloud providers use container to create isolated computing environment for FaaS users. However, containers share a same kernel for all instances run on top of that, which cannot guarantee an ABI-level security as virtual machine does. Therefore, a new kind of virtual machine with container-level low overhead, named as “micro VM” or “light weight virtual machine” comes. But using virtual machines means trade off. Comparing to the high performance and lightweight containers, virtual machines usually have unavoidable problems like I/O (input and output), and some existing problems of containers like the cold start latency may become more severe. But how much it takes and if it is deserving? This paper provides a comparison between traditional virtual machine, container, and the new light weight virtual machine (named micro VM) in terms of scalability and performance, aiming to determine whether the micro VM can be the suitable computing platform for FaaS.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125775165","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543111
R. Yang, Peixu Cai, Luming Wang
Aiming at problems such as poor target orientation, redundant path inflection points and collision risk in sampling-based planning algorithm such as RRT and RRT*. Strategies for solving those problems are presented in recent work of papers which based on improving Bi-RRT that is an extension of RRT with faster convergence. This paper provides a comparison and analytical review of those strategies correspond to those problems which the performance of the strategies in terms of path length, processing time and total number of nodes in tree are presented through MATLAB simulation. Moreover, the optimal strategies are selected and implemented in Bi-RRT* which has faster convergence speed as compared to its predecessor of Bi-RRT. Further, certain aspects of improved Bi-RRT* based on selected strategies are found to be improved by comparing to traditional Bi-RRT*.
{"title":"Comparison of Strategies for Optimizing Bi-RRT* on Mobile Robots","authors":"R. Yang, Peixu Cai, Luming Wang","doi":"10.1109/CSAIEE54046.2021.9543111","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543111","url":null,"abstract":"Aiming at problems such as poor target orientation, redundant path inflection points and collision risk in sampling-based planning algorithm such as RRT and RRT*. Strategies for solving those problems are presented in recent work of papers which based on improving Bi-RRT that is an extension of RRT with faster convergence. This paper provides a comparison and analytical review of those strategies correspond to those problems which the performance of the strategies in terms of path length, processing time and total number of nodes in tree are presented through MATLAB simulation. Moreover, the optimal strategies are selected and implemented in Bi-RRT* which has faster convergence speed as compared to its predecessor of Bi-RRT. Further, certain aspects of improved Bi-RRT* based on selected strategies are found to be improved by comparing to traditional Bi-RRT*.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"186 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126024221","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 : 2021-08-20DOI: 10.1109/csaiee54046.2021.9543299
{"title":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE 2021)","authors":"","doi":"10.1109/csaiee54046.2021.9543299","DOIUrl":"https://doi.org/10.1109/csaiee54046.2021.9543299","url":null,"abstract":"","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115788308","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543103
Hao Peng, T. Xiang, Zhehui Huang, Chenye Tang
Among all the rapidly developing pathological diagnosis methods, particular cell therapy has become one of the most popular research tasks. In some occasions of cell detection and segmentation, several methods of separating presented touching and overlapping cell structures need to be utilized. Applying and developing these methods has become one of the most crucial and error-prone tasks in further analysis of brightfield images. In this work, we choose HeLa cells in a specific cell tracking dataset to detect HeLa cells in brightfield images and describe an approach to do cell detection and further analysis. Given a set of brightfield HeLa cell images in the cell cycle, we separate them into the border, centre, and blank sessions as the labels. Patches are extracted from images after binarization. When they are distinguished and labelled, we utilize different filters as pre-process labels and carry on data augmentation to obtain abundant patches as our training dataset. We find that SVM is a desirable model for classification since it performs well in most datasets, and LeNet, which is able to respond to a part of the surrounding units, can also be applied in our experiment. Therefore, we prefer SVM and LeNet as our models to do classification and prediction. In optical microscopy, especially when transmitted light and fluorescence microscopy are related to the specific cell structure segmentation, the distinct approach that we introduced in this work about separating touching and overlapping cell structures represents a desirable performance with high efficiency and robustness
{"title":"The Detection of Hela Cells in Brightfield Images","authors":"Hao Peng, T. Xiang, Zhehui Huang, Chenye Tang","doi":"10.1109/CSAIEE54046.2021.9543103","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543103","url":null,"abstract":"Among all the rapidly developing pathological diagnosis methods, particular cell therapy has become one of the most popular research tasks. In some occasions of cell detection and segmentation, several methods of separating presented touching and overlapping cell structures need to be utilized. Applying and developing these methods has become one of the most crucial and error-prone tasks in further analysis of brightfield images. In this work, we choose HeLa cells in a specific cell tracking dataset to detect HeLa cells in brightfield images and describe an approach to do cell detection and further analysis. Given a set of brightfield HeLa cell images in the cell cycle, we separate them into the border, centre, and blank sessions as the labels. Patches are extracted from images after binarization. When they are distinguished and labelled, we utilize different filters as pre-process labels and carry on data augmentation to obtain abundant patches as our training dataset. We find that SVM is a desirable model for classification since it performs well in most datasets, and LeNet, which is able to respond to a part of the surrounding units, can also be applied in our experiment. Therefore, we prefer SVM and LeNet as our models to do classification and prediction. In optical microscopy, especially when transmitted light and fluorescence microscopy are related to the specific cell structure segmentation, the distinct approach that we introduced in this work about separating touching and overlapping cell structures represents a desirable performance with high efficiency and robustness","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121168648","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 : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543211
Haomin Pang, Zhaoxu Wu, Haibo Luo, Biwu Yi
With the vigorous development of the Internet to combat criminal activities such as black and gray production, the problem of data classification is gradually being taken seriously. Therefore, by modeling and analyzing the browser history records that have been acquired, in which Chinese word separation in the field of neuro-linguistic programming (NLP) is used for word separation, feature extraction using a vocabulary table model, and classification processing by a neural network algorithm. Simulation experiments on browser history data through feature extraction and neural networks are conducted to train the accuracy of the model for analyzing browser history records and classifying the test data.
{"title":"Deep learning-based browser record analysis research","authors":"Haomin Pang, Zhaoxu Wu, Haibo Luo, Biwu Yi","doi":"10.1109/CSAIEE54046.2021.9543211","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543211","url":null,"abstract":"With the vigorous development of the Internet to combat criminal activities such as black and gray production, the problem of data classification is gradually being taken seriously. Therefore, by modeling and analyzing the browser history records that have been acquired, in which Chinese word separation in the field of neuro-linguistic programming (NLP) is used for word separation, feature extraction using a vocabulary table model, and classification processing by a neural network algorithm. Simulation experiments on browser history data through feature extraction and neural networks are conducted to train the accuracy of the model for analyzing browser history records and classifying the test data.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127717473","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：