Pub Date : 2021-08-20DOI: 10.1109/CSAIEE54046.2021.9543134
Jiaming Gu
With the development of the novel coronavirus epidemic, virus detection and research has gradually become a hot research direction. The structure of the virus is mainly divided into protein shell and ribonucleic acid (RNA). RNA is an important information-carrying biopolymer within biological cells that plays a key role in regulatory processes and transcription control. Studies of RNA-induced conditions, including human immunodeficiency viruses, neocymavirus, and even Alzheimer's and Parkinson's disease, require an understanding of the structure and function of RNA. As a result, the study of RNA is becoming increasingly important in a range of applications, including biology and medicine. The function of RNA is determined primarily by the thermodynamic three-stage folding of a sequence of nucleotides. The hydrogen bond between nucleotides determines the main driving force for the formation of a three-stage structure. Smaller folds around the hydrogen bond are called secondary structures of RNA. The three-stage structure determines the function and nature of RNA, and traditional manual exploration of RNA tertiary structures, such as X-ray crystal diffraction, and MRI to determine RNA tertiary structures, while accurate and reliable, is labor-intensive and time-consuming. Accurate judgment of secondary structures has greatly influenced the study of RNA tertiary structures and deeper studies, and the exploration of RNA secondary structures with artificial intelligence can lead to more accurate, rapid and efficient results. In the current field, artificial intelligence algorithms to predict RNA secondary structures usually use deep learning, genetic algorithms and other means, through neural network fitting to obtain prediction results. This approach is supervised learning, requiring a large amount of RNA secondary structure data to be collated prior to the study, while the models trained are not explanatory. As we all know, RNA folding is driven primarily by thermodynamics, can we train a model that learns the principles of RNA folding on its own, based on limited structural data? The main research direction of this paper is to explore the secondary structure model of ribonucleic acid independently by using algorithms in the way of computer deep-strengthening learning. Deep-enhanced learning primarily transforms the prediction process of the RNA secondary structure into the process of intelligent decision-making to explore optimal decision-making. Due to the limited training set and computing power, this paper explores the feasibility and development potential of deep-enhanced learning algorithms in RNA secondary structure prediction. In the current field, artificial intelligence algorithms to predict RNA secondary structures usually use deep learning, genetic algorithms and other means, through neural network fitting to obtain prediction results. This approach is supervised learning, requiring a large amount of RNA secondary structure data to be co
{"title":"Application of deep intensive learning in RNA secondary structure prediction","authors":"Jiaming Gu","doi":"10.1109/CSAIEE54046.2021.9543134","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543134","url":null,"abstract":"With the development of the novel coronavirus epidemic, virus detection and research has gradually become a hot research direction. The structure of the virus is mainly divided into protein shell and ribonucleic acid (RNA). RNA is an important information-carrying biopolymer within biological cells that plays a key role in regulatory processes and transcription control. Studies of RNA-induced conditions, including human immunodeficiency viruses, neocymavirus, and even Alzheimer's and Parkinson's disease, require an understanding of the structure and function of RNA. As a result, the study of RNA is becoming increasingly important in a range of applications, including biology and medicine. The function of RNA is determined primarily by the thermodynamic three-stage folding of a sequence of nucleotides. The hydrogen bond between nucleotides determines the main driving force for the formation of a three-stage structure. Smaller folds around the hydrogen bond are called secondary structures of RNA. The three-stage structure determines the function and nature of RNA, and traditional manual exploration of RNA tertiary structures, such as X-ray crystal diffraction, and MRI to determine RNA tertiary structures, while accurate and reliable, is labor-intensive and time-consuming. Accurate judgment of secondary structures has greatly influenced the study of RNA tertiary structures and deeper studies, and the exploration of RNA secondary structures with artificial intelligence can lead to more accurate, rapid and efficient results. In the current field, artificial intelligence algorithms to predict RNA secondary structures usually use deep learning, genetic algorithms and other means, through neural network fitting to obtain prediction results. This approach is supervised learning, requiring a large amount of RNA secondary structure data to be collated prior to the study, while the models trained are not explanatory. As we all know, RNA folding is driven primarily by thermodynamics, can we train a model that learns the principles of RNA folding on its own, based on limited structural data? The main research direction of this paper is to explore the secondary structure model of ribonucleic acid independently by using algorithms in the way of computer deep-strengthening learning. Deep-enhanced learning primarily transforms the prediction process of the RNA secondary structure into the process of intelligent decision-making to explore optimal decision-making. Due to the limited training set and computing power, this paper explores the feasibility and development potential of deep-enhanced learning algorithms in RNA secondary structure prediction. In the current field, artificial intelligence algorithms to predict RNA secondary structures usually use deep learning, genetic algorithms and other means, through neural network fitting to obtain prediction results. This approach is supervised learning, requiring a large amount of RNA secondary structure data to be co","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115864836","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}
Deep learning technologies have shown impressive performance in many areas. However, deep learning systems can be deceived by using intentionally crafted data, says, adversarial samples. This inherent vulnerability limits its application in safety-critical domains such as automatic driving, military applications and so on. As a kind of defense measures, various approaches have been proposed to detect adversarial samples, among which their efficiency should be further improved to accomplish practical application requirements. In this paper, we proposed a neuron coverage-based approach which detect adversarial samples by distinguishing the activated neurons' distribution features in classifier layer. The analysis and experiments showed that this approach achieves high accuracy while having relatively low computation and storage cost.
{"title":"Detecting Adversarial Samples with Neuron Coverage","authors":"Huayang Cao, Wei Kong, Xiaohui Kuang, Jianwen Tian","doi":"10.1109/CSAIEE54046.2021.9543451","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543451","url":null,"abstract":"Deep learning technologies have shown impressive performance in many areas. However, deep learning systems can be deceived by using intentionally crafted data, says, adversarial samples. This inherent vulnerability limits its application in safety-critical domains such as automatic driving, military applications and so on. As a kind of defense measures, various approaches have been proposed to detect adversarial samples, among which their efficiency should be further improved to accomplish practical application requirements. In this paper, we proposed a neuron coverage-based approach which detect adversarial samples by distinguishing the activated neurons' distribution features in classifier layer. The analysis and experiments showed that this approach achieves high accuracy while having relatively low computation and storage cost.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123194570","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/CSAIEE54046.2021.9543209
Wei Guo, Lirong Hu, Dong Yu
The threshold voltage variations due to the random doping fluctuation and the increasing disturb due to the large number of cells on the chip affect the stability of the read, write, and hold operations of the SRAM circuit. As the smallest size and most numerous module within the chip, the stability of the SRAM memory cell, whether it is used in the configuration chain or in the storage array, is a prerequisite of the whole chip correct function. Most of the articles about the immunity of SRAM circuits focus on changing the circuit structure, such as using 7T, 8T, or 10T memory cells, or adding read and write assist circuits. However, such structural changes can significantly increase the area and the chip leakage current. In this paper, the impact of the size and threshold voltage on its operational stability is analyzed from the perspective of quantitative analysis and simulation. And the optimal SRAM 6T memory cell size is selected based on 28nm CMOS.
{"title":"A Size Optimization Scheme of SRAM 6T Memory Cell","authors":"Wei Guo, Lirong Hu, Dong Yu","doi":"10.1109/CSAIEE54046.2021.9543209","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543209","url":null,"abstract":"The threshold voltage variations due to the random doping fluctuation and the increasing disturb due to the large number of cells on the chip affect the stability of the read, write, and hold operations of the SRAM circuit. As the smallest size and most numerous module within the chip, the stability of the SRAM memory cell, whether it is used in the configuration chain or in the storage array, is a prerequisite of the whole chip correct function. Most of the articles about the immunity of SRAM circuits focus on changing the circuit structure, such as using 7T, 8T, or 10T memory cells, or adding read and write assist circuits. However, such structural changes can significantly increase the area and the chip leakage current. In this paper, the impact of the size and threshold voltage on its operational stability is analyzed from the perspective of quantitative analysis and simulation. And the optimal SRAM 6T memory cell size is selected based on 28nm CMOS.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126075375","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/CSAIEE54046.2021.9543388
Jiaorui Shen
Software test plays an important role in software engineering, and dynamic symbolic execution (DSE) has become a popular technique in white-box testing. However, the efficiency of DSE is a big challenge of this technique. Compiler optimizations may have a big impact on DSE in some cases. In this paper, we introduce two small examples to visually show the impact of compiler optimizations on constraints solving and path exploration of DSE. After that, we propose a series of experiments using KLEE and LL VM compiler as a case to test real C programs in Coreutils-8.32. We use a simple model to assess the impact of different compiler optimizations and we also study on the combinations of compiler optimizations. The results show compiler optimizations can have both positive and negative effects, and some optimizations like FI may have greater influences than others. Moreover, some combinations of compiler optimizations can better improve the efficiency of DSE than single compiler optimization, which can be further studied.
{"title":"The Impact of Compiler Optimizations on Symbolic Execution","authors":"Jiaorui Shen","doi":"10.1109/CSAIEE54046.2021.9543388","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543388","url":null,"abstract":"Software test plays an important role in software engineering, and dynamic symbolic execution (DSE) has become a popular technique in white-box testing. However, the efficiency of DSE is a big challenge of this technique. Compiler optimizations may have a big impact on DSE in some cases. In this paper, we introduce two small examples to visually show the impact of compiler optimizations on constraints solving and path exploration of DSE. After that, we propose a series of experiments using KLEE and LL VM compiler as a case to test real C programs in Coreutils-8.32. We use a simple model to assess the impact of different compiler optimizations and we also study on the combinations of compiler optimizations. The results show compiler optimizations can have both positive and negative effects, and some optimizations like FI may have greater influences than others. Moreover, some combinations of compiler optimizations can better improve the efficiency of DSE than single compiler optimization, which can be further studied.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127881346","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}
With the rapid development of artificial intelligence, the convolution of the neural network has been deploying across a range of industries, and has certain applications in national defense security, transportation monitoring and medical research. However, due to the constraints of speed and the power consumption, convolutional neural networks are still limited to a great extent in the implementation of the edge computing and mobile devices, etc. For this reason, we design a lightweight convolutional neural network based on FPGA. In this paper, we use the YOLOv3- Tiny algorithm, which is fast in execution, small in computation and small in size, is suitable for deployments on embedded devices such as FPGA. This paper uses 16-bit fixed-point quantization, special data storage and hardware circuit for convolutional computation written in verilog - hardware description language, consumes 512 DSPs, consumes 37.037 ms to recognize a single image frame, and consumes 9.611W. The system basically achieves the design goal of target detection.
{"title":"Lightweight convolutional neural network of YOLO v3- Tiny algorithm on FPGA for target detection","authors":"Jitong Xin, Meiyi Cha, Luojia Shi, Chunyu Long, Hairong Li, Fangcong Wang, Peng Wang","doi":"10.1109/CSAIEE54046.2021.9543128","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543128","url":null,"abstract":"With the rapid development of artificial intelligence, the convolution of the neural network has been deploying across a range of industries, and has certain applications in national defense security, transportation monitoring and medical research. However, due to the constraints of speed and the power consumption, convolutional neural networks are still limited to a great extent in the implementation of the edge computing and mobile devices, etc. For this reason, we design a lightweight convolutional neural network based on FPGA. In this paper, we use the YOLOv3- Tiny algorithm, which is fast in execution, small in computation and small in size, is suitable for deployments on embedded devices such as FPGA. This paper uses 16-bit fixed-point quantization, special data storage and hardware circuit for convolutional computation written in verilog - hardware description language, consumes 512 DSPs, consumes 37.037 ms to recognize a single image frame, and consumes 9.611W. The system basically achieves the design goal of target detection.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121896428","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/CSAIEE54046.2021.9543136
Gao Yuan, Tong Zhang, Wanlu Zhang, Hongsheng Li
At present, the prediction of stock market is one of the most popular and valuable research fields in the financial field. More and more scholars are engaged in the research of stock market forecast, exploring the law of stock market development, and new science and technology are constantly applied to the stock price forecast. In this paper, we proposed a stock closing price prediction model based on the XGBoost and Grid SearchCV algorithms. Experimental results show that our idea represents better performance than the other machine learning methods. Specifically, the RMSE value is 1.39%, 2.43% and 8.33% lower than SVM algorithm, neural network algorithm and LightGBM algorithm, respectively. In addition, we also give the importance ranking of the characteristics that affect the stock closing price, and obtain some interesting and instructive suggestions. For example, the “EMA-9” and “SMA-15” feature has the biggest and smallest impact on stock prices, which can guide our future work.
{"title":"Analysis of Stock Price Based On the XGBoost Algorithm With EMA-19 and SMA-15 Features","authors":"Gao Yuan, Tong Zhang, Wanlu Zhang, Hongsheng Li","doi":"10.1109/CSAIEE54046.2021.9543136","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543136","url":null,"abstract":"At present, the prediction of stock market is one of the most popular and valuable research fields in the financial field. More and more scholars are engaged in the research of stock market forecast, exploring the law of stock market development, and new science and technology are constantly applied to the stock price forecast. In this paper, we proposed a stock closing price prediction model based on the XGBoost and Grid SearchCV algorithms. Experimental results show that our idea represents better performance than the other machine learning methods. Specifically, the RMSE value is 1.39%, 2.43% and 8.33% lower than SVM algorithm, neural network algorithm and LightGBM algorithm, respectively. In addition, we also give the importance ranking of the characteristics that affect the stock closing price, and obtain some interesting and instructive suggestions. For example, the “EMA-9” and “SMA-15” feature has the biggest and smallest impact on stock prices, which can guide our future work.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115843508","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/CSAIEE54046.2021.9543196
Xi Sun, Z. Lv
The click-through rate prediction is with significance in the field of recommendation systems, especially in advertising recommendation systems. At present, some sequence models based on deep learning have been directly used in the field of the click-through rate prediction to dig out the rule of user behavior and have achieved good results, but they ignored the influence of time information on the rule of user behavior. To solve the above problems, we propose a model named Time Interval Encoding Deep Session Interest Network (TIED-DSIN). In the TIED-DSIN model, a time interval encoding method is designed to integrate time interval information into the sequence model, and time decay factor is introduced in the encoding process to make the model consider the influence of time information fully when mining the rule of users' dynamic behaviors. Correspondingly, a comparative experiment is conducted on the real Alimama public data set, and the results show that the accuracy of the TIED-DSIN model is better than other models that commonly used in the click-through rate prediction.
{"title":"Deep Session Interest Network Based on the Time Interval Encoding for the Click-through Rate Prediction","authors":"Xi Sun, Z. Lv","doi":"10.1109/CSAIEE54046.2021.9543196","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543196","url":null,"abstract":"The click-through rate prediction is with significance in the field of recommendation systems, especially in advertising recommendation systems. At present, some sequence models based on deep learning have been directly used in the field of the click-through rate prediction to dig out the rule of user behavior and have achieved good results, but they ignored the influence of time information on the rule of user behavior. To solve the above problems, we propose a model named Time Interval Encoding Deep Session Interest Network (TIED-DSIN). In the TIED-DSIN model, a time interval encoding method is designed to integrate time interval information into the sequence model, and time decay factor is introduced in the encoding process to make the model consider the influence of time information fully when mining the rule of users' dynamic behaviors. Correspondingly, a comparative experiment is conducted on the real Alimama public data set, and the results show that the accuracy of the TIED-DSIN model is better than other models that commonly used in the click-through rate prediction.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132209778","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/CSAIEE54046.2021.9543296
Gongliu Yang, Kun Zhang, Ruizhao Cheng, Yongfeng Zhang
Most MEMS inertial navigation systems (INS) need to meet a wide working temperature range of - 20 ~+ 55°C. MEMS gyroscope is the core component of MEMS INS and the accuracy of MEMS gyros directly affects the navigation performance. However, the output of MEMS gyros is inevitably affected by temperature. Due to the limited temperature error compensation accuracy of traditional method based on least squares polynomial fitting, the paper presents a new temperature error compensation method based on Whale Optimization Algorithm (WOA) optimized support vector regression (SVR), which is achieved the optimization of SVR by WOA. After simulation and MEMS gyros temperature error compensation test. The results show that the WOA-SVR can effectively compensate the temperature error of MEMS gyros. And the accuracy is significantly improved compared with the traditional methods.
{"title":"A Novel Temperature Error Compensation method for MEMS Gyros Based on WOA-SVR","authors":"Gongliu Yang, Kun Zhang, Ruizhao Cheng, Yongfeng Zhang","doi":"10.1109/CSAIEE54046.2021.9543296","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543296","url":null,"abstract":"Most MEMS inertial navigation systems (INS) need to meet a wide working temperature range of - 20 ~+ 55°C. MEMS gyroscope is the core component of MEMS INS and the accuracy of MEMS gyros directly affects the navigation performance. However, the output of MEMS gyros is inevitably affected by temperature. Due to the limited temperature error compensation accuracy of traditional method based on least squares polynomial fitting, the paper presents a new temperature error compensation method based on Whale Optimization Algorithm (WOA) optimized support vector regression (SVR), which is achieved the optimization of SVR by WOA. After simulation and MEMS gyros temperature error compensation test. The results show that the WOA-SVR can effectively compensate the temperature error of MEMS gyros. And the accuracy is significantly improved compared with the traditional methods.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124896572","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/CSAIEE54046.2021.9543322
Yizhou Yang
Ensemble learning is a system which used to train multiple learning models and combine their results, treating them as a “committee” of decision makers. To explore effect of ensemble learning, this paper applied two basic ensemble systems of encoder to natural language processing. To compare the individual models and ensemble systems, this paper varied the number models which used to calculate ensemble accuracies. The result is that the decision of the model, with all models combined, usually have better overall accuracy, on average, than any single model. It shown that ensemble system used all models usually have better performance. This paper given explanation in the conclusion section of this result.
{"title":"Basic Ensemble Learning of Encoder Representations from Transformer for Disaster-mentioning Tweets Classification","authors":"Yizhou Yang","doi":"10.1109/CSAIEE54046.2021.9543322","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543322","url":null,"abstract":"Ensemble learning is a system which used to train multiple learning models and combine their results, treating them as a “committee” of decision makers. To explore effect of ensemble learning, this paper applied two basic ensemble systems of encoder to natural language processing. To compare the individual models and ensemble systems, this paper varied the number models which used to calculate ensemble accuracies. The result is that the decision of the model, with all models combined, usually have better overall accuracy, on average, than any single model. It shown that ensemble system used all models usually have better performance. This paper given explanation in the conclusion section of this result.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114669459","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/CSAIEE54046.2021.9543368
Ling Zheng, Peixin Xu, Jiayin Bai
In the power grid system, the accurate determination and precise positioning of faulty equipment is an important foundation for realizing the self-healing function of the power grid. With the increasing complexity of the power grid topology, traditional fault location methods are prone to misjudge the faulty equipment and the normal equipment in the station, causing problems such as low positioning accuracy and poor stability. To locate faulty equipment quickly and accurately, this paper proposes a power grid fault location method based on pre-training of convolutional autoencoder. The method is based on the synchronous phasor measurement unit (PMU), which converts its sequence data into images to reduce noise interference. The method pre-trains lots of samples by using convolutional autoencoder, and then uses a classifier to fine-tune small batches of balanced samples. Finally, the faulty equipment in the power grid can be accurately located by dividing the area multiple times and positioning the faulty equipment gradually. Experiments and simulations show that compared with other mainstream methods, this method has higher robustness and stability, can effectively alleviate the impact of data imbalance, and improve the accuracy and accuracy of faulty equipment location.
{"title":"Power Grid Fault Location Method Based on Pretraining of Convolutional Autoencoder","authors":"Ling Zheng, Peixin Xu, Jiayin Bai","doi":"10.1109/CSAIEE54046.2021.9543368","DOIUrl":"https://doi.org/10.1109/CSAIEE54046.2021.9543368","url":null,"abstract":"In the power grid system, the accurate determination and precise positioning of faulty equipment is an important foundation for realizing the self-healing function of the power grid. With the increasing complexity of the power grid topology, traditional fault location methods are prone to misjudge the faulty equipment and the normal equipment in the station, causing problems such as low positioning accuracy and poor stability. To locate faulty equipment quickly and accurately, this paper proposes a power grid fault location method based on pre-training of convolutional autoencoder. The method is based on the synchronous phasor measurement unit (PMU), which converts its sequence data into images to reduce noise interference. The method pre-trains lots of samples by using convolutional autoencoder, and then uses a classifier to fine-tune small batches of balanced samples. Finally, the faulty equipment in the power grid can be accurately located by dividing the area multiple times and positioning the faulty equipment gradually. Experiments and simulations show that compared with other mainstream methods, this method has higher robustness and stability, can effectively alleviate the impact of data imbalance, and improve the accuracy and accuracy of faulty equipment location.","PeriodicalId":376014,"journal":{"name":"2021 IEEE International Conference on Computer Science, Artificial Intelligence and Electronic Engineering (CSAIEE)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116104669","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：