Pub Date : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642186
Yulin He, Chuandong Li, Xingxing Ju
As an important application of emotion artificial intelligence, emotion classification provides the basis for the realization of affective brain-computer interface (aBCI). In this study, the NeuCube is used to learn and classify Electroencephalogram (EEG) data from the DEAP dataset. NeuCube is a type of spiking neural network (SNN) framework developed based on the real human brain. It is very suitable for analyzing and processing spatio-temporal data. Based on the 10-fold cross-validation method, we obtain a mean accuracy of 68.91 % in the emotional binary valence classification problem. Meanwhile, the EEG data recorded from F3 and F4 electrode channels provide more information compared with Fp1 and Fp2. The results prove that the spiking neural network can be applied to the task of emotion classification effectively.
{"title":"Emotion Classification Using EEG Data in a Brain-Inspired Spiking Neural Network","authors":"Yulin He, Chuandong Li, Xingxing Ju","doi":"10.1109/ICICIP53388.2021.9642186","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642186","url":null,"abstract":"As an important application of emotion artificial intelligence, emotion classification provides the basis for the realization of affective brain-computer interface (aBCI). In this study, the NeuCube is used to learn and classify Electroencephalogram (EEG) data from the DEAP dataset. NeuCube is a type of spiking neural network (SNN) framework developed based on the real human brain. It is very suitable for analyzing and processing spatio-temporal data. Based on the 10-fold cross-validation method, we obtain a mean accuracy of 68.91 % in the emotional binary valence classification problem. Meanwhile, the EEG data recorded from F3 and F4 electrode channels provide more information compared with Fp1 and Fp2. The results prove that the spiking neural network can be applied to the task of emotion classification effectively.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"131 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127292373","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 : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642220
Yunbo Yang, Na Liu, Sitian Qin
This paper studies the properties of a class of multi- agent systems. First of all, this article lists some symbols and lemmas to be used afterwards. The first main part of this article studies a class of continuous-time multi-agent systems with event-trigger mechanism and gives its consensus analysis by applying Lyapunov method. At the same time, this article also gives a modified trigger mechanism, consisting of both time intervals and event-trigger intervals for this kind of continuous system. And it is proved that under the proposed trigger mechanism, the state solutions of the given system can finally reach a consensus and the Zeno effect does not appear. Moreover, the problem of average consensus of differential privacy is also studied with an event-trigger mechanism in a discrete-time multi-agent system. The consensus and accuracy of this discrete system in the sense of mean square is studied. Through the research, it is concluded that the output states of the discrete system under the given event-trigger mechanism finally reach a consensus in the mean square sense, and the state solutions converge to the weighted average of the initial state. At the end of this paper, numerical simulations are made to illustrate the feasibility of the algorithm of this paper in practice.
{"title":"The Consensus Research on a Class of Event-Triggered Multi-Agent System","authors":"Yunbo Yang, Na Liu, Sitian Qin","doi":"10.1109/ICICIP53388.2021.9642220","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642220","url":null,"abstract":"This paper studies the properties of a class of multi- agent systems. First of all, this article lists some symbols and lemmas to be used afterwards. The first main part of this article studies a class of continuous-time multi-agent systems with event-trigger mechanism and gives its consensus analysis by applying Lyapunov method. At the same time, this article also gives a modified trigger mechanism, consisting of both time intervals and event-trigger intervals for this kind of continuous system. And it is proved that under the proposed trigger mechanism, the state solutions of the given system can finally reach a consensus and the Zeno effect does not appear. Moreover, the problem of average consensus of differential privacy is also studied with an event-trigger mechanism in a discrete-time multi-agent system. The consensus and accuracy of this discrete system in the sense of mean square is studied. Through the research, it is concluded that the output states of the discrete system under the given event-trigger mechanism finally reach a consensus in the mean square sense, and the state solutions converge to the weighted average of the initial state. At the end of this paper, numerical simulations are made to illustrate the feasibility of the algorithm of this paper in practice.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130503048","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}
The excellent performance of deep learning depends on the strong representation ability of its backbone. As a conventional means of most backbones, pretraining can make the model obtain high accuracy, but it also brings some disadvantages that can not be ignored: first, the structures of the backbones need pretraining are fixed, they are difficult to modify and migrate across tasks; second, the pretraining process needs to consume huge computing power. To solve this problem, we propose a backbone named RVNet (Residual VGGNet), which can make the model converge quickly without pretraining. The design of RVNet is divided into the following two steps: firstly, the residual convolutional layer (RCL) is designed by referring to the residual skill and BN layer, which can prevent the gradient from disappearing and restrain the data distribution. At the same time, The introduced 1* 1 convolution layer can improve the nonlinearity of the model while controlling the number of feature maps’ channels; then, based on VGGNet-19, the designed RCLs replace the original 3* 3 convolution layer to improve the representation ability of the backbone. We take the person re-identification (Re-ID) task as the research object, and prove the effectiveness and superiority of RVNet through a series of ablation experiments.
{"title":"Design of A Backbone without Pretraining","authors":"Shaoqi Hou, Wenyi Du, Yiyin Ding, Yuhao Zeng, Chunyu Wang, Guangqiang Yin","doi":"10.1109/ICICIP53388.2021.9642216","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642216","url":null,"abstract":"The excellent performance of deep learning depends on the strong representation ability of its backbone. As a conventional means of most backbones, pretraining can make the model obtain high accuracy, but it also brings some disadvantages that can not be ignored: first, the structures of the backbones need pretraining are fixed, they are difficult to modify and migrate across tasks; second, the pretraining process needs to consume huge computing power. To solve this problem, we propose a backbone named RVNet (Residual VGGNet), which can make the model converge quickly without pretraining. The design of RVNet is divided into the following two steps: firstly, the residual convolutional layer (RCL) is designed by referring to the residual skill and BN layer, which can prevent the gradient from disappearing and restrain the data distribution. At the same time, The introduced 1* 1 convolution layer can improve the nonlinearity of the model while controlling the number of feature maps’ channels; then, based on VGGNet-19, the designed RCLs replace the original 3* 3 convolution layer to improve the representation ability of the backbone. We take the person re-identification (Re-ID) task as the research object, and prove the effectiveness and superiority of RVNet through a series of ablation experiments.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"30 10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125874170","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 : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642165
Y. Kong, Jiajia Wu, Shiyong Chen, Junwen Zhou
A distributed manipulability optimization (DMO) scheme based on a finite time neural network is proposed in this paper to solve the cooperative motion planning of redundant manipulators. In this proposed kinematic scheme, the end-effectors of the manipulators can complete the specific task in a cooperative manner under peer-to-peer communication and the optimal kinematic time of redundant manipulators has achieved. The DMO scheme is formulated into a quadratic program and is solved by Lagrange multiplier theorem. The stability and finiteness of the proposed DMO scheme have been proved in theory. Simulation results on three redundant manipulators show the validity and accuracy of this new DMO scheme. method
{"title":"Distributed Manipulability optimization in a Finite Time Neural Network for Redundant Manipulators","authors":"Y. Kong, Jiajia Wu, Shiyong Chen, Junwen Zhou","doi":"10.1109/ICICIP53388.2021.9642165","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642165","url":null,"abstract":"A distributed manipulability optimization (DMO) scheme based on a finite time neural network is proposed in this paper to solve the cooperative motion planning of redundant manipulators. In this proposed kinematic scheme, the end-effectors of the manipulators can complete the specific task in a cooperative manner under peer-to-peer communication and the optimal kinematic time of redundant manipulators has achieved. The DMO scheme is formulated into a quadratic program and is solved by Lagrange multiplier theorem. The stability and finiteness of the proposed DMO scheme have been proved in theory. Simulation results on three redundant manipulators show the validity and accuracy of this new DMO scheme. method","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"83 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121907039","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}
Text classification is a classical and basic task of natural language processing (NLP). In recent years, machine learning has been widely used for text classification. However traditional machine learning methods depend heavily on high quality feature engineering. Recently, deep learning methods have contributed to improving text classification performance. Graph convolution network (GCN) has been proved to be able to capture spatial feature of documents. However, the ability of GCN to capture sentence local features and context information is limited. In this paper, we propose a novel method using local features to enhance GCN for text classification. The combination methods based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) (such as Bi-LSTM, C-LSTM and ServeNet) are used to capture local features to enrich feature information, and a weight value is used to adjust the intensity of enhancement. We conducted extensive experiments on 5 benchmark datasets (WSDataset, Ohsumed, R52, R8, 20NG), proving the proposed method significantly outperforms the baseline deep learning methods.
{"title":"A Novel Method Using Local Feature to Enhance GCN for Text Classification","authors":"Chunlian Yang, Yuchen Guo, Xiaowei Li, Benhui Chen","doi":"10.1109/ICICIP53388.2021.9642171","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642171","url":null,"abstract":"Text classification is a classical and basic task of natural language processing (NLP). In recent years, machine learning has been widely used for text classification. However traditional machine learning methods depend heavily on high quality feature engineering. Recently, deep learning methods have contributed to improving text classification performance. Graph convolution network (GCN) has been proved to be able to capture spatial feature of documents. However, the ability of GCN to capture sentence local features and context information is limited. In this paper, we propose a novel method using local features to enhance GCN for text classification. The combination methods based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) (such as Bi-LSTM, C-LSTM and ServeNet) are used to capture local features to enrich feature information, and a weight value is used to adjust the intensity of enhancement. We conducted extensive experiments on 5 benchmark datasets (WSDataset, Ohsumed, R52, R8, 20NG), proving the proposed method significantly outperforms the baseline deep learning methods.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125157489","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 : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642223
B. Zhang, X. Lan, Ye Li, Xi Yu Zhang
In view of the problem of recognition of active motion intention of human upper limb, based on the EMG signal of the upper limb surface, this paper proposes a method of predicting the angle of upper limb joint based on RBF neural network. The motion intention of shoulder joint, elbow joint and wrist joint in sagittal plane of human body is predicted and recognized effectively. The simulation results show that the RBF method proposed in this paper can better predict the angle of the upper limb, and verified that the RBF neural network method proposed in this paper can improve the accuracy of the angle prediction of the upper limb joint, which lays the algorithm framework and theoretical foundation for the human-computer interaction control of the upper limb rehabilitation robot.
{"title":"A novel RBF neural network based recognition of human upper limb active motion intention","authors":"B. Zhang, X. Lan, Ye Li, Xi Yu Zhang","doi":"10.1109/ICICIP53388.2021.9642223","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642223","url":null,"abstract":"In view of the problem of recognition of active motion intention of human upper limb, based on the EMG signal of the upper limb surface, this paper proposes a method of predicting the angle of upper limb joint based on RBF neural network. The motion intention of shoulder joint, elbow joint and wrist joint in sagittal plane of human body is predicted and recognized effectively. The simulation results show that the RBF method proposed in this paper can better predict the angle of the upper limb, and verified that the RBF neural network method proposed in this paper can improve the accuracy of the angle prediction of the upper limb joint, which lays the algorithm framework and theoretical foundation for the human-computer interaction control of the upper limb rehabilitation robot.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"70 4","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114130664","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 : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642215
Keyi Chen, Hangjun Che, Xuanhao Yang, Man-Fai Leung
Graph nonnegative matrix factorization (GNMF) is superior for mining the intrinsic geometric structure embedded in high-dimensional data. As the sparsity of the factorized matrices is crucial for clustering, the l0 norm is commonly used in the formulated optimization problem to enforce the sparseness which makes the problem NP-hard and discontinuous. In this paper, the sparse graph nonnegative matrix factorization (SGNMF) is formulated as a global optimization problem by using the sum of inverted Gaussian functions to approximate the l0 norm, the multiplicative update rules are developed to solve the problem with guaranteed convergence. The superior performance of the proposed approach is substantiated by clustering tests on four public datasets.
{"title":"Sparsity-constrained Graph Nonnegative Matrix Factorization for Clustering","authors":"Keyi Chen, Hangjun Che, Xuanhao Yang, Man-Fai Leung","doi":"10.1109/ICICIP53388.2021.9642215","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642215","url":null,"abstract":"Graph nonnegative matrix factorization (GNMF) is superior for mining the intrinsic geometric structure embedded in high-dimensional data. As the sparsity of the factorized matrices is crucial for clustering, the l0 norm is commonly used in the formulated optimization problem to enforce the sparseness which makes the problem NP-hard and discontinuous. In this paper, the sparse graph nonnegative matrix factorization (SGNMF) is formulated as a global optimization problem by using the sum of inverted Gaussian functions to approximate the l0 norm, the multiplicative update rules are developed to solve the problem with guaranteed convergence. The superior performance of the proposed approach is substantiated by clustering tests on four public datasets.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116030663","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 : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642172
Zheqi Zhu, Yingjia Gao, Shenshen Gu
With the development of artificial intelligence, the utilization of robots based on AI is widespread in our daily life, especially in the area of sports. In the aspect of tennis, collecting tennis balls on the ground after a fierce match or training would be tiresome work, so an automatic tennis ball picking robot becomes useful. Three main aspects should be considered in the research of the tennis ball collection robot: the recognition and localization of tennis balls, path planning for collecting every tennis ball, and the global positioning and navigation of the robot. Firstly, computer vision based on deep learning algorithms has excellent reliability, and the MobileNet-SSD model can be quantized and deployed on Raspberry Pi. Therefore, we choose the MobileNet-SSD model with a monocular camera catching pictures to recognize tennis balls. Secondly, perspective transformation is used to get the precise location of the target tennis ball. We propose a regional traversal algorithm to plan the path to collect as many tennis balls as possible. Thirdly, we utilize ultra-wide-band (UWB) supplemented by triangle centroid methods to locate the robot in a global position. After proper training, the tennis ball collection robot performs well and has excellent potential.
{"title":"Tennis Ball Collection Robot Based on MobileNet-SSD","authors":"Zheqi Zhu, Yingjia Gao, Shenshen Gu","doi":"10.1109/ICICIP53388.2021.9642172","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642172","url":null,"abstract":"With the development of artificial intelligence, the utilization of robots based on AI is widespread in our daily life, especially in the area of sports. In the aspect of tennis, collecting tennis balls on the ground after a fierce match or training would be tiresome work, so an automatic tennis ball picking robot becomes useful. Three main aspects should be considered in the research of the tennis ball collection robot: the recognition and localization of tennis balls, path planning for collecting every tennis ball, and the global positioning and navigation of the robot. Firstly, computer vision based on deep learning algorithms has excellent reliability, and the MobileNet-SSD model can be quantized and deployed on Raspberry Pi. Therefore, we choose the MobileNet-SSD model with a monocular camera catching pictures to recognize tennis balls. Secondly, perspective transformation is used to get the precise location of the target tennis ball. We propose a regional traversal algorithm to plan the path to collect as many tennis balls as possible. Thirdly, we utilize ultra-wide-band (UWB) supplemented by triangle centroid methods to locate the robot in a global position. After proper training, the tennis ball collection robot performs well and has excellent potential.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122953767","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 : 2021-12-03DOI: 10.1109/ICICIP53388.2021.9642179
Yunong Zhang, Jielong Chen, Haosen Lu
Future prediction is a branch of information processing. An attempt to predict some future event year is presented in this work via combining the addition-subtraction frequency (ASF) method, i.e., ASF algorithms with 3 inputs, and multiple mathematical modeling methods (e.g., polynomial curve fitting, exponential curve fitting, and smoothing spline). The 3-input ASF algorithms using full-traversal, equal-half-traversal, and unequal-half-traversal are applied in the numerical experiments. The difficult challenge we face is that the raw data set size is small, i.e., only 4. Thus, we process the limited information in a variety of ways, i.e., we handle the small data set by using multiple methods. We finally predict that 2021, 2022, or 2027 is of relatively high possibility to be a future year of such a small-data sequence. There may be errors of one to two years, and it may be avoided if some proper measures are taken.
{"title":"Predicting Future Event via Small Data (e.g., 4 Data) by ASF and Curve Fitting Methods","authors":"Yunong Zhang, Jielong Chen, Haosen Lu","doi":"10.1109/ICICIP53388.2021.9642179","DOIUrl":"https://doi.org/10.1109/ICICIP53388.2021.9642179","url":null,"abstract":"Future prediction is a branch of information processing. An attempt to predict some future event year is presented in this work via combining the addition-subtraction frequency (ASF) method, i.e., ASF algorithms with 3 inputs, and multiple mathematical modeling methods (e.g., polynomial curve fitting, exponential curve fitting, and smoothing spline). The 3-input ASF algorithms using full-traversal, equal-half-traversal, and unequal-half-traversal are applied in the numerical experiments. The difficult challenge we face is that the raw data set size is small, i.e., only 4. Thus, we process the limited information in a variety of ways, i.e., we handle the small data set by using multiple methods. We finally predict that 2021, 2022, or 2027 is of relatively high possibility to be a future year of such a small-data sequence. There may be errors of one to two years, and it may be avoided if some proper measures are taken.","PeriodicalId":435799,"journal":{"name":"2021 11th International Conference on Intelligent Control and Information Processing (ICICIP)","volume":"2004 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127309747","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：