Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005475
Yan Gao, Qiang Wang, Liang Cai
In the traditional Byzantine consensus schemes, the efficiency and throughput of the whole topology will be limited by transmission delay in nodes, which depends on the complexity of message and bandwidth constraints. This paper proposes a Byzantine consensus scheme based on weight modified Treap topology that reduces the complexity and delays of message aggregation by quickly constructing and maintaining Treap node tree. We have the master node organize other active nodes into a balanced tree rooted at itself to allocate communication and computing costs. We propose a failure detection mechanism for Treap that makes the master node record the weights of the nodes for each of its direct child nodes. The master node establishes a delay function to set the weight of each node based on the delay between receiving PREPARE message and completing its own calculation to sending its share. With a node's weight over threshold, it is assumed that the node has failed and been marked. The failure detection mechanism of the marked nodes starts to maintain availability by replacing the replication of the marked nodes. It also prevents the parent node from doing evil by authorizing nodes to mark only its direct child node. We only update the Treap topology on a regular basis or in case of failure to reduce the update overhead.
{"title":"Byzantine Consensus Based on Modified Treap Topology","authors":"Yan Gao, Qiang Wang, Liang Cai","doi":"10.1109/ICSAI57119.2022.10005475","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005475","url":null,"abstract":"In the traditional Byzantine consensus schemes, the efficiency and throughput of the whole topology will be limited by transmission delay in nodes, which depends on the complexity of message and bandwidth constraints. This paper proposes a Byzantine consensus scheme based on weight modified Treap topology that reduces the complexity and delays of message aggregation by quickly constructing and maintaining Treap node tree. We have the master node organize other active nodes into a balanced tree rooted at itself to allocate communication and computing costs. We propose a failure detection mechanism for Treap that makes the master node record the weights of the nodes for each of its direct child nodes. The master node establishes a delay function to set the weight of each node based on the delay between receiving PREPARE message and completing its own calculation to sending its share. With a node's weight over threshold, it is assumed that the node has failed and been marked. The failure detection mechanism of the marked nodes starts to maintain availability by replacing the replication of the marked nodes. It also prevents the parent node from doing evil by authorizing nodes to mark only its direct child node. We only update the Treap topology on a regular basis or in case of failure to reduce the update overhead.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129140060","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005421
Liu Jin, Zhaochun Sun, Huifang Ma
In some fields such as e-commerce and social media platforms and sentiment analysis, efficient short text classification is crucial to enable users to locate pertinent information effectively. Along with the increasing number of short texts, classifying short texts with brief contents and sparse features has become a major research topic in recent years. Towards this end, a short text classification method based on a dual channel hypergraph convolutional network is proposed to flexibly capture the complex higher-order relationships among short texts and words. Specifically, our method firstly models the pre-processed short text data into short text hypergraph and short text association graph; secondly, two different short text feature representations are learned via a dual channel hypergraph convolutional network and fused by an attention network to enhance the short text embedding; at last, a classification model is adopted to perform short text classification. Extensive experimental results indicate that the method has superior short text classification effect and stability compared with the existing model, which has better performance among comparable short text classification models.
{"title":"Short text classification method with dual channel hypergraph convolution networks","authors":"Liu Jin, Zhaochun Sun, Huifang Ma","doi":"10.1109/ICSAI57119.2022.10005421","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005421","url":null,"abstract":"In some fields such as e-commerce and social media platforms and sentiment analysis, efficient short text classification is crucial to enable users to locate pertinent information effectively. Along with the increasing number of short texts, classifying short texts with brief contents and sparse features has become a major research topic in recent years. Towards this end, a short text classification method based on a dual channel hypergraph convolutional network is proposed to flexibly capture the complex higher-order relationships among short texts and words. Specifically, our method firstly models the pre-processed short text data into short text hypergraph and short text association graph; secondly, two different short text feature representations are learned via a dual channel hypergraph convolutional network and fused by an attention network to enhance the short text embedding; at last, a classification model is adopted to perform short text classification. Extensive experimental results indicate that the method has superior short text classification effect and stability compared with the existing model, which has better performance among comparable short text classification models.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116339171","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}
Existing classroom behavior detection methods for students are mainly based on the network model to extract key common features to directly determine behavior types, which cannot provide a higher fine-grained understanding of interaction relationships in the classroom. This paper proposes a classroom behavior detection method for students based on the Human-Object Interaction (HOI) model, which further utilizes human-object relationship features to infer interaction relationships based on object detection. In the study, the cell phone is selected as the detected object to interact with the students, and the HOI model is trained and tested for two types of behaviors—Use and No interaction. The results show that the average accuracy of the trained HOI model reaches about 83.4% in the test, which promotes a higher fine-grained perception and understanding of classroom behavior detection and provides a new perspective for building smart classrooms and exploring personalized teaching and learning paths.
{"title":"Students’ Classroom Behavior Detection Based on Human-Object Interaction Model","authors":"Yonghe Zhang, Wenjiao Qu, Guocheng Zhong, Yundan Xiao","doi":"10.1109/ICSAI57119.2022.10005457","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005457","url":null,"abstract":"Existing classroom behavior detection methods for students are mainly based on the network model to extract key common features to directly determine behavior types, which cannot provide a higher fine-grained understanding of interaction relationships in the classroom. This paper proposes a classroom behavior detection method for students based on the Human-Object Interaction (HOI) model, which further utilizes human-object relationship features to infer interaction relationships based on object detection. In the study, the cell phone is selected as the detected object to interact with the students, and the HOI model is trained and tested for two types of behaviors—Use and No interaction. The results show that the average accuracy of the trained HOI model reaches about 83.4% in the test, which promotes a higher fine-grained perception and understanding of classroom behavior detection and provides a new perspective for building smart classrooms and exploring personalized teaching and learning paths.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"141 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125950994","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005455
Boran Wang, Minghao Gao
There has been a recent interest in employing reinforcement learning for training end-to-end goal-driven robot navigation policies. However, implementing reinforcement learning in end-to-end navigation may result in inefficient policies that exhibit redundant turning actions when attempting to avoid obstacles. This work proposes a two branches network to learn efficient policies with less turning action when robots cross the obstacles. We first employ supervised learning to train a robot action classification network with optical flow. We then combine this classifier with an RGBD optical encoder to develop an action-decision network. Ultimately, we evaluate our approach in a visually realistic simulation environment. The results show that our method can reduce unnecessary steering actions and improve efficiency while ensuring navigation capabilities. We further show that our approach can reduce energy consumption during navigation and extend the robot's work time. Experiment results in the iGibson® simulator over hand-made paths reveal that our method can reduce 13.1% of the action number in the training set and 12.9% in the testing set compared with the baseline approaches. It also can reduce 8.3% energy consumption in the training set and 9.6% in the testing set and only has a 4.2% and 8.1% difference compared with the human path.
{"title":"End-to-End Efficient Indoor Navigation with Optical Flow","authors":"Boran Wang, Minghao Gao","doi":"10.1109/ICSAI57119.2022.10005455","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005455","url":null,"abstract":"There has been a recent interest in employing reinforcement learning for training end-to-end goal-driven robot navigation policies. However, implementing reinforcement learning in end-to-end navigation may result in inefficient policies that exhibit redundant turning actions when attempting to avoid obstacles. This work proposes a two branches network to learn efficient policies with less turning action when robots cross the obstacles. We first employ supervised learning to train a robot action classification network with optical flow. We then combine this classifier with an RGBD optical encoder to develop an action-decision network. Ultimately, we evaluate our approach in a visually realistic simulation environment. The results show that our method can reduce unnecessary steering actions and improve efficiency while ensuring navigation capabilities. We further show that our approach can reduce energy consumption during navigation and extend the robot's work time. Experiment results in the iGibson® simulator over hand-made paths reveal that our method can reduce 13.1% of the action number in the training set and 12.9% in the testing set compared with the baseline approaches. It also can reduce 8.3% energy consumption in the training set and 9.6% in the testing set and only has a 4.2% and 8.1% difference compared with the human path.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"213 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133734723","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005476
Yi Zhang, Liang Zhou
The prevalence of GPS positioning software has led to the generation of massive trajectory data, so it is necessary to take measures to compress the data. This paper proposes an efficient trajectory simplification algorithm based on Essential Inflection Point Extraction (EIPE). EIPE algorithm adopts synchronous Euclidean distance and direction angle deviation to preserve the temporal-spatial information while compressing the trajectories. However, considering the direction angle deviation will improve the compression rate for trajectories with high randomness in direction. To address this issue, we set a range threshold for the EIPE algorithm to extract essential inflection points from trajectories. The evaluation results on two real-life datasets indicate that our algorithm can improve compression efficiency and achieve satisfactory performance on both average direction angle deviation error and running time.
{"title":"An Efficient Approach for Trajectory Simplification Based on Essential Inflection Point Extraction","authors":"Yi Zhang, Liang Zhou","doi":"10.1109/ICSAI57119.2022.10005476","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005476","url":null,"abstract":"The prevalence of GPS positioning software has led to the generation of massive trajectory data, so it is necessary to take measures to compress the data. This paper proposes an efficient trajectory simplification algorithm based on Essential Inflection Point Extraction (EIPE). EIPE algorithm adopts synchronous Euclidean distance and direction angle deviation to preserve the temporal-spatial information while compressing the trajectories. However, considering the direction angle deviation will improve the compression rate for trajectories with high randomness in direction. To address this issue, we set a range threshold for the EIPE algorithm to extract essential inflection points from trajectories. The evaluation results on two real-life datasets indicate that our algorithm can improve compression efficiency and achieve satisfactory performance on both average direction angle deviation error and running time.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"147 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132776250","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005351
Shanwu Shao, Ji Li, Ping Shao, Xiangyuan Zhu
Image encryption can protect cloud image privacy. However, the usability of encrypted images has not received much attention. To this end, an image encryption scheme with an associated thumbnail is proposed, it generates both the encrypted image and the associated thumbnail, which can be stored in the cloud. The associated thumbnail is only partially encrypted, which has a small capacity, and maintains certain usability with less download and decryption time. Therefore, the users can download and decrypt the associated thumbnail instead of the encrypted image to meet general usage requirements, and most of the time there is no need to download the encrypted image from the cloud. Experiments show that this scheme balances image privacy protection and usability, and can bring convenience to the browsing and management of encrypted images in the cloud.
{"title":"An Image Encryption Scheme with the Associated Thumbnail","authors":"Shanwu Shao, Ji Li, Ping Shao, Xiangyuan Zhu","doi":"10.1109/ICSAI57119.2022.10005351","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005351","url":null,"abstract":"Image encryption can protect cloud image privacy. However, the usability of encrypted images has not received much attention. To this end, an image encryption scheme with an associated thumbnail is proposed, it generates both the encrypted image and the associated thumbnail, which can be stored in the cloud. The associated thumbnail is only partially encrypted, which has a small capacity, and maintains certain usability with less download and decryption time. Therefore, the users can download and decrypt the associated thumbnail instead of the encrypted image to meet general usage requirements, and most of the time there is no need to download the encrypted image from the cloud. Experiments show that this scheme balances image privacy protection and usability, and can bring convenience to the browsing and management of encrypted images in the cloud.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132978222","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005501
Jieru Ding, Min Wang, Xinghui Wu, Zhiyi Wang
Automotive radar plays a significant role in un-manned auto-drive system, and most vehicle-mounted radars improve the angular resolution by the MIMO radar. Two-dimension (2D) fast Fourier transform (FFT) is usually used to extract the range frequency and Doppler frequency. When there is few sampling points in the observed signal, imaging results of range-Doppler rapidly deteriorates. In this paper, we exploit the sparsity of scattering points in space and the robustness of l1 norm, to finish the super-resolution imaging of range-Doppler (RD) map. l1 is employed to update the sparse result by introducing the Lagrange multiplier. Finally, the algorithm has been validated by the simulated data, and it has demonstrated the algorithm’s effectiveness.
{"title":"Two-dimension Super-resolution Range Doppler Imaging in Automotive Radar","authors":"Jieru Ding, Min Wang, Xinghui Wu, Zhiyi Wang","doi":"10.1109/ICSAI57119.2022.10005501","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005501","url":null,"abstract":"Automotive radar plays a significant role in un-manned auto-drive system, and most vehicle-mounted radars improve the angular resolution by the MIMO radar. Two-dimension (2D) fast Fourier transform (FFT) is usually used to extract the range frequency and Doppler frequency. When there is few sampling points in the observed signal, imaging results of range-Doppler rapidly deteriorates. In this paper, we exploit the sparsity of scattering points in space and the robustness of l1 norm, to finish the super-resolution imaging of range-Doppler (RD) map. l1 is employed to update the sparse result by introducing the Lagrange multiplier. Finally, the algorithm has been validated by the simulated data, and it has demonstrated the algorithm’s effectiveness.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124923972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005509
Jiaye Lin, Yanjie Liu
Compared to the vision-based approach, LiDAR-based SLAM has shown a great advantage in depicting geometric characteristics but still suffers from accumulated localization errors during long-term operation in large-scale scenarios. Introducing semantic information to the current system helps to discover higher-level features and establish a stronger association of features in different frames. In this paper, we utilize semantic information to present an integral LiDAR odometry that combines adaptive downsampling feature with label-specified registration to boost the performance of odometry estimation, together with Scan Context as the loop closure module to constrain the amplification of cumulative errors. Experiments are conducted based on the well-known KITTI dataset, which reveals that the proposed framework achieves higher accuracy with an average RTE of 0.97% in real-time and shows great robustness toward various scenarios.
{"title":"Semantic Assisted LiDAR Odometry with Loop Closure in Large Scale Urban Environment","authors":"Jiaye Lin, Yanjie Liu","doi":"10.1109/ICSAI57119.2022.10005509","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005509","url":null,"abstract":"Compared to the vision-based approach, LiDAR-based SLAM has shown a great advantage in depicting geometric characteristics but still suffers from accumulated localization errors during long-term operation in large-scale scenarios. Introducing semantic information to the current system helps to discover higher-level features and establish a stronger association of features in different frames. In this paper, we utilize semantic information to present an integral LiDAR odometry that combines adaptive downsampling feature with label-specified registration to boost the performance of odometry estimation, together with Scan Context as the loop closure module to constrain the amplification of cumulative errors. Experiments are conducted based on the well-known KITTI dataset, which reveals that the proposed framework achieves higher accuracy with an average RTE of 0.97% in real-time and shows great robustness toward various scenarios.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122512914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005401
Quanhong Tian
The purpose of Sequential Recommendation (SR) is to recommend the next commodities that a user wants to buy based on their historical interaction sequence. The current approach for SR focuses only on mining the user’s interest preferences, while they all fail to consider the influence of item prices on users’ purchase decisions and suffer from the data sparsity problem. In this paper, a Multi-channel Contrastive Learning method for SR (MCLSR) is proposed, which can effectively extract users’ interest preferences and price preferences and alleviate the sparsity issues. Specifically, first, a heterogeneous knowledge graph is constructed from all interaction sequence and the item attribute (i.e., item price and category) by us. Then, we leverage a heterogeneous graph neural network mechanism to learn user, item, and price node embeddings. Next, users’ price preferences and interest preferences are extracted by an attention network. Finally, a multi-channel contrastive learning mechanism is employed to build price and interest preferences’ relations and generate high-quality recommendation results. Experiments on both real datasets show that MCLSR obtains more sophisticated performance than the existing baseline.
{"title":"Multi-channel Contrastive Learning for Sequential Recommendation","authors":"Quanhong Tian","doi":"10.1109/ICSAI57119.2022.10005401","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005401","url":null,"abstract":"The purpose of Sequential Recommendation (SR) is to recommend the next commodities that a user wants to buy based on their historical interaction sequence. The current approach for SR focuses only on mining the user’s interest preferences, while they all fail to consider the influence of item prices on users’ purchase decisions and suffer from the data sparsity problem. In this paper, a Multi-channel Contrastive Learning method for SR (MCLSR) is proposed, which can effectively extract users’ interest preferences and price preferences and alleviate the sparsity issues. Specifically, first, a heterogeneous knowledge graph is constructed from all interaction sequence and the item attribute (i.e., item price and category) by us. Then, we leverage a heterogeneous graph neural network mechanism to learn user, item, and price node embeddings. Next, users’ price preferences and interest preferences are extracted by an attention network. Finally, a multi-channel contrastive learning mechanism is employed to build price and interest preferences’ relations and generate high-quality recommendation results. Experiments on both real datasets show that MCLSR obtains more sophisticated performance than the existing baseline.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121269275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-10DOI: 10.1109/ICSAI57119.2022.10005355
Hongxing Zhu, Anmin Zhu
Financia1 data are non-stationary and nonlinear. Machine learning makes it easier to classify financial data than traditional models. With the development of machine learning, improving the accuracy of machine learning models for stock price prediction has gradually become a hot research topic. This paper uses the XGBoost (eXtreme gradient boosting) model and the SVM (support vector machine) model to predict the rising, falling and fluctuating of CSI 300, SSE 50 and CSI 500 stock index futures respectively. Then it constructs the XGBoost-SVM combination model and designs a quantitative investment strategy to trade stock index futures in order to research the effectiveness of the models in quantitative investment strategies. The research shows that the proposed method can stably outperform the benchmark returns by combining the investment strategies of the three-price-trend classifications. The constructed XGBoost-SVM model performs better than the original model. It gets higher returns.
{"title":"Application Research of the XGBoost-SVM Combination Model in Quantitative Investment Strategy","authors":"Hongxing Zhu, Anmin Zhu","doi":"10.1109/ICSAI57119.2022.10005355","DOIUrl":"https://doi.org/10.1109/ICSAI57119.2022.10005355","url":null,"abstract":"Financia1 data are non-stationary and nonlinear. Machine learning makes it easier to classify financial data than traditional models. With the development of machine learning, improving the accuracy of machine learning models for stock price prediction has gradually become a hot research topic. This paper uses the XGBoost (eXtreme gradient boosting) model and the SVM (support vector machine) model to predict the rising, falling and fluctuating of CSI 300, SSE 50 and CSI 500 stock index futures respectively. Then it constructs the XGBoost-SVM combination model and designs a quantitative investment strategy to trade stock index futures in order to research the effectiveness of the models in quantitative investment strategies. The research shows that the proposed method can stably outperform the benchmark returns by combining the investment strategies of the three-price-trend classifications. The constructed XGBoost-SVM model performs better than the original model. It gets higher returns.","PeriodicalId":339547,"journal":{"name":"2022 8th International Conference on Systems and Informatics (ICSAI)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127646210","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：