Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137919
Yu Bai, Shihu Liu, Yi Tang
Graph Convolutional Networks (GCNs) have recently received a lot of attention, owing to their ability to handle graph-structured data. To improve the expressive power of GCNs, several recent studies has concentrated on the stacking of multiple layers, such as convolutional neural networks. However, simply stacking multiple GCN layers will lead to over-fitting and over-smoothing issues. To integrate deeper information and solve the above problems, this paper proposes Multi-Hop Diffusion-Based Graph Convolutional Networks (MD-GCNs), a method for aggregating and stacking multi-hop neighbors of varying orders into one layer, allowing for the capture of long-distance interactions between remote nodes at each layer of GCNs. In order to calculate the weight between neighbor nodes with multi-hop in the same layer, Multi-Hop Diffusion (MD) mechanism introduces the graph diffusion to spread the weight, the receptive field of each layer of GCNs is increased. On this basis, we introduce the MD-GCNs architecture that can be stacked in multiple layers and has the ability to be expressed. Experimental results on node classification tasks in both transductive and inductive learning settings demonstrate the superiority of the proposed method.
{"title":"Multi-Hop Diffusion-Based Graph Convolutional Networks","authors":"Yu Bai, Shihu Liu, Yi Tang","doi":"10.1109/ACAIT56212.2022.10137919","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137919","url":null,"abstract":"Graph Convolutional Networks (GCNs) have recently received a lot of attention, owing to their ability to handle graph-structured data. To improve the expressive power of GCNs, several recent studies has concentrated on the stacking of multiple layers, such as convolutional neural networks. However, simply stacking multiple GCN layers will lead to over-fitting and over-smoothing issues. To integrate deeper information and solve the above problems, this paper proposes Multi-Hop Diffusion-Based Graph Convolutional Networks (MD-GCNs), a method for aggregating and stacking multi-hop neighbors of varying orders into one layer, allowing for the capture of long-distance interactions between remote nodes at each layer of GCNs. In order to calculate the weight between neighbor nodes with multi-hop in the same layer, Multi-Hop Diffusion (MD) mechanism introduces the graph diffusion to spread the weight, the receptive field of each layer of GCNs is increased. On this basis, we introduce the MD-GCNs architecture that can be stacked in multiple layers and has the ability to be expressed. Experimental results on node classification tasks in both transductive and inductive learning settings demonstrate the superiority of the proposed method.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127191597","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137858
Chuanyun Xu, Yu Zheng, Yang Zhang, Chengjie Sun, Gang Li, Zhaohan Zhu
As real-world data grows fast, the problem of data imbalance has become more prominent. Thus the long-tail problem in deep learning has received lots of attention recently. One of the solutions is to apply a class rebalancing strategy, such as directly using the inverse of the class sample size for reweighting. In past studies, the setting of weights only relates to the number of class samples. Only relying on the information of the number of class samples to determine the size of the weight is very crude in the sensitive method of re-weighting. In this paper, we implement adaptive re-weighting for three essential attributes of the dataset considering several factors: the number of classes, the number of samples, and the degree of class imbalance. We conducted experiments on the commonly used sample imbalance problem solution and proposed a new sample reweighting method. Specifically, a novel re-weighting idea is proposed to optimize Class-Balanced Loss Based on an Effective Number of Samples. Experiments show that the method is superior in re-weighting imbalanced datasets on deep neural networks. We hope our work will stimulate a rethinking of the number-of-samples-based convention in re-weighting.
{"title":"Adaptive Class-Balanced Loss Based on Re-Weighting","authors":"Chuanyun Xu, Yu Zheng, Yang Zhang, Chengjie Sun, Gang Li, Zhaohan Zhu","doi":"10.1109/ACAIT56212.2022.10137858","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137858","url":null,"abstract":"As real-world data grows fast, the problem of data imbalance has become more prominent. Thus the long-tail problem in deep learning has received lots of attention recently. One of the solutions is to apply a class rebalancing strategy, such as directly using the inverse of the class sample size for reweighting. In past studies, the setting of weights only relates to the number of class samples. Only relying on the information of the number of class samples to determine the size of the weight is very crude in the sensitive method of re-weighting. In this paper, we implement adaptive re-weighting for three essential attributes of the dataset considering several factors: the number of classes, the number of samples, and the degree of class imbalance. We conducted experiments on the commonly used sample imbalance problem solution and proposed a new sample reweighting method. Specifically, a novel re-weighting idea is proposed to optimize Class-Balanced Loss Based on an Effective Number of Samples. Experiments show that the method is superior in re-weighting imbalanced datasets on deep neural networks. We hope our work will stimulate a rethinking of the number-of-samples-based convention in re-weighting.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132207663","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137942
Wanting Wu, Jishan Piao
The classification and prediction of enterprise financial data can improve the cost and benefit optimization management level of enterprises. In order to improve the ability of enterprise financial data classification, an optimization algorithm of enterprise financial data classification based on decision tree is proposed. The global data model is adopted to establish the management model of enterprise financial database. Based on the heterogeneous parameters among enterprise financial data sources, combined with the structural feature analysis of data sources, the characteristic analysis method of dynamic allocation and correlation constraints of resources such as human, material and financial resources is adopted to establish the allocation model of influencing factors of enterprise financial data. Based on the decision tree classification algorithm, the correlation features of cost and income of enterprise financial data are extracted. According to the pattern change of compliance management income, cluster analysis and pattern recognition of expected income dynamic characteristics of enterprise financial data are realized. By constructing a dynamic allocation model of enterprise financial data and enterprise financial cost and income, cash flow data analysis method is adopted, according to quantitative parameter analysis of realtime operating cash flow, semantic similarity measurement method is adopted, and based on online observation data cleaning, correlation characteristics recognition and cluster analysis of enterprise financial data cost and income are realized, and enterprise financial data is optimally classified. The empirical analysis and simulation results show that this method is highly reliable in classifying enterprise financial data, and has strong ability to dynamically allocate resources such as manpower, material resources and financial resources and control income and cost, thus improving the quality level of enterprise financial data management.
{"title":"Research on Optimization Algorithm of Enterprise Financial Data Classification Based on Decision Tree","authors":"Wanting Wu, Jishan Piao","doi":"10.1109/ACAIT56212.2022.10137942","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137942","url":null,"abstract":"The classification and prediction of enterprise financial data can improve the cost and benefit optimization management level of enterprises. In order to improve the ability of enterprise financial data classification, an optimization algorithm of enterprise financial data classification based on decision tree is proposed. The global data model is adopted to establish the management model of enterprise financial database. Based on the heterogeneous parameters among enterprise financial data sources, combined with the structural feature analysis of data sources, the characteristic analysis method of dynamic allocation and correlation constraints of resources such as human, material and financial resources is adopted to establish the allocation model of influencing factors of enterprise financial data. Based on the decision tree classification algorithm, the correlation features of cost and income of enterprise financial data are extracted. According to the pattern change of compliance management income, cluster analysis and pattern recognition of expected income dynamic characteristics of enterprise financial data are realized. By constructing a dynamic allocation model of enterprise financial data and enterprise financial cost and income, cash flow data analysis method is adopted, according to quantitative parameter analysis of realtime operating cash flow, semantic similarity measurement method is adopted, and based on online observation data cleaning, correlation characteristics recognition and cluster analysis of enterprise financial data cost and income are realized, and enterprise financial data is optimally classified. The empirical analysis and simulation results show that this method is highly reliable in classifying enterprise financial data, and has strong ability to dynamically allocate resources such as manpower, material resources and financial resources and control income and cost, thus improving the quality level of enterprise financial data management.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"100 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132225307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137953
Dongli Wang, JiaLiu, Yan Zhou
Dynamic modeling of different dimensional features in video is the key element of group activity recognition. In the past years, a lot of work has been devoted to the modeling of character features, these methods have achieved good results, but most of them ignore that group activity is a continuous motion closely related to the scene, and underestimated the importance of the relationship between frames. This paper proposes a Person-Frame Dynamic Feature Graph Network to model group activity information from two levels: video frame level and individual level: Temporal Semantic sub-Graph (TSG) channel constructs temporal semantic relation subgraph for video frame features, and Person-level Dynamic Feature Map (PDFM) models personal dynamic characteristics. In addition, in order to alleviate the problem of slow training speed of group activity model, we use lightweight mobilenet-v2 as the backbone, and embed the Initial Feature Preprocessing Module (IFPM) in it to improve the training efficiency while maintaining the recognition accuracy. A lot of experiments have been done on this model with the most widely used dataset in the field of group activity recognition, and excellent results are obtained, which proves the effectiveness of the model.
{"title":"Person-Frame Dynamic Feature Graph Network for Group Activity Recognition","authors":"Dongli Wang, JiaLiu, Yan Zhou","doi":"10.1109/ACAIT56212.2022.10137953","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137953","url":null,"abstract":"Dynamic modeling of different dimensional features in video is the key element of group activity recognition. In the past years, a lot of work has been devoted to the modeling of character features, these methods have achieved good results, but most of them ignore that group activity is a continuous motion closely related to the scene, and underestimated the importance of the relationship between frames. This paper proposes a Person-Frame Dynamic Feature Graph Network to model group activity information from two levels: video frame level and individual level: Temporal Semantic sub-Graph (TSG) channel constructs temporal semantic relation subgraph for video frame features, and Person-level Dynamic Feature Map (PDFM) models personal dynamic characteristics. In addition, in order to alleviate the problem of slow training speed of group activity model, we use lightweight mobilenet-v2 as the backbone, and embed the Initial Feature Preprocessing Module (IFPM) in it to improve the training efficiency while maintaining the recognition accuracy. A lot of experiments have been done on this model with the most widely used dataset in the field of group activity recognition, and excellent results are obtained, which proves the effectiveness of the model.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132884501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137826
Lingyue Kong, Kun Jiang, Yuanda Wang
Accurate prediction of pedestrian trajectory is crucial for the autonomous driving system and service robots. In this paper, we further analyze the pedestrian interaction patterns and propose a novel model, named GL-Net, based on the graph structure with two encoders and one decoder. Our model first formulates the short-term spatio-temporal interaction between pedestrians within a single frame by the single sequence encoder. In this module, we utilize a graph attention network (GAT) and a graph-based transformer in parallel to extract both local and global spatial interaction features respectively. A set of candidate trajectories are then generated by the long sequence encoder, which can extract entire temporal dependence in historical pedestrian trajectory and Figure out long-term pedestrian intention. To rectify the inherent uncertainty caused by the multimodal nature, we introduce a Gaussian noise to our spatio-temporal embedding. Evaluations of ETH and UCY datasets show that our model achieves better performance than the previous graph-based models. Moreover, our model produces more reasonable trajectories at the point of social interaction and has a better balance of capturing spatial interaction features and generating temporal sequences than other models.
{"title":"Transformer with Global and Local Interaction for Pedestrian Trajectory Prediction","authors":"Lingyue Kong, Kun Jiang, Yuanda Wang","doi":"10.1109/ACAIT56212.2022.10137826","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137826","url":null,"abstract":"Accurate prediction of pedestrian trajectory is crucial for the autonomous driving system and service robots. In this paper, we further analyze the pedestrian interaction patterns and propose a novel model, named GL-Net, based on the graph structure with two encoders and one decoder. Our model first formulates the short-term spatio-temporal interaction between pedestrians within a single frame by the single sequence encoder. In this module, we utilize a graph attention network (GAT) and a graph-based transformer in parallel to extract both local and global spatial interaction features respectively. A set of candidate trajectories are then generated by the long sequence encoder, which can extract entire temporal dependence in historical pedestrian trajectory and Figure out long-term pedestrian intention. To rectify the inherent uncertainty caused by the multimodal nature, we introduce a Gaussian noise to our spatio-temporal embedding. Evaluations of ETH and UCY datasets show that our model achieves better performance than the previous graph-based models. Moreover, our model produces more reasonable trajectories at the point of social interaction and has a better balance of capturing spatial interaction features and generating temporal sequences than other models.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"22 5","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113964319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137965
Shengnan Wang
BIM construction of prefabricated buildings is the key facility to ensure the safety of prefabricated buildings. In order to improve BIM construction technology and schedule optimization control performance of prefabricated buildings, an optimization control method of BIM construction schedule based on improved differential evolution algorithm is proposed. ANSYS software analysis model is used to build the constraint parameter model of BIM construction schedule optimization control of prefabricated buildings, and the distribution model of management experience and technical ability related to prefabricated buildings is built. Combined with technical system, construction technology and technical level, the construction technology optimization parameter design of prefabricated buildings BIM is carried out, and the construction cost of BIM construction schedule management of prefabricated buildings is established by improved differential evolution algorithm. The information management structure model of unit economic benefits, financing risks, production, transportation and installation costs, taxes and fees, etc. is built, and the improved differential evolution algorithm is adopted to realize the optimal control of BIM construction process of prefabricated buildings. The simulation results show that this method can be used to optimize and control the BIM construction technology of prefabricated buildings, improve the construction quality, and the optimization ability of each parameter describing the construction quality is good, and the process control quality is high, which has a good guiding role in promoting the BIM construction quality optimization of prefabricated buildings.
{"title":"BIM Construction Schedule Optimization of Prefabricated Buildings Based on Improved Differential Evolution Algorithm","authors":"Shengnan Wang","doi":"10.1109/ACAIT56212.2022.10137965","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137965","url":null,"abstract":"BIM construction of prefabricated buildings is the key facility to ensure the safety of prefabricated buildings. In order to improve BIM construction technology and schedule optimization control performance of prefabricated buildings, an optimization control method of BIM construction schedule based on improved differential evolution algorithm is proposed. ANSYS software analysis model is used to build the constraint parameter model of BIM construction schedule optimization control of prefabricated buildings, and the distribution model of management experience and technical ability related to prefabricated buildings is built. Combined with technical system, construction technology and technical level, the construction technology optimization parameter design of prefabricated buildings BIM is carried out, and the construction cost of BIM construction schedule management of prefabricated buildings is established by improved differential evolution algorithm. The information management structure model of unit economic benefits, financing risks, production, transportation and installation costs, taxes and fees, etc. is built, and the improved differential evolution algorithm is adopted to realize the optimal control of BIM construction process of prefabricated buildings. The simulation results show that this method can be used to optimize and control the BIM construction technology of prefabricated buildings, improve the construction quality, and the optimization ability of each parameter describing the construction quality is good, and the process control quality is high, which has a good guiding role in promoting the BIM construction quality optimization of prefabricated buildings.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124175859","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 similarity evaluation method based on neural network has achieved good results, but it has higher requirements on the scale and quality of the corpus. Based on this problem, this paper proposes a semantic similarity evaluation method based on text generation data augmentation. This method combines Seq2Seq with a masked language model for data augmentation, and uses the expanded data to fine-tune the pre-trained language model. The pre-trained language model and the Siamese network are combined to build a semantic similarity evaluation model. Finally, experiments were carried out on the standard sentence similarity evaluation data set SentEva12012-2016. Compared with the benchmark model, the Spearman correlation coefficient improved by 3.11%. Experiments show that the semantic similarity evaluation method based on data augmentation can effectively solve the problem of low accuracy due to lack of data.
{"title":"Semantic Similarity Evaluation Method Based on Text Generation Data Augmentation","authors":"Jiangfeng Zhou, Dafei Lin, Xinlai Xing, Xiaochuan Zhang","doi":"10.1109/ACAIT56212.2022.10137987","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137987","url":null,"abstract":"The similarity evaluation method based on neural network has achieved good results, but it has higher requirements on the scale and quality of the corpus. Based on this problem, this paper proposes a semantic similarity evaluation method based on text generation data augmentation. This method combines Seq2Seq with a masked language model for data augmentation, and uses the expanded data to fine-tune the pre-trained language model. The pre-trained language model and the Siamese network are combined to build a semantic similarity evaluation model. Finally, experiments were carried out on the standard sentence similarity evaluation data set SentEva12012-2016. Compared with the benchmark model, the Spearman correlation coefficient improved by 3.11%. Experiments show that the semantic similarity evaluation method based on data augmentation can effectively solve the problem of low accuracy due to lack of data.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121288914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137956
Jin Shen, Lin Mei, Yating Sun
Aiming at the low efficiency and accuracy of K-means algorithm in processing massive data, an improved K-means clustering algorithm based on dissimilarity function was proposed. The Euclidean distance internal weighted method was used to improve the traditional distance algorithm, and a new dissimilarity function was constructed to calculate the distance of the cluster center. Experimental results showed that compared with the traditional K-means clustering algorithm, the improved K-means clustering algorithm has a faster convergence speed and higher accuracy in the algorithm verification. In practical applications, after cluster analysis is performed on the proportion of page access times, more accurate user consumption behavior characteristics are obtained. Therefore, based on the improved K-means clustering algorithm, the consumption behavior characteristics of business users can be described and analyzed well.
{"title":"Analysis of Consumption Behavior Characteristics of Business Users Based on Dissimilarity Function Improved K-Means Clustering Algorithm","authors":"Jin Shen, Lin Mei, Yating Sun","doi":"10.1109/ACAIT56212.2022.10137956","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137956","url":null,"abstract":"Aiming at the low efficiency and accuracy of K-means algorithm in processing massive data, an improved K-means clustering algorithm based on dissimilarity function was proposed. The Euclidean distance internal weighted method was used to improve the traditional distance algorithm, and a new dissimilarity function was constructed to calculate the distance of the cluster center. Experimental results showed that compared with the traditional K-means clustering algorithm, the improved K-means clustering algorithm has a faster convergence speed and higher accuracy in the algorithm verification. In practical applications, after cluster analysis is performed on the proportion of page access times, more accurate user consumption behavior characteristics are obtained. Therefore, based on the improved K-means clustering algorithm, the consumption behavior characteristics of business users can be described and analyzed well.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116485058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137932
Chongjun Zhao
Aiming at the problem of translation accuracy of traditional auxiliary translation software, this paper proposed to construct an auxiliary translation model based on logarithmic position representation and self-attention. This model used the self-attention mechanism (SA) to capture the semantic relevance of contextual words. Then, the distance information and direction information between words were retained by the logarithmic position representation (LPR), so as to improve the translation accuracy of the model. Experimental results showed that the BLEU score of the proposed model is 31.59, which is 8.04 and 3.65 higher than that of GNMT RL model and existing SOTA model, respectively. In English-French machine translation task, the BLEU score of the proposed model is 42.98, which is higher than that of the other two models. Therefore, the deep learning machine translation model constructed in this paper has higher accuracy and can improve the efficiency of machine translation.
{"title":"Aided Translation Model Based on Logarithmic Position Representation Method and Self-Attention Mechanism","authors":"Chongjun Zhao","doi":"10.1109/ACAIT56212.2022.10137932","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137932","url":null,"abstract":"Aiming at the problem of translation accuracy of traditional auxiliary translation software, this paper proposed to construct an auxiliary translation model based on logarithmic position representation and self-attention. This model used the self-attention mechanism (SA) to capture the semantic relevance of contextual words. Then, the distance information and direction information between words were retained by the logarithmic position representation (LPR), so as to improve the translation accuracy of the model. Experimental results showed that the BLEU score of the proposed model is 31.59, which is 8.04 and 3.65 higher than that of GNMT RL model and existing SOTA model, respectively. In English-French machine translation task, the BLEU score of the proposed model is 42.98, which is higher than that of the other two models. Therefore, the deep learning machine translation model constructed in this paper has higher accuracy and can improve the efficiency of machine translation.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116092817","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137948
Lijia Yang
In order to improve the scheduling ability of urban cold chain multi-series distributed logistics, it is necessary to carry out path optimization planning and design. This paper puts forward the shortest path optimization planning algorithm of urban cold chain multi-series distributed logistics based on particle swarm optimization. The particle swarm optimization method is adopted to sample the environmental information of urban cold chain multi-serial point distributed logistics area, the collected data of urban cold chain multi-serial point distributed logistics area is dynamically weighted and the shortest path optimization control is carried out, and the path space area grid block planning detection model of urban cold chain multi-serial point distributed logistics area is established. According to the task requirements, Particle swarm optimization (PSO) shortest path detection method is used to optimize the shortest path planning and block search of urban cold chain multi-series distributed logistics. The pheromone features of the shortest path planning of urban cold chain multi-series distributed logistics are extracted. The shortest path planning method is used to analyze the characteristics of urban cold chain multi-series distributed logistics, and the global evolution game features of logistics trolley are analyzed. Particle swarm optimization (PSO) algorithm is used to carry out adaptive optimization in the shortest path planning process of urban cold chain multi-series distributed logistics, so as to realize independent planning and shortest optimization of the global path of urban cold chain multi-series distributed logistics. The simulation results show that the shortest path planning of urban cold chain multi-series distributed logistics with this method has good optimization ability, which improves the response ability of urban cold chain multi-series distributed logistics and reduces the cost of distribution time.
{"title":"Self-Planning Method for Global Path of Logistics Trolley Considering Task Requirements","authors":"Lijia Yang","doi":"10.1109/ACAIT56212.2022.10137948","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137948","url":null,"abstract":"In order to improve the scheduling ability of urban cold chain multi-series distributed logistics, it is necessary to carry out path optimization planning and design. This paper puts forward the shortest path optimization planning algorithm of urban cold chain multi-series distributed logistics based on particle swarm optimization. The particle swarm optimization method is adopted to sample the environmental information of urban cold chain multi-serial point distributed logistics area, the collected data of urban cold chain multi-serial point distributed logistics area is dynamically weighted and the shortest path optimization control is carried out, and the path space area grid block planning detection model of urban cold chain multi-serial point distributed logistics area is established. According to the task requirements, Particle swarm optimization (PSO) shortest path detection method is used to optimize the shortest path planning and block search of urban cold chain multi-series distributed logistics. The pheromone features of the shortest path planning of urban cold chain multi-series distributed logistics are extracted. The shortest path planning method is used to analyze the characteristics of urban cold chain multi-series distributed logistics, and the global evolution game features of logistics trolley are analyzed. Particle swarm optimization (PSO) algorithm is used to carry out adaptive optimization in the shortest path planning process of urban cold chain multi-series distributed logistics, so as to realize independent planning and shortest optimization of the global path of urban cold chain multi-series distributed logistics. The simulation results show that the shortest path planning of urban cold chain multi-series distributed logistics with this method has good optimization ability, which improves the response ability of urban cold chain multi-series distributed logistics and reduces the cost of distribution time.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123709630","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：