With the increasing number of vehicles, traffic management has put forward higher requirements for vehicle monitoring, thus the technology of vehicle detection based on drone images has received increasing attention. Firstly, we construct a new vehicle detection data set of 600 drone images so that to solve the vehicle detection tasks in real world. Secondly, aiming at the problem of false detection and missed detection in vehicle detection, the Faster R-CNN is improved by using ResNet and constructing Feature Pyramid Networks (FPN) to extract the image features. Finally, based on the vehicle detection data set, the improved Faster R-CNN can be used to detect vehicle targets. The experiment results show that the accuracy of improved method is 96.83%, which is 3.86% higher than that of the original Faster R-CNN method.
{"title":"Vehicle Detection Based on Drone Images with the Improved Faster R-CNN","authors":"Lixin Wang, Junguo Liao, Chaoqian Xu","doi":"10.1145/3318299.3318383","DOIUrl":"https://doi.org/10.1145/3318299.3318383","url":null,"abstract":"With the increasing number of vehicles, traffic management has put forward higher requirements for vehicle monitoring, thus the technology of vehicle detection based on drone images has received increasing attention. Firstly, we construct a new vehicle detection data set of 600 drone images so that to solve the vehicle detection tasks in real world. Secondly, aiming at the problem of false detection and missed detection in vehicle detection, the Faster R-CNN is improved by using ResNet and constructing Feature Pyramid Networks (FPN) to extract the image features. Finally, based on the vehicle detection data set, the improved Faster R-CNN can be used to detect vehicle targets. The experiment results show that the accuracy of improved method is 96.83%, which is 3.86% higher than that of the original Faster R-CNN method.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116439024","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}
This article reviews relevant theories and literature on big data management, management decision-making, execution, and other aspects, discusses the two significant factors of decision-making force and executive power that are the realization of corporate strategic goals, and puts forward the corporate data in the context of big data. The operating model (mainly for the enterprise's decision-making and implementation) faces new opportunities and challenges, that is, through in-depth analysis and exploration of big data management can effectively improve the company's decision-making ability and execution efficiency, and promote the realization of corporate strategic goals.
{"title":"Research on the Application of Big Data Management in Enterprise Management Decision-making and Execution Literature Review","authors":"Zhiyi Zhuo, Shanhu Zhang","doi":"10.1145/3318299.3318388","DOIUrl":"https://doi.org/10.1145/3318299.3318388","url":null,"abstract":"This article reviews relevant theories and literature on big data management, management decision-making, execution, and other aspects, discusses the two significant factors of decision-making force and executive power that are the realization of corporate strategic goals, and puts forward the corporate data in the context of big data. The operating model (mainly for the enterprise's decision-making and implementation) faces new opportunities and challenges, that is, through in-depth analysis and exploration of big data management can effectively improve the company's decision-making ability and execution efficiency, and promote the realization of corporate strategic goals.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"213 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114743465","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 machine learning regression model is based on the assumption of normal distribution. In this paper, we mainly study the probability distribution of the machine learning model and the effect of the convergence values of different loss functions on the probability distribution model. Based on the idea of robust regression and the assumption of homogeneous variance of the model, we solved the statistical solution of two-dimensional regression problem by using least square method. The maximum likelihood estimation parameters of the probabilistic model are obtained by using the maximum likelihood estimation method. In order to compare the solving parameters of the two methods, the convergence values of L1 loss function and L2 loss function are used for the regression verification. Through the mathematical and statistical rigorous derivation, obtained two important conclusions; First, under the condition that the data satisfies normal distribution and is based on the assumption of homogeneous variance, the probability model conforms to the multivariate gaussian distribution. Secondly, the model satisfying the multi-gaussian distribution has little influence on the parameter estimation under the condition of the large number theorem, that is, the multi-gaussian distribution model has good tolerance to the loss function.
{"title":"Model Loss and Distribution Analysis of Regression Problems in Machine Learning","authors":"Nan Yang, Zeyu Zheng, Tianran Wang","doi":"10.1145/3318299.3318367","DOIUrl":"https://doi.org/10.1145/3318299.3318367","url":null,"abstract":"The machine learning regression model is based on the assumption of normal distribution. In this paper, we mainly study the probability distribution of the machine learning model and the effect of the convergence values of different loss functions on the probability distribution model. Based on the idea of robust regression and the assumption of homogeneous variance of the model, we solved the statistical solution of two-dimensional regression problem by using least square method. The maximum likelihood estimation parameters of the probabilistic model are obtained by using the maximum likelihood estimation method. In order to compare the solving parameters of the two methods, the convergence values of L1 loss function and L2 loss function are used for the regression verification. Through the mathematical and statistical rigorous derivation, obtained two important conclusions; First, under the condition that the data satisfies normal distribution and is based on the assumption of homogeneous variance, the probability model conforms to the multivariate gaussian distribution. Secondly, the model satisfying the multi-gaussian distribution has little influence on the parameter estimation under the condition of the large number theorem, that is, the multi-gaussian distribution model has good tolerance to the loss function.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116825324","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}
Some money laundering activities had periodic fund transfer behaviors, and discovering these cyclical behaviors was conducive to narrowing the scope of investigation. This paper treated the capital transaction data as a time series and found each periodic subsequence in the time series through the sub-period discovery algorithm, and designed the tolerance index to improve the robustness of the algorithm. In money laundering activities, there maight be linkage between related accounts. Through the relevant sub-period discovery algorithm, the highly correlated periodic behavior between different accounts were found, and then the suspicious accounts were found. A data set based on police investigation experience is constructed, and on this data set, the algorithm is validated to be effective.
{"title":"Research on the Periodical Behavior Discovery of Funds in Anti-money Laundering Investigation","authors":"Shiliang He, Zhenxin Qu","doi":"10.1145/3318299.3318356","DOIUrl":"https://doi.org/10.1145/3318299.3318356","url":null,"abstract":"Some money laundering activities had periodic fund transfer behaviors, and discovering these cyclical behaviors was conducive to narrowing the scope of investigation. This paper treated the capital transaction data as a time series and found each periodic subsequence in the time series through the sub-period discovery algorithm, and designed the tolerance index to improve the robustness of the algorithm. In money laundering activities, there maight be linkage between related accounts. Through the relevant sub-period discovery algorithm, the highly correlated periodic behavior between different accounts were found, and then the suspicious accounts were found. A data set based on police investigation experience is constructed, and on this data set, the algorithm is validated to be effective.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"169 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126008781","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}
In recent years, the Encoder-Decoder framework has been widely used in image captioning. In the forecast period, many methods regard the input of the usage model at the previous moment as the output at the moment, which may cause the generated words to get worse. This paper proposes to use the correct rate of the preceding words to constrain the weight of the back words, making the loss weight of the back words increase as the preceding word error rate decreases, namely Automatic Constraint Loss (ACL), reducing the difference in the training and test phase. The experimental results on the MSCOCO dataset show that the addition of the proposed method to the original model, the bleu_1 and bleu_2 scores are greatly improved, and the attention mechanism can more accurately select the image region.
{"title":"Image Captioning Based on Automatic Constraint Loss","authors":"Chaoqian Xu, G. Zhu, Lixin Wang","doi":"10.1145/3318299.3318375","DOIUrl":"https://doi.org/10.1145/3318299.3318375","url":null,"abstract":"In recent years, the Encoder-Decoder framework has been widely used in image captioning. In the forecast period, many methods regard the input of the usage model at the previous moment as the output at the moment, which may cause the generated words to get worse. This paper proposes to use the correct rate of the preceding words to constrain the weight of the back words, making the loss weight of the back words increase as the preceding word error rate decreases, namely Automatic Constraint Loss (ACL), reducing the difference in the training and test phase. The experimental results on the MSCOCO dataset show that the addition of the proposed method to the original model, the bleu_1 and bleu_2 scores are greatly improved, and the attention mechanism can more accurately select the image region.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129206129","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}
In this article, we discuss a class of stochastic partial differential systems with nonlinear impulsive and Markovian switching. Some new sufficient conditions proving asymptotic stability in p-th moment of stochastic systems are derived by employing some inequality and the fixed point technique. Some well-known results are generalized and improved.
{"title":"Asymptotic Stability of Nonlinear Impulsive Stochastic Systems with Markovian Switching","authors":"Xinwen Zhang, Chao Jia, Weiguo Liu","doi":"10.1145/3318299.3318304","DOIUrl":"https://doi.org/10.1145/3318299.3318304","url":null,"abstract":"In this article, we discuss a class of stochastic partial differential systems with nonlinear impulsive and Markovian switching. Some new sufficient conditions proving asymptotic stability in p-th moment of stochastic systems are derived by employing some inequality and the fixed point technique. Some well-known results are generalized and improved.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"90 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122816123","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the development of the Internet, the increase of information sources and speed of information release and transmission have led to a sharp increase in the amount of information. To enable users finding more accurate and reliable information in the large heterogeneous multi-source data, data fusion technology becomes more and more important. Data fusion technology structuralizes and integrates heterogeneous data from different sources which greatly improves the comprehensiveness, availability and extensibility of data. This paper proposes a general multi-source data fusion framework. The framework transforms multi-source structured data, semi-structured data and unstructured data into unified data format described by RDF (Resource Description Framework) standard, and then realizes information fusion through data fusion algorithm, to solve the heterogeneity and semantic conflict in multi-source data fusion under the big data environment.
{"title":"A General Multi-Source Data Fusion Framework","authors":"Wei-Ming Liu, Chen Zhang, Bin Yu, Yitong Li","doi":"10.1145/3318299.3318394","DOIUrl":"https://doi.org/10.1145/3318299.3318394","url":null,"abstract":"With the development of the Internet, the increase of information sources and speed of information release and transmission have led to a sharp increase in the amount of information. To enable users finding more accurate and reliable information in the large heterogeneous multi-source data, data fusion technology becomes more and more important. Data fusion technology structuralizes and integrates heterogeneous data from different sources which greatly improves the comprehensiveness, availability and extensibility of data. This paper proposes a general multi-source data fusion framework. The framework transforms multi-source structured data, semi-structured data and unstructured data into unified data format described by RDF (Resource Description Framework) standard, and then realizes information fusion through data fusion algorithm, to solve the heterogeneity and semantic conflict in multi-source data fusion under the big data environment.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"449 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132690588","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}
Synthesizing photorealistic frontal face images from multiple-view profile face images has a wide range of applications in the field of face recognition. However, existing models still have some disadvantages such as high cost and high computational complexity. At present, the Two-Pathway Generative Adversarial Network (TP-GAN) is the state-of-the-art face synthesis model, which can perceive the global structure and local details at the same time. It solves the prier problems but has disadvantages such as training difficulty and lack of diversity of generated samples. Based on Wasserstein GAN with Gradient Penalty (WGAN-GP), this paper proposes a novel Two-Pathway Wasserstein GAN with Gradient Penalty (TPWGAN-GP) model to tackle these defects. TPWGAN-GP uses a gradient penalty method to satisfy the Lipschitz continuity condition, which solves the problems of difficulty in hyper-parameter adjustment and gradient explosion in the TP-GAN, making the convergence speed faster and the model more stable in training process. The generated samples are of higher quality, resulting in more photorealistic faces for recognition tasks.
{"title":"An Improved Face Synthesis Model for Two-Pathway Generative Adversarial Network","authors":"Changlin Li, Zhangjin Huang","doi":"10.1145/3318299.3318346","DOIUrl":"https://doi.org/10.1145/3318299.3318346","url":null,"abstract":"Synthesizing photorealistic frontal face images from multiple-view profile face images has a wide range of applications in the field of face recognition. However, existing models still have some disadvantages such as high cost and high computational complexity. At present, the Two-Pathway Generative Adversarial Network (TP-GAN) is the state-of-the-art face synthesis model, which can perceive the global structure and local details at the same time. It solves the prier problems but has disadvantages such as training difficulty and lack of diversity of generated samples. Based on Wasserstein GAN with Gradient Penalty (WGAN-GP), this paper proposes a novel Two-Pathway Wasserstein GAN with Gradient Penalty (TPWGAN-GP) model to tackle these defects. TPWGAN-GP uses a gradient penalty method to satisfy the Lipschitz continuity condition, which solves the problems of difficulty in hyper-parameter adjustment and gradient explosion in the TP-GAN, making the convergence speed faster and the model more stable in training process. The generated samples are of higher quality, resulting in more photorealistic faces for recognition tasks.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126518713","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}
This article introduces the OsgEarth open source project and the establishment of three-dimensional (3D) cluster situation. On account of multiple nodes and heavy task, the simulation visual effect in the 3D situation is not smooth. Aiming at the problems mentioned above, a multi process service architecture and a dynamic load balancing agent are proposed to deal with heavy task. Simultaneously, a visual optimization scheme based on callback and multithread interpolation is proposed to settle the caton phenomenon caused by the multi nodes in the 3D situation. On this basis, we verify the cluster simulation scene of 40 and 200 nodes. The experiments demonstrates a favourable visual impact with high performance.
{"title":"Visual Optimization of Cluster Simulation Based on Multi Process Service and Load Balancing Agent","authors":"Y. Xiao, Mei-Min Wu, Qian Bi","doi":"10.1145/3318299.3318306","DOIUrl":"https://doi.org/10.1145/3318299.3318306","url":null,"abstract":"This article introduces the OsgEarth open source project and the establishment of three-dimensional (3D) cluster situation. On account of multiple nodes and heavy task, the simulation visual effect in the 3D situation is not smooth. Aiming at the problems mentioned above, a multi process service architecture and a dynamic load balancing agent are proposed to deal with heavy task. Simultaneously, a visual optimization scheme based on callback and multithread interpolation is proposed to settle the caton phenomenon caused by the multi nodes in the 3D situation. On this basis, we verify the cluster simulation scene of 40 and 200 nodes. The experiments demonstrates a favourable visual impact with high performance.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114871062","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}
To further increase prediction accuracy, improve power management and reduce waste, this paper proposes a hybrid electric load forecasting model based on wavelet analysis (WA) and least squares support vector regression (LSSVR) with particle swarm optimization (PSO) algorithm. Where wavelet analysis is used to transform the original electric data sequence into multi-resolution subsets during the preprocessing stage and then the decomposed subsets are inserted into LSSVR to realize prediction, finally the ultimate prediction results are obtained via the wavelet reconstruction with all the independent prediction results. However, the key to influence forecasting accuracy is the parameters used in the LSSVR, in this paper PSO is used to optimize the kernel parameter Δ and the regularization parameter γ of LSSVR and choose the appropriate parameters for the hybrid forecasting model. The effectiveness of the proposed hybrid model has been proved in electric load prediction; the prediction results show that the proposed hybrid model outperforms the Elman networks model, the radial basis function (RBF) neural network model and LSSVR optimized only with PSO. The hybrid model achieves satisfying results, the mean absolute percentage error (MAPE) with 0.907% and the coefficient of determination (R 2) with 0.9936, it offers a higher forecasting precision.
{"title":"A Hybrid Model of Least Squares Support Vector Regression Optimized by Particle Swarm Optimization for Electricity Demand Prediction","authors":"Zirong Li, Lian Li","doi":"10.1145/3318299.3318332","DOIUrl":"https://doi.org/10.1145/3318299.3318332","url":null,"abstract":"To further increase prediction accuracy, improve power management and reduce waste, this paper proposes a hybrid electric load forecasting model based on wavelet analysis (WA) and least squares support vector regression (LSSVR) with particle swarm optimization (PSO) algorithm. Where wavelet analysis is used to transform the original electric data sequence into multi-resolution subsets during the preprocessing stage and then the decomposed subsets are inserted into LSSVR to realize prediction, finally the ultimate prediction results are obtained via the wavelet reconstruction with all the independent prediction results. However, the key to influence forecasting accuracy is the parameters used in the LSSVR, in this paper PSO is used to optimize the kernel parameter Δ and the regularization parameter γ of LSSVR and choose the appropriate parameters for the hybrid forecasting model. The effectiveness of the proposed hybrid model has been proved in electric load prediction; the prediction results show that the proposed hybrid model outperforms the Elman networks model, the radial basis function (RBF) neural network model and LSSVR optimized only with PSO. The hybrid model achieves satisfying results, the mean absolute percentage error (MAPE) with 0.907% and the coefficient of determination (R 2) with 0.9936, it offers a higher forecasting precision.","PeriodicalId":164987,"journal":{"name":"International Conference on Machine Learning and Computing","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121258390","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：