With the continuing development and innovation of modern information and communication technologies in recent years, social media platforms, such as WeChat and Microblog, have been witnessed to play a key role in employee management for enterprises or organizations, enabling individual or groups of employees to express their viewpoints or report their works in a real-time fashion. The resulting Cyber-workspace in social media, which exists in parallel with employees' physical workspace, has the potential of greatly changing the forms and functions of the organizations, as well as the knowledge structures and behavioral patterns of employees, thus bringing unprecedented challenges to the existing research efforts on the management of organizational behavior. In this paper, we propose an approach of behavioral profiling for analyzing and better understanding employees, with the input of large amounts of real-time collected data generated by employees' daily reported works on social media. Our aim is to characterize the diversified behavior patterns of employees in an automatic and accurate fashion for organization management. We also validate our proposed method using a real-world dataset collected from WeChat, and the experimental results prove our analysis, as well as the effectiveness of our approach.
{"title":"Behavioral profiling for employees using social media: A case study based on wechat","authors":"Xiaochun Ni, Shuai Zeng, Rui Qin, Juanjuan Li, Yong Yuan, Feiyue Wang","doi":"10.1109/CAC.2017.8244176","DOIUrl":"https://doi.org/10.1109/CAC.2017.8244176","url":null,"abstract":"With the continuing development and innovation of modern information and communication technologies in recent years, social media platforms, such as WeChat and Microblog, have been witnessed to play a key role in employee management for enterprises or organizations, enabling individual or groups of employees to express their viewpoints or report their works in a real-time fashion. The resulting Cyber-workspace in social media, which exists in parallel with employees' physical workspace, has the potential of greatly changing the forms and functions of the organizations, as well as the knowledge structures and behavioral patterns of employees, thus bringing unprecedented challenges to the existing research efforts on the management of organizational behavior. In this paper, we propose an approach of behavioral profiling for analyzing and better understanding employees, with the input of large amounts of real-time collected data generated by employees' daily reported works on social media. Our aim is to characterize the diversified behavior patterns of employees in an automatic and accurate fashion for organization management. We also validate our proposed method using a real-world dataset collected from WeChat, and the experimental results prove our analysis, as well as the effectiveness of our approach.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125595447","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 : 2017-10-20DOI: 10.1109/CAC.2017.8244164
Rui Qin, Yong Yuan, Xiaochun Ni, Feiyue Wang
When online advertising met the big data technology, programmatic buying has become more and more popular, in which Real Time Bidding (RTB) is regarded as one of the most important formats of programmatic buying advertising. In RTB advertising markets, there is a two-stage auction process for each ad impression, in which Demand-Side Platforms (DSPs) adopt a two-stage resale model to get their revenues. Thus, for each DSP, how to design effective auction mechanisms in the two-stage auction process so as to get higher revenues for both itself and its advertisers has become a critical issue. This paper aims to study this issue, and propose a new bidding and pricing mechanism for the DSP. We also utilize the computational experiment approach to evaluate our proposed mechanism, and the experimental results show that our new mechanism can improve the revenues of both the DSP and the advertisers.
当网络广告遇到大数据技术，程序化购买越来越受欢迎，其中实时竞价(Real Time Bidding, RTB)被认为是程序化购买广告最重要的形式之一。在RTB广告市场中，每个广告印象都有一个两阶段的拍卖过程，其中需求方平台(dsp)采用两阶段转售模式来获得收入。因此，对于每个DSP来说，如何在两阶段的拍卖过程中设计出有效的拍卖机制，从而为自己和广告商获得更高的收益就成为了一个关键问题。本文旨在对这一问题进行研究，并提出一种新的DSP竞价定价机制。我们还利用计算实验的方法来评估我们提出的机制，实验结果表明，我们的新机制可以提高DSP和广告商的收入。
{"title":"Exploring new mechanisms for demand-side platforms in real time bidding markets","authors":"Rui Qin, Yong Yuan, Xiaochun Ni, Feiyue Wang","doi":"10.1109/CAC.2017.8244164","DOIUrl":"https://doi.org/10.1109/CAC.2017.8244164","url":null,"abstract":"When online advertising met the big data technology, programmatic buying has become more and more popular, in which Real Time Bidding (RTB) is regarded as one of the most important formats of programmatic buying advertising. In RTB advertising markets, there is a two-stage auction process for each ad impression, in which Demand-Side Platforms (DSPs) adopt a two-stage resale model to get their revenues. Thus, for each DSP, how to design effective auction mechanisms in the two-stage auction process so as to get higher revenues for both itself and its advertisers has become a critical issue. This paper aims to study this issue, and propose a new bidding and pricing mechanism for the DSP. We also utilize the computational experiment approach to evaluate our proposed mechanism, and the experimental results show that our new mechanism can improve the revenues of both the DSP and the advertisers.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131973416","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 : 2017-10-01DOI: 10.1109/CAC.2017.8244133
Chunyang Sheng, Jun Zhao
Aiming at the optimal scheduling problem of byproduct gas system in steel industry, a knowledge and mathematical programming-based optimal scheduling method is proposed in this study. On one hand, a fuzzy model is designed to extract the expert scheduling knowledge from the historical data of the industrial process. And then, a great deal of scheduling knowledge is employed to compose a fuzzy rules base, which can be used for fuzzy inference of operation scheme with a new input. On the other hand, a mixed integer linear program (MILP) method is built to further optimize the operation scheme. Thus, a more reasoning and optimal operation scheme can be achieved with the consideration of both the expert knowledge and the mathematical programming method. Finally, a byproduct gas system of one steel industry is studied for experiments to verify the effectiveness and practicability of the proposed method.
{"title":"Knowledge and mathematical programming-based optimal scheduling for byproduct gas system in steel industry","authors":"Chunyang Sheng, Jun Zhao","doi":"10.1109/CAC.2017.8244133","DOIUrl":"https://doi.org/10.1109/CAC.2017.8244133","url":null,"abstract":"Aiming at the optimal scheduling problem of byproduct gas system in steel industry, a knowledge and mathematical programming-based optimal scheduling method is proposed in this study. On one hand, a fuzzy model is designed to extract the expert scheduling knowledge from the historical data of the industrial process. And then, a great deal of scheduling knowledge is employed to compose a fuzzy rules base, which can be used for fuzzy inference of operation scheme with a new input. On the other hand, a mixed integer linear program (MILP) method is built to further optimize the operation scheme. Thus, a more reasoning and optimal operation scheme can be achieved with the consideration of both the expert knowledge and the mathematical programming method. Finally, a byproduct gas system of one steel industry is studied for experiments to verify the effectiveness and practicability of the proposed method.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115227308","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}
Fault detection and isolation (FDI), which is a critical part of modern industrial systems, plays a key role in the maintainability, safety, and reliability of processes. Existing FDI approaches are dependent on varying degrees of knowledge of the process, limiting their implementation in practical industrial processes. Based on the least absolute shrinkage and selection operator (lasso), this paper proposes Bayesian fused lasso to overcome the above-mentioned problem. The fault isolation problem is converted into a quadratic programming problem with constraints, which can be solved satisfactorily by the Bayesian fused lasso. Ultimately, the probability distribution of every fault variable can be obtained. In the case of unknown fault directions, fault isolation is carried out. Therefore, the possibility of misdiagnosis was reduced. The reliability and effectiveness of the proposed method are illustrated with the case.
{"title":"Fault isolation based on Bayesian fused lasso","authors":"Shenbo Zhang, Zhengbing Yan, Ping Wu, Zhengjiang Zhang","doi":"10.1109/CAC.2017.8243248","DOIUrl":"https://doi.org/10.1109/CAC.2017.8243248","url":null,"abstract":"Fault detection and isolation (FDI), which is a critical part of modern industrial systems, plays a key role in the maintainability, safety, and reliability of processes. Existing FDI approaches are dependent on varying degrees of knowledge of the process, limiting their implementation in practical industrial processes. Based on the least absolute shrinkage and selection operator (lasso), this paper proposes Bayesian fused lasso to overcome the above-mentioned problem. The fault isolation problem is converted into a quadratic programming problem with constraints, which can be solved satisfactorily by the Bayesian fused lasso. Ultimately, the probability distribution of every fault variable can be obtained. In the case of unknown fault directions, fault isolation is carried out. Therefore, the possibility of misdiagnosis was reduced. The reliability and effectiveness of the proposed method are illustrated with the case.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115322268","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 : 2017-10-01DOI: 10.1109/CAC.2017.8244010
Lu Chang, Liang Shan, Siyuan Xu, Chenglin Liu
Pedestrians evacuation is key to improve the traffic efficiency and guarantee the safety of pedestrians in high density area, which have been widely concerned by scholars. This paper establishes a macroscopic model for T-shaped channel consisting of some corridors and nodes, which is very common in real traffic system. Then, with the goal of avoiding congestion and maximizing the pedestrian flow of the whole model, a feedback controller is designed and the stability of the system is analyzed. Finally, the stability range is obtained and the simulation result is given to show the efficiency of the controller.
{"title":"State feedback control of T-shaped pedestrian channel","authors":"Lu Chang, Liang Shan, Siyuan Xu, Chenglin Liu","doi":"10.1109/CAC.2017.8244010","DOIUrl":"https://doi.org/10.1109/CAC.2017.8244010","url":null,"abstract":"Pedestrians evacuation is key to improve the traffic efficiency and guarantee the safety of pedestrians in high density area, which have been widely concerned by scholars. This paper establishes a macroscopic model for T-shaped channel consisting of some corridors and nodes, which is very common in real traffic system. Then, with the goal of avoiding congestion and maximizing the pedestrian flow of the whole model, a feedback controller is designed and the stability of the system is analyzed. Finally, the stability range is obtained and the simulation result is given to show the efficiency of the controller.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115485263","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 : 2017-10-01DOI: 10.1109/CAC.2017.8243123
Cunwu Han, Mengqi Li, Lei Liu, S. Bi, Zhilei Jia
In cellular mobile wireless networks, nonlinear channel fading is always a critical factor which will scientifically deteriorate the Quality of Service (QoS) for the users. However, it is difficult to describe the nonlinear channel fading by precise mathematical model because of its complexity. So, the control of wireless networks becomes an especial tough work. In this paper, a power and rate control system model is established which considers general nonlinearity of the channel fading which is seldom considered in the literature. And a neural network controller is proposed. The effectiveness and feasibility of the proposed controller are verified by simulation results. The presented method yields superior performance with shorter transient time and higher signal-to-interference-plus-noise ratio (SINR) than traditional predictive control scheme.
{"title":"Neural network control for wireless networks with nonlinear channel fading","authors":"Cunwu Han, Mengqi Li, Lei Liu, S. Bi, Zhilei Jia","doi":"10.1109/CAC.2017.8243123","DOIUrl":"https://doi.org/10.1109/CAC.2017.8243123","url":null,"abstract":"In cellular mobile wireless networks, nonlinear channel fading is always a critical factor which will scientifically deteriorate the Quality of Service (QoS) for the users. However, it is difficult to describe the nonlinear channel fading by precise mathematical model because of its complexity. So, the control of wireless networks becomes an especial tough work. In this paper, a power and rate control system model is established which considers general nonlinearity of the channel fading which is seldom considered in the literature. And a neural network controller is proposed. The effectiveness and feasibility of the proposed controller are verified by simulation results. The presented method yields superior performance with shorter transient time and higher signal-to-interference-plus-noise ratio (SINR) than traditional predictive control scheme.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115716448","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 : 2017-10-01DOI: 10.1109/CAC.2017.8243935
Hong Zhang, H Zhao, Guojiang Zhang, Yanlin Li
A switching circuit system based on FPGA is designed and implemented in this paper to provide high-frequency adjustable-phase voltages for the ultrasonic motor. For the existing drive systems of the ultrasonic motor, the bus voltage is usually 12V DC or 24V DC supporting by an external adapter, the driving voltages of the ultrasonic motor, however, are required to be boosted yet. Then, the step-up transformers are always necessary. Different from that, the novel half-bridge based drive designed in this paper directly applies grid voltage of 220V, 50Hz as the AC input; that is, the external power adapter and the transformers are not necessary anymore. The cost and volume of the drive are subsequently reduced. Furthermore, the input voltages of the ultrasonic motor can be adjusted easily by changing the switching frequency and the conducting time. Meanwhile, a high-order matching circuit is introduced for stabilizing the voltage gain within its operating frequency bandwidth, which leads to a more stable speed regulation performance.
{"title":"Design and implementation of a novel drive for the ultrasonic motor","authors":"Hong Zhang, H Zhao, Guojiang Zhang, Yanlin Li","doi":"10.1109/CAC.2017.8243935","DOIUrl":"https://doi.org/10.1109/CAC.2017.8243935","url":null,"abstract":"A switching circuit system based on FPGA is designed and implemented in this paper to provide high-frequency adjustable-phase voltages for the ultrasonic motor. For the existing drive systems of the ultrasonic motor, the bus voltage is usually 12V DC or 24V DC supporting by an external adapter, the driving voltages of the ultrasonic motor, however, are required to be boosted yet. Then, the step-up transformers are always necessary. Different from that, the novel half-bridge based drive designed in this paper directly applies grid voltage of 220V, 50Hz as the AC input; that is, the external power adapter and the transformers are not necessary anymore. The cost and volume of the drive are subsequently reduced. Furthermore, the input voltages of the ultrasonic motor can be adjusted easily by changing the switching frequency and the conducting time. Meanwhile, a high-order matching circuit is introduced for stabilizing the voltage gain within its operating frequency bandwidth, which leads to a more stable speed regulation performance.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123122597","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 : 2017-10-01DOI: 10.1109/CAC.2017.8242803
Miao Li, Lijuan Li, Risheng Jiao, Hongguang Xiao
With the developing of modern training, advanced requirements of training process such as creative, economical, realistic, safety have been proposed. It is hard for tradition training method to satisfy the training requirement of new technology and equipment in the background of modern industry. Traditional training methods were facing severe challenges. In order to solve the complex technical training problem in the modern industry background, we put forward the opinion that virtual reality and artificial intelligence support the future intelligent training. In this paper, we first presented virtual reality technology as the foundation of future training facilities. Then, we put forward artificial intelligence were the method of generation of intelligent training cases. Following that, we stated rewards the mode that it is resort to advanced big data and information technology to complete base data collection of the case and it is intelligent to generate case model with artificial intelligence and machine learning. To illustrate those points, we provided a case of virtual reality applying in training system simulation field for power system two overhaul, including problem proposition, solution concept, solution result and application prospect. In the end, we summed up the whole paper. The strategies and opinions put forward in this paper can not only help us to understand virtual reality and artificial intelligence's effects in the process of building intelligent training in the future but also render new ideas and inspirations for experts working in virtual reality researching, machine learning and artificial intelligent decision-making, power system simulation and some related areas. Furthermore, it provided the conceptual support for constructing intelligent training, intelligent manufacture, as well as intelligent China.
{"title":"Virtrul reality and artificial intelligence support future training development","authors":"Miao Li, Lijuan Li, Risheng Jiao, Hongguang Xiao","doi":"10.1109/CAC.2017.8242803","DOIUrl":"https://doi.org/10.1109/CAC.2017.8242803","url":null,"abstract":"With the developing of modern training, advanced requirements of training process such as creative, economical, realistic, safety have been proposed. It is hard for tradition training method to satisfy the training requirement of new technology and equipment in the background of modern industry. Traditional training methods were facing severe challenges. In order to solve the complex technical training problem in the modern industry background, we put forward the opinion that virtual reality and artificial intelligence support the future intelligent training. In this paper, we first presented virtual reality technology as the foundation of future training facilities. Then, we put forward artificial intelligence were the method of generation of intelligent training cases. Following that, we stated rewards the mode that it is resort to advanced big data and information technology to complete base data collection of the case and it is intelligent to generate case model with artificial intelligence and machine learning. To illustrate those points, we provided a case of virtual reality applying in training system simulation field for power system two overhaul, including problem proposition, solution concept, solution result and application prospect. In the end, we summed up the whole paper. The strategies and opinions put forward in this paper can not only help us to understand virtual reality and artificial intelligence's effects in the process of building intelligent training in the future but also render new ideas and inspirations for experts working in virtual reality researching, machine learning and artificial intelligent decision-making, power system simulation and some related areas. Furthermore, it provided the conceptual support for constructing intelligent training, intelligent manufacture, as well as intelligent China.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"140 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124423619","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 paper presents an intelligent operation and maintenance training system for power grid operation staff of substation based on augmented reality (AR), whose functions, characteristics, software configuration and implementation are introduced in details. As the training and examination system for operation and maintenance personnel, the system was composed of intelligent training system and intelligent operation and maintenance system. The realistic settings are re-created by augmented reality effectively, which could realize the “panoramic” experience, in order to strengthen the operability of the training process. Based on this, the intelligent level of skills training continues to make progress. While the intelligent training simulation system with the operation and maintenance mode of state maintenance is built based on big data analysis, which conducts to provide strong support for intelligent substation engineering applications and practice.
{"title":"Design and development of intelligent operation and maintenance training system for substation based on augmented reality","authors":"Yong Peng, G. Yu, Wei-Chun Ni, Zhengquan Lv, Yuliang Jiang, Jing Chen","doi":"10.1109/CAC.2017.8243621","DOIUrl":"https://doi.org/10.1109/CAC.2017.8243621","url":null,"abstract":"This paper presents an intelligent operation and maintenance training system for power grid operation staff of substation based on augmented reality (AR), whose functions, characteristics, software configuration and implementation are introduced in details. As the training and examination system for operation and maintenance personnel, the system was composed of intelligent training system and intelligent operation and maintenance system. The realistic settings are re-created by augmented reality effectively, which could realize the “panoramic” experience, in order to strengthen the operability of the training process. Based on this, the intelligent level of skills training continues to make progress. While the intelligent training simulation system with the operation and maintenance mode of state maintenance is built based on big data analysis, which conducts to provide strong support for intelligent substation engineering applications and practice.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124431731","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 : 2017-10-01DOI: 10.1109/CAC.2017.8244044
Xianlun Tang, Deyi Xiong, Jiaxin Li, Yali Wan
Traditional pairwise learning to rank algorithms pay little attention to top ranked documents in the query list, and do not work well when they are used on a data set with multiple rating grades. In this paper, a novel pairwise learning to rank algorithm is proposed to solve this problem. This algorithm defines a bounded loss function and introduces the preference weights between document pairs into it. Because the batch gradient descent method will lead to slow optimization and the stochastic gradient descent method will be easily affected by noises, a mini-batch gradient descent method is proposed to optimize the algorithm, which makes the number of iteration no longer dependent on the size of samples. Finally, experiments on OHSUMED data set and MQ2008 data set demonstrate the effectiveness of the proposed algorithm.
{"title":"A pairwise learning to rank algorithm based on bounded loss and preference weight","authors":"Xianlun Tang, Deyi Xiong, Jiaxin Li, Yali Wan","doi":"10.1109/CAC.2017.8244044","DOIUrl":"https://doi.org/10.1109/CAC.2017.8244044","url":null,"abstract":"Traditional pairwise learning to rank algorithms pay little attention to top ranked documents in the query list, and do not work well when they are used on a data set with multiple rating grades. In this paper, a novel pairwise learning to rank algorithm is proposed to solve this problem. This algorithm defines a bounded loss function and introduces the preference weights between document pairs into it. Because the batch gradient descent method will lead to slow optimization and the stochastic gradient descent method will be easily affected by noises, a mini-batch gradient descent method is proposed to optimize the algorithm, which makes the number of iteration no longer dependent on the size of samples. Finally, experiments on OHSUMED data set and MQ2008 data set demonstrate the effectiveness of the proposed algorithm.","PeriodicalId":116872,"journal":{"name":"2017 Chinese Automation Congress (CAC)","volume":"176 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124261170","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：