Publications | Shengjun Zhang

2024

TUP
梯度提升算法实战 : 基于XGBoost和scikit-learn

(Corey Wade) 科里·韦德, and 张生军

Apr 2024

中国国家版本馆CIP数据核字(2024)第065073号

Abs Bib

本书首先介绍了scikit-learn中的机器学习和XGBoost技术，然后探讨了梯度提升背后的理论知识。包括决策树、装袋技术及XGBoost超参数。本书将从头开始构建梯度提升模型，并将梯度提升扩展到大数据领域。本书重点探讨XGBoost的细节，着重于速度增强和通过数学推导导出参数。通过详细案例研究，练习使用 scikit-learn 、原始的 Python API 构建、以及微调 XGBoost 分类器和回归器。利用XGBoost超参数来提高评分、纠正缺失值、缩放不平衡数据集以及微调备选基学习器。最后，应用高级 XGBoost 技术，如构建非相关的集成模型、堆叠模型，并使用稀疏矩阵、定制的转换器和管道为行业部署准备模型。全书共分3篇：第1篇（第1～4章）为入门篇，介绍XGBoost背后的实用理论，包括装袋和提升模型结构、数据预处理、回归和分类模型、XGBoost基本模型及超参数微调；第2篇（第5～7章）为进阶篇，介绍XGBoost框架构成及超参数优化；第3篇（第8 10章）为进阶篇，着重讨论微调备选基学习器、创新技巧、特征工程，并使用稀疏矩阵、定制转换器和管道，练习构建适合行业部署的模型。全书提供了大量应用实例。本书适合作为高等院校计算机、软件工程专业高年级本科生、研究生的教材，同时适合有一定机器学习基础的数据科学家、机器学习工程师和研究人员阅读，为解决复杂问题提供实用指导。
@book{GradientBoosting2024, title = {梯度提升算法实战 : 基于XGBoost和scikit-learn}, language = {zh}, author = {科里·韦德, (Corey Wade) and 张生军}, translator = {张生军}, year = {2024}, month = apr, publisher = {清华大学出版社}, address = {北京}, isbn = {978-7-302-65951-8}, note = {中国国家版本馆CIP数据核字(2024)第065073号}, series = {机器学习－算法}, classification = {TP181}, }
PATENT
A Road Perception Method and System Based on Millimeter Wave Radar

Shengjun Zhang, Jun Wang, Shixun Wu, and 3 more authors

May 2024

Publication Number: CN118091649A

Abs Bib HTML

The invention discloses a road perception method and system based on millimeter wave radar, comprising the following steps: S1, the vehicle detects whether there is a base station in the surrounding environment; S2, determining whether networking is required; S3, performing signal analysis on the network formed in step S2, determining the network status, if the network is normal, performing target detection processing, if the network is abnormal, continuing to detect surrounding base stations that can form a network; S4, classifying the detected targets, performing behavior analysis and condition analysis on the classified targets; S5, determining whether the target will affect driving, if it does not affect normal driving, continue detection. The road perception method and system based on millimeter wave radar not only can detect road conditions, but also enable vehicles to communicate using RF networking, perform network detection using multiple base stations, and share communication results, thereby improving detection accuracy and expanding the detection range.
@patent{CN118091649A, title = {A Road Perception Method and System Based on Millimeter Wave Radar}, number = {CN118091649A}, year = {2024}, month = may, note = {Publication Number: CN118091649A}, author = {Zhang, Shengjun and Wang, Jun and Wu, Shixun and Zheng, Borun and Tan, Wei and Zhang, Kaihong}, assignee = {Chongqing Jiaotong University}, address = {66 Xuefu Avenue, Nan'an District, Chongqing 400060}, application_number = {2024101841224}, application_date = {2024-02-19}, classification = {G01S13/88(2006.01)I}, }
PATENT
A Personnel Re-identification System and Method Based on RF Signals in Foundation Pit Construction Scenarios

Shengjun Zhang, Zengqiang Zuo, Xinxin Liu, and 2 more authors

May 2024

Publication Number: CN118018961B

Abs Bib HTML

The invention discloses a personnel re-identification system based on RF signals in foundation pit construction scenarios, including an RF transceiver base station and an RF receiver base station. The RF transceiver base station is used to emit RF signals to form a signal field and receive reflected RF signals to obtain data in the signal field. The RF receiver base station is used to receive the remaining scattered waves. The RF transceiver base station includes an RF transmitting unit and an RF receiving unit. The invention adopts a personnel re-identification system and method based on RF signals in foundation pit construction scenarios, which achieves personnel tracking, positioning, and behavior analysis through RF coverage and data collection, reducing operational error risks, saving equipment costs, and improving safety performance.
@patent{CN118018961A, title = {A Personnel Re-identification System and Method Based on RF Signals in Foundation Pit Construction Scenarios}, number = {CN118018961B}, year = {2024}, month = may, note = {Publication Number: CN118018961B}, author = {Zhang, Shengjun and Zuo, Zengqiang and Liu, Xinxin and Wang, Jiawei and Li, Yicai}, assignee = {Chongqing Jiaotong University}, address = {66 Xuefu Avenue, Nan'an District, Chongqing 400060}, application_number = {2024101664094}, application_date = {2024-02-06}, classification = {H04W4/029(2018.01)I}, }

2023

PATENT
A Geotechnical Sample Image Acquisition Device

Shengjun Zhang, Shixun Wu, Kai Xu, and 2 more authors

Jul 2023

Publication Number: CN116380900A

Abs Bib

The invention proposes a geotechnical sample image acquisition device, relating to the technical field of image acquisition equipment. It includes a shooting tent and a base, with the top of the base rotatably connected to a rotating chassis. The shooting tent includes a sliding cover and a fixed cover arranged oppositely, forming a closed display space with the top surface of the base. The fixed cover has a chamber accommodating the sliding cover, and the fixed cover is slidably connected to the sliding cover. The inner side wall of the shooting tent is provided with a lighting component. A longitudinal shooting track is provided inside the fixed cover, and the shooting track abuts against the inner side wall of the fixed cover. A detachable shooting stand is provided on the shooting track, and a camera is provided on any shooting stand. This device can complete all-around image acquisition of geotechnical samples, ensuring image continuity. It features a simple structure, reasonable design, elegant appearance, and is used for 3D panoramic image acquisition. It can effectively complete full-view material acquisition with high efficiency and good quality.
@patent{CN116380900A, title = {A Geotechnical Sample Image Acquisition Device}, number = {CN116380900A}, year = {2023}, month = jul, note = {Publication Number: CN116380900A}, author = {Zhang, Shengjun and Wu, Shixun and Xu, Kai and Xu, Dengyuan and Li, Sheng}, assignee = {Chongqing Jiaotong University}, address = {66 Xuefu Avenue, Nan'an District, Chongqing 400000}, application_number = {2023104050719}, application_date = {2023-04-14}, classification = {G01N21/84(2006.01)I}, }
PATENT
A Hemisphere Cover

Shengjun Zhang, Sheng Li, Kexin Li, and 3 more authors

Jun 2023

Publication Number: CN219240864U

Abs Bib

The utility model proposes a hemisphere cover, which relates to the field of covering device technology. It includes a first arc-shaped plate and a second arc-shaped plate, wherein the second arc-shaped plate is rotatably connected to the first arc-shaped plate through a rotation mechanism; the first arc-shaped plate has a cavity for accommodating the second arc-shaped plate, and the second arc-shaped plate is slidably arranged in the cavity of the first arc-shaped plate; the rotation mechanism includes a fixing member and a rotating member, the rotating member is a hollow tube with limit pieces provided on the inner wall, the outer wall of the fixing member is provided with wedges adapted to the limit pieces, the rotating member is sleeved outside the fixing member, the rotating member is fixedly connected to the second arc-shaped plate, and the fixing member is fixedly connected to the first arc-shaped plate. The first arc-shaped plate and the second arc-shaped plate of the device are connected through the rotation mechanism, which is divided into a fixing member and a rotating member. By fixing the fixing member and the rotating member respectively on the first arc-shaped plate and the second arc-shaped plate, the rotation of the second arc-shaped plate is driven by the rotating member, thereby accurately controlling the opening and closing angle of the hemisphere cover.
@patent{CN219240864U, title = {A Hemisphere Cover}, number = {CN219240864U}, year = {2023}, month = jun, note = {Publication Number: CN219240864U}, author = {Zhang, Shengjun and Li, Sheng and Li, Kexin and Wang, Hao and Feng, Shiyang and Gong, Shaolin}, assignee = {Chongqing Jiaotong University}, address = {66 Xuefu Avenue, Nan'an District, Chongqing 400000}, application_number = {2023208515829}, application_date = {2023-04-14}, classification = {E04B1/343(2006.01)I}, }

2022

PATENT

Automatic Driving Curve Generation Method for High-speed Trains Based on Temperature and Load Conditions

Kai Xu, Hao Zheng, Shixun Wu, and 1 more author

Jun 2022

Bib

@patent{xu2022automatic,
  title = {Automatic Driving Curve Generation Method for High-speed Trains Based on Temperature and Load Conditions},
  author = {Xu, Kai and Zheng, Hao and Wu, Shixun and Zhang, Shengjun},
  year = {2022},
  pages = {5},
  address = {Fuxing Avenue, Shuangfu New Area, Jiangjin District, Chongqing 402247}
}

PATENT

Curve Generation Method for Train Automatic Driving Based on Target and Passenger Flow Differences

Kai Xu, Rui Yang, Shixun Wu, and 2 more authors

Jun 2022

Bib

@patent{xu2022curve,
  title = {Curve Generation Method for Train Automatic Driving Based on Target and Passenger Flow Differences},
  author = {Xu, Kai and Yang, Rui and Wu, Shixun and Zhang, Shengjun and Xu, Dengyuan},
  year = {2022},
  pages = {5},
  address = {Fuxing Avenue, Shuangfu New Area, Jiangjin District, Chongqing 402247}
}

PATENT

Method to Prevent Collision of High-speed Train Groups in Communication Interrupts

Kai Xu, Qingshan Liu, Shixun Wu, and 2 more authors

Jun 2022

Bib

@patent{xu2022collision,
  title = {Method to Prevent Collision of High-speed Train Groups in Communication Interrupts},
  author = {Xu, Kai and Liu, Qingshan and Wu, Shixun and Huang, Darong and Zhang, Shengjun},
  year = {2022},
  pages = {8},
  address = {Fuxing Avenue, Shuangfu New Area, Jiangjin District, Chongqing 402247}
}

PATENT

Control Method of Unmanned Train Control System

Kai Xu, Zhen Chen, Wei Li, and 4 more authors

Jun 2022

Bib

@patent{xu2022control,
  title = {Control Method of Unmanned Train Control System},
  author = {Xu, Kai and Chen, Zhen and Li, Wei and Wu, Shixun and Zhang, Shengjun and Xu, Dengyuan and Tan, Wei},
  year = {2022},
  pages = {18},
  address = {Fuxing Avenue, Shuangfu New Area, Jiangjin District, Chongqing 402247}
}

2021

PATENT

Monitoring and Auxiliary System for High-speed Train Drivers

Kai Xu, Shiwen Zhou, Shixun Wu, and 2 more authors

Jun 2021

Bib

@patent{xu2021monitoring,
  title = {Monitoring and Auxiliary System for High-speed Train Drivers},
  author = {Xu, Kai and Zhou, Shiwen and Wu, Shixun and Zhang, Shengjun and Xu, Dengyuan},
  year = {2021},
  pages = {8},
  address = {Fuxing Avenue, Shuangfu New Area, Jiangjin District, Chongqing 402247}
}

2020

PATENT

Shore-based Video-assisted Positioning Unmanned Aerial Vehicle Large-scale Flow Field Measurement Method and Device

Shengjun Zhang, Honggang Wang, and Dengyuan Xu

Jun 2020

Bib HTML

@patent{CN109859269A,
  title = {Shore-based Video-assisted Positioning Unmanned Aerial Vehicle Large-scale Flow Field Measurement Method and Device},
  author = {Zhang, Shengjun and Wang, Honggang and Xu, Dengyuan},
  year = {2020},
  pages = {14},
  address = {66 Xuefu Avenue, Nanan District, Chongqing 400074}
}

PATENT

Cloud-edge Collaborative Unmanned Train Control System and its Control Method

Kai Xu, Yongchao Tu, Shixun Wu, and 2 more authors

Jun 2020

Bib

@patent{xu2020cloud,
  title = {Cloud-edge Collaborative Unmanned Train Control System and its Control Method},
  author = {Xu, Kai and Tu, Yongchao and Wu, Shixun and Huang, Darong and Zhang, Shengjun},
  year = {2020},
  pages = {15},
  address = {Fuxing Avenue, Shuangfu New Area, Jiangjin District, Chongqing 402247}
}

2019

Neural Network Localization With TOA Measurements Based on Error Learning and Matching

Shixun Wu, Shengjun Zhang, Kai Xu, and 1 more author

IEEE Access, 2019

Abs DOI Bib HTML

Due to the widespread application of location information, the neural network localization method with the advantage of high localization accuracy has received significant interests in recent years. In this paper, we present two new neural network localization methods with time-of-arrival (TOA) measurements. In order to deal with three types of error about TOA measurements such as measurement error, non-line of sight (NLOS) error, and synchronization error, the proposed methods contain an offline training stage and an online localization stage. In the offline stage, the artificial neural network (ANN) or radial basis function (RBF) neural network is utilized to train the range measurements with the output of range errors rather than the position of the mobile terminal (MT). Moreover, due to the unknown signal propagation condition whether it is the line of sight or NLOS propagation, the k -mean clustering algorithm is used to classifying the range errors into different clusters. In the online stage, the range errors are predicted and updated, and then, the linear least square algorithm with the adjusted range measurements is applied for the position estimate of MT. Comparing with the ANN or RBF neural network localization methods, the simulation results show that the proposed methods can effectively reduce the localization error, especially when the training sample is not adequate. In addition, they are insensitive to measurement error, synchronization error, and the distribution of NLOS error. Finally, the memory requirement and computational complexity about different algorithms are analyzed and compared.
@article{8633857, author = {Wu, Shixun and Zhang, Shengjun and Xu, Kai and Huang, Darong}, journal = {IEEE Access}, title = {Neural Network Localization With TOA Measurements Based on Error Learning and Matching}, year = {2019}, volume = {7}, number = {}, pages = {19089-19099}, keywords = {Neural networks;Nonlinear optics;Synchronization;Measurement uncertainty;Position measurement;Wireless sensor networks;Feature extraction;Artificial neural network (ANN);localization;linear least square (LLS);non-line of sight (NLOS);radial basis function (RBF);time of arrival (TOA)}, doi = {10.1109/ACCESS.2019.2897153}, issn = {2169-3536}, month = {}, }

JISE

A Weighting Localization Algorithm with LOS and One-Bound NLOS Identification in Multipath Environments.

Shi-Xun Wu, Sheng-Jun Zhang, Kai Xu, and 1 more author

Journal of Information Science & Engineering, 2019

Bib

@article{wu2019weighting,
  title = {A Weighting Localization Algorithm with LOS and One-Bound NLOS Identification in Multipath Environments.},
  author = {Wu, Shi-Xun and Zhang, Sheng-Jun and Xu, Kai and Huang, Da-Rong},
  journal = {Journal of Information Science \& Engineering},
  volume = {35},
  number = {6},
  year = {2019}
}

A Method, Apparatus, and Electronic Device for Measuring Surface Flow Field in Water Bodies

Shengjun Zhang, Shixun Wu, and Jiang Hu

Apr 2019

Publication Number: CN109584314A

Abs Bib HTML

The invention discloses a method for measuring the surface flow field in water bodies, which relates to the field of image processing and video analysis technology. The method includes: setting and recording the position of an auxiliary calibration camera, using the auxiliary calibration camera to capture a large-scale video of the water surface scene, performing three-dimensional calibration of the auxiliary calibration camera to obtain three-dimensional calibration data; adjusting the position of a measurement camera according to the position of the auxiliary calibration camera, using the measurement camera to capture a video of the local wave details on the water surface; mapping the coordinates of the large-scale water surface video and the local wave detail video, obtaining the relative coordinates of each frame of the local wave detail video image in the large-scale water surface video image; performing wave tracking processing on the local wave detail video to obtain the motion vector information of the waves, and then calibrating according to the three-dimensional calibration data to obtain the flow velocity vector data; stitching each frame of the local wave detail video image based on the relative coordinates and flow velocity vector data to obtain wide-area flow field data, thereby achieving high-precision measurement of the wide-area flow field.
@patent{CN109584314A, title = {A Method, Apparatus, and Electronic Device for Measuring Surface Flow Field in Water Bodies}, number = {CN109584314A}, year = {2019}, month = apr, note = {Publication Number: CN109584314A}, author = {Zhang, Shengjun and Wu, Shixun and Hu, Jiang}, assignee = {Chongqing Jiaotong University}, address = {66 Xuefu Avenue, Nan'an District, Chongqing 400074}, application_number = {2018116156658}, application_date = {2018-12-27}, classification = {G06T7/80(2017.01)I}, }

2016

JCJU

Research on Hand Segmentation Based on Mixed Features in Unmarked Gesture Recognition

Shengjun Zhang, Shixun Wu, Honggang Wang, and 2 more authors

Journal of Chongqing Jiaotong University (Natural Science), Apr 2016

Bib

@article{zhang2016hand,
  title = {Research on Hand Segmentation Based on Mixed Features in Unmarked Gesture Recognition},
  author = {Zhang, Shengjun and Wu, Shixun and Wang, Honggang and Xu, Dengyuan and Huang, Darong},
  journal = {Journal of Chongqing Jiaotong University (Natural Science)},
  volume = {35},
  number = {5},
  pages = {185},
  year = {2016}
}

JUP
基于视觉的无标记手势识别

张生军

Jun 2016

中国版本图书馆CIP数据核字(2016)第146115号

Abs Bib

本书从手势识别全局角度入手，通盘考虑识别过程中的每一个模块。在每个功能模块单独设计基础上，协调各个模块之间的关系，并范化系统的适用范围和手势识别的准确性。同时作为理论研究的测试平台，该系统为时间、空间复杂度较高的算法提供了测试接口。
@book{HandGestureRecognition2016, title = {基于视觉的无标记手势识别}, author = {张生军}, year = {2016}, month = jun, publisher = {吉林大学出版社}, address = {长春}, isbn = {978-7-5677-6761-4}, note = {中国版本图书馆CIP数据核字(2016)第146115号}, series = {计算机视觉－研究}, classification = {TP302.7}, }

2015

Research on Performance of FTN New Coding Technology

Shengjun Zhang

Information and Communication, Jun 2015

Bib

@article{zhang2015ftn,
  title = {Research on Performance of FTN New Coding Technology},
  author = {Zhang, Shengjun},
  journal = {Information and Communication},
  number = {5},
  pages = {12--13},
  year = {2015}
}

Gesture Annotation Based on Fuzzy Conditional Random Fields

Shengjun Zhang, Shixun Wu, Darong Huang, and 1 more author

Science Technology and Engineering, Jun 2015

Bib

@article{zhang2015gesture,
  title = {Gesture Annotation Based on Fuzzy Conditional Random Fields},
  author = {Zhang, Shengjun and Wu, Shixun and Huang, Darong and others},
  journal = {Science Technology and Engineering},
  volume = {15},
  number = {24},
  year = {2015}
}

2012

OE
Fuzzy-based latent-dynamic conditional random fields for continuous gesture recognition

Zhang Shengjun, He Xiaohai, and Teng Qizhi

Optical Engineering, Jun 2012

Abs DOI Bib HTML

We show an original method for automatic hand gesture recognition that makes use of fuzzified latent-dynamic conditional random fields (LDCRF). In this method, fuzzy linguistic variables are used to model the features of hand gestures and then to modify the potential function in LDCRFs. By combining LDCRFs and fuzzy sets, these fuzzy-based LDCRFs (FLDCRF) have the advantages of LDCRFs in sequence labeling along with the advantage of retaining the imprecise character of gestures. The efficiency of the proposed method was tested with unsegmented gesture sequences in three different hand gesture data sets. The experimental results demonstrate that FLDCRFs compare favorably with support vector machines, hidden conditional random fields, and LDCRFs on hand gesture recognition tasks.
@article{10.1117/1.OE.51.6.067202, author = {Shengjun, Zhang and Xiaohai, He and Qizhi, Teng}, title = {Fuzzy-based latent-dynamic conditional random fields for continuous gesture recognition}, volume = {51}, journal = {Optical Engineering}, number = {6}, publisher = {SPIE}, pages = {067202}, keywords = {Fuzzy logic, Gesture recognition, Data modeling, Motion models, Optical engineering, Sensors, Video, Computing systems, Eye models, Performance modeling}, year = {2012}, doi = {10.1117/1.OE.51.6.067202}, url = {https://doi.org/10.1117/1.OE.51.6.067202}, }

Removal of Motion Human Body Shadows Based on Weak Threshold Segmentation Algorithm

Kewei Li, Xiaomin Yang, Xiaohai He, and 1 more author

Computer Engineering, Jun 2012

Bib

@article{li2012removal,
  title = {Removal of Motion Human Body Shadows Based on Weak Threshold Segmentation Algorithm},
  author = {Li, Kewei and Yang, Xiaomin and He, Xiaohai and Zhang, Shengjun},
  journal = {Computer Engineering},
  volume = {38},
  number = {5},
  pages = {173--175},
  year = {2012}
}

2011

JCAR

Improved GMM-based Motion Object Detection Method

Gang Li, Xiaohai He, Shengjun Zhang, and 1 more author

Journal of Computer Applications Research, Jun 2011

Bib

@article{ligang2011improve,
  title = {Improved GMM-based Motion Object Detection Method},
  author = {Li, Gang and He, Xiaohai and Zhang, Shengjun and Gao, Mingliang},
  journal = {Journal of Computer Applications Research},
  volume = {28},
  number = {12},
  pages = {4738--4741},
  year = {2011}
}

JCS

Region Template Matching Based Multi-target Real-time Tracking

Yuegong Lin, Xiaohai He, and Shengjun Zhang

Journal of Computer Simulation, Jun 2011

Bib

@article{lin2011region,
  title = {Region Template Matching Based Multi-target Real-time Tracking},
  author = {Lin, Yuegong and He, Xiaohai and Zhang, Shengjun},
  journal = {Journal of Computer Simulation},
  volume = {28},
  number = {1},
  pages = {277--280},
  year = {2011}
}

2009

PATENT

Method and Device for Non-steady Flow Measurement of Small Flow Rate

Weiliej, Shuntai Yang, Hong Ji, and 6 more authors

Jun 2009

Bib

@manual{weiliej2009method,
  title = {Method and Device for Non-steady Flow Measurement of Small Flow Rate},
  author = {Weiliej and Yang, Shuntai and Ji, Hong and Li, Shaonian and Yu, Zhenyan and Zhang, Liqiang and Zhang, Shengjun and Qiang, Yan and others},
  year = {2009}
}

2007

JST

Plug and Play of Traditional and Intelligent Sensors Based on 1451 Protocol

Shengjun Zhang, Shuyang Ran, Jian Zheng, and 3 more authors

Journal of Sensor Technology, Jun 2007

Bib

@article{zhang2007plug,
  title = {Plug and Play of Traditional and Intelligent Sensors Based on 1451 Protocol},
  author = {Zhang, Shengjun and Ran, Shuyang and Zheng, Jian and Shi, Yi and Zeng, Jun and Man, Yi},
  journal = {Journal of Sensor Technology},
  volume = {20},
  number = {2},
  pages = {354--357},
  year = {2007}
}