模式识别与人工智能
Home      About Journal      Editorial Board      Instructions      Ethics Statement      Contact Us                   中文
Pattern Recognition and Artificial Intelligence
22 Judgement and Disposal of Academic Misconduct Article
22 Copyright Transfer Agreement
22 Proof of Confidentiality
22 Requirements for Electronic Version
More....
22 Chinese Association of Automation
22 National ResearchCenter for Intelligent Computing System
22 Institute of Intelligent Machines,Chinese Academy of Sciences
More....
 
 
2023 Vol.36 Issue.10, Published 2023-10-25
Papers and Reports    Researches and Applications    Surveys and Reviews   
   
Papers and Reports
867 Parallel Operating Rooms: A New Model of Perioperative Nursing Process and Smart Surgical Platform Management
WANG Huizhen, ZHANG Jie, YU Yi, ZHAO Lin, LI Kuinan, MA Huiying, QI Xiaojing, WANG Jing, WANG Yutong, LIN Yilun, XU Li, SHEN Le, LI Hanzhong, WANG Fei-Yue

With the advancement of medical technology, the operating room is confronted with increasing complexity challenges, especially in the areas of teaching and training, nursing coordination, scientific research innovation and management. In this paper, the concept of a parallel operating room is proposed based on parallel healthcare theory to optimize perioperative nursing processes by artificial intelligence and comprehensive data analysis, thereby improving surgical platform management efficiency. Real systems and virtual systems are integrated in the parallel operating room through online learning, offline computing, and virtual-physical interaction to achieve precise management and control of surgical procedures. The application of parallel operating rooms in clinical nursing and management as well as nursing education and research is discussed in detail, and the potential of parallel operating rooms in future medical fields is prospected. The proposed integrated solution alleviates the problems faced by traditional operating rooms and lays a foundation for the development of smart healthcare, while enhancing safety and operational efficiency in the operating room.
2023 Vol. 36 (10): 867-876 [Abstract] ( 426 ) [HTML 1KB] [ PDF 2095KB] ( 808 )
877 Working Memory Theory Driven Natural Attribute Prediction Model for Social Media User Profiling
LIU Jinhang, LI Lin, LONG Sijie, WANG Conghui
Constructing user profiling systems using contents generated by social media user can offer personalized services and precise marketing for e-commerce platform. It is a significant research direction in the field of social media analysis. In this paper, the document-level multimodal data formed by users publishing content chronologically is studied, and the challenges brought by that to user profiling are analyzed. Aiming at the natural attribute primarily related to user gender and birth year, how to deal with and analyze the document-level multimodal data posted by social media users efficiently is studied as well. A natural attribute prediction model for social media user profiling is proposed. Inspired by cognitive psychology, an effective data chunking method is designed via working memory theory to alleviate the problems of semantics broken and synthetic discourse in traditional methods. To solve the problem of user content preference, an attention mechanism is employed to balance task contributions between intra-modal and inter-modal data. Experiments show that the proposed model is superior in user gender and birth year prediction.
2023 Vol. 36 (10): 877-889 [Abstract] ( 248 ) [HTML 1KB] [ PDF 1288KB] ( 1090 )
890 Chinese-Vietnamese Cross-Language Event Retrieval Incorporating Event Knowledge
HUANG Yuxin, DENG Tongjie, YU Zhengtao, XIAN Yantuan
The goal of Chinese-Vietnamese cross-language event retrieval task is to retrieve Vietnamese documents expressing the same event based on the input Chinese query. Existing cross-language retrieval models exhibit poor alignment in Chinese-Vietnamese low-resource retrieval, and simple semantic matching retrieval struggles to comprehend the event semantic information of complex queries. To address this issue, a Chinese-Vietnamese cross-language event retrieval method incorporating event knowledge is proposed. A Chinese-Vietnamese cross-language event pre-training module is built for continuous pre-training to improve the representation performance of the model on Chinese-Vietnamese low-resource languages. The difference between the masked predicted values and the true values of the event is discriminated based on contrastive learning to encourage the model to better understand and capture the event knowledge features. Experiments on cross-language event retrieval tasks and cross-language question-and-answer tasks demonstrate the performance improvement of the proposed method.
2023 Vol. 36 (10): 890-901 [Abstract] ( 265 ) [HTML 1KB] [ PDF 1098KB] ( 719 )
Surveys and Reviews
902 Reinforcement Learning and Its Application in Robot Task Planning: A Survey
ZHANG Xiaoming, GAO Shijie, YAO Changyu, CHU Yu, PENG Shuo
Reinforcement learning enables robots learn the optimal action policy through the interaction with the environment, representing an important frontier direction in the field of robotics. In this paper, the formal modeling of robot task planning problem is briefly introduced, and the main methods of reinforcement learning are analyzed, including model-free reinforcement learning, model-based reinforcement learning and hierarchical reinforcement learning. The research progress in robot task planning based on reinforcement learning is explored. Various reinforcement learning methods and their applications are discussed as well. Finally, the key problems of reinforcement learning in robot applications are summarized, and the future research directions are prospected.
2023 Vol. 36 (10): 902-917 [Abstract] ( 537 ) [HTML 1KB] [ PDF 900KB] ( 1049 )
Researches and Applications
918 Renewal Theory Model of First Hitting Time Analysis for Continuous Evolutionary Algorithms
ZHOU Zhensheng, WANG Lin, FENG Fujian, TAN Mian, HE Xing, ZHANG Zaijun
In the research on upper bound of the first hitting time for continuous evolutionary algorithms,strong assumptions are required and less attention is given to its lower bound . In this paper, martingale theory and renewal process are introduced and combined with Wald's inequality and renewal theorem. A renewal theory model based on progress rate is proposed to estimate the upper and lower bounds of the expected first hitting time of evolution strategies. The renewal theory model relies on the initial population and the probability density function of the progress rate, providing an estimation advantage for the analysis of the first hitting time of evolutionary strategies. To verify the validity of the proposed renewal theory model, experiments are conducted to estimate the expected first hitting time. Firstly, the expected first hitting time of (1,λ)evolution strategies with uniform mutation on a two-dimensional inclined plane problem is calculated. The closed-form expression for the relationship between(1,λ)evolution strategies population size and the time upper and lower bounds is obtained. It is proved that the expected first hitting time is not negatively correlated with the population size. Next, the expected first hitting time of evolution strategies with uniform mutation on a five-dimensional hyperplane problem is calculated, and the closed-form expression for theoretical upper and lower bounds is derived. Numerical experiments show that the theoretically calculated upper and lower bounds are consistent with the actual expected first hitting time, which provides a theoretical tool for analyzing the first hitting time of evolution strategies.
2023 Vol. 36 (10): 918-930 [Abstract] ( 238 ) [HTML 1KB] [ PDF 845KB] ( 852 )
931 Siamese Contrastive Network Based Multilingual Parallel Sentence Pair Extraction between Chinese and Southeast Asian Languages
ZHOU Yuanzhuo, MAO Cunli, SHEN Zheng, ZHANG Siqi, YU Zhengtao, WANG Zhenhan
The poor performance of parallel sentence pair extraction application on Southeast Asian languages with scarce resources is primarily due to the weak representation capabilities of the sentence pair extraction models caused by the lack of training corpora. Therefore, a siamese contrastive network based multilingual parallel sentence pair extraction between Chinese and Southeast Asian languages is proposed to optimize model structure, training strategy and data. Firstly, a siamese contrastive network framework is employed, integrating contrastive learning concept into the siamese network to enhance the representation capability for parallel sentence pairs. Next, a strategy of joint training with similar languages is introduced to share knowledge effectively and improve the learning ability of the model. Finally, Chinese-mixed Southeast Asian parallel sentence pairs are constructed by multilingual word replacement, providing abundant sample information for training. Experiments on Chinese-Thai and Chinese-Lao datasets demonstrate that the proposed method effectively enhances the performance of parallel sentence pair extraction.
2023 Vol. 36 (10): 931-941 [Abstract] ( 258 ) [HTML 1KB] [ PDF 1114KB] ( 1012 )
942 Social Recommendation Model Based on Self-Supervised Graph Masked Neural Networks
ZANG Xiubo, XIA Hongbin, LIU Yuan
The existing self-supervised social recommendation models mostly construct self-supervised signals through the strategies of manual heuristic graph enhancement and contrasts between single-relational views. Thus, the performance of the model is easily affected by the quality of enhanced self-supervised signals, making it challenging to adaptively suppress noise. To solve these problems, a social recommendation model based on self-supervised graph masked neural networks(SGMN) is proposed. Firstly, single-relational views for user-social interaction and item classification are constructed respectively, as well as high-order connected heterogeneous graphs. Specifically, the graph masked learning paradigm is adopted to guide adaptive and learnable data augmentation for the user-social graph. Secondly, a heterogeneous graph encoder is designed to learn the latent semantics of the views, and cross-view contrastive learning is performed on user and item embeddings to complete self-supervised tasks. Then, weighted fusion is conducted on the user and item embeddings separately for recommendation task. Finally, a multi-task training strategy is employed to jointly optimize self-supervised learning, recommendation and graph masked tasks. Experiments on three real datasets demonstrate a certain performance improvement of SGMN.
2023 Vol. 36 (10): 942-952 [Abstract] ( 322 ) [HTML 1KB] [ PDF 782KB] ( 892 )
953 RGB-D SLAM Algorithm Based on Delayed Semantic Information in Dynamic Environment
WANG Hao, ZHOU Shenchao, FANG Baofu
Visual simultaneous localization and mapping(SLAM) cannot be applied to dynamic environment. The mainstream solution is to combine segmentation network and SLAM. However, real-time operation of SLAM systems is not guaranteed due to the processing speed constraints of segmentation network. Therefore, a RGB-D SLAM algorithm based on delayed semantic information in dynamic environment is proposed. Firstly, tracking and segmentation threads run in parallel. To obtain the latest delayed semantic information, a cross-frame segmentation strategy is employed for image processing, and real-time semantic information for the current frame is generated by the tracking thread according to the delaysemantic information. Then, the dynamic point set of the current frame is selected and the real motion state of the prior dynamic object in the environment is determined by combining successful tracking count and epipolar constraints. When the object is determined as moving, the object area is further subdivided into rectangular grids, and dynamic feature points are removed with the grid as the minimum unit. Finally, the camera pose is tracked by static feature points and an environment map is constructed. Experiments on TUM RGB-D dynamic scene dataset and real scenes show that the proposed algorithm performs well and its effectiveness is verified.
2023 Vol. 36 (10): 953-966 [Abstract] ( 239 ) [HTML 1KB] [ PDF 6538KB] ( 355 )
模式识别与人工智能
 

Supervised by
China Association for Science and Technology
Sponsored by
Chinese Association of Automation
NationalResearchCenter for Intelligent Computing System
Institute of Intelligent Machines, Chinese Academy of Sciences
Published by
Science Press
 
Copyright © 2010 Editorial Office of Pattern Recognition and Artificial Intelligence
Address: No.350 Shushanhu Road, Hefei, Anhui Province, P.R. China Tel: 0551-65591176 Fax:0551-65591176 Email: bjb@iim.ac.cn
Supported by Beijing Magtech 　Email:support@magtech.com.cn