检索结果-内蒙古大学图书馆

International Joint Conference on Neural Networks (IJCNN)

作者： Zikang Wang Linjing Li Daniel Dajun Zeng The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China Shenzhen Artificial Intelligence and Data Science Institute (Longhua) Shenzhen China

ISBN: (数字)9781728169262

ISBN: (纸本)9781728169279

Knowledge graph completion, one of the most important research questions in knowledge graphs, aims at predicting missing links in a given graph. Current mainstream approaches adopt high-quality embeddings of entities and relations of the graph to improve their performances. However, it is not easy to devise a universal embedding learner that can fit various scenarios. In this paper, we propose a general-purpose framework which can be employed to improve the performance of knowledge graph completion. Specifically, given an arbitrary knowledge graph completion model, we first run the original model to get a ranked entity list. Then, we combine the query and the top ranked entities with attention mechanism, re-rank all these entities by feeding the combined vector into a neural network. The proposed re-ranking phase can be conveniently added to a variety of models to improve their performance without substantial modification. We conduct experiments on four datasets: WN18, FB15k, WN18RR, and FB15k-237. We choose TransE, TransH, TransD, DistMult, and ANALOGY as base models. Experiments on these datasets and models validate the effectiveness of the proposed re-ranking framework. We further explore the influence of the number of top ranked entities used in the re-ranking phase. We also test other attention mechanism to determine the most effective one, and found that vanilla attention mechanism can balance accuracy and complexity.

关键词： Computational modeling Task analysis Neural networks Semantics complexity theory Measurement Magnetic heads

来源：评论

学校读者我要写书评

暂无评论

Identification of muscle morphology with noncontact capacitive sensing: Preliminary study

Identification of muscle morphology with noncontact capaciti...

引用

Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)

作者： Enhao Zheng Jiacheng Wan Dongfang Xu Qining Wang Hong Qiao The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China School of Information Engineering China university of geosciences (Beijing) Beijing China The Robotics Research Group College of Engineering Peking University Beijing China

ISBN: (数字)9781728119908

ISBN: (纸本)9781728119915

Human-machine interface with muscle signals serves as an important role in the field of wearable robotics. To compensate for the limitations of the existing surface Electromyography (sEMG) based technologies, we previously proposed a noncontact capacitive sensing approach that could record the limb shape changes. The sensing approach frees the human skin from contacting to the metal electrodes, thus enabling the measurement of muscle signals by dressing the sensing front-ends outside of the clothes. We validated the capacitive sensing in human motion intent recognition tasks with the wearable robots and produced comparable results to existing studies. However, the biological significance of the capacitance signals is still unrevealed, which is an indispensable issue for robot intuitive control. In this study, we address the problems of identifying the relationships between the muscle morphological parameters and the capacitance signals. We constructed a measurement system that recorded the noncontact capacitive sensing signals and the muscle ultrasound (US) images simultaneously. With the designed device, five subjects were employed and the US images from the gastrocnemius muscle (GM) and the tibialis anterior (TA) muscle during level walking were sampled. We fitted the calculated muscle morphological parameters (the pinnation angles and the muscle fascicle length) and the capacitance signals of the same gait phases. The results demonstrated that at least one-channel capacitance signal strongly correlated to the muscle morphological parameters (R 2 > 0.5, quadratic regression). The average R 2 s of the most correlated channels were up to 0.86 for pinnation angles and 0.83 for the muscle fascicle length changes. The interesting findings in this preliminary study suggest the biological physical significance of the capacitance signals during human locomotion. Future efforts are worth being paid in this new research direction for more promising results.

关键词： Muscles Robot sensing systems Capacitance Legged locomotion Electrodes Ultrasonic variables measurement

来源：评论

学校读者我要写书评

暂无评论

Feature Aggregation Attention Network for Single Image Dehazing

Feature Aggregation Attention Network for Single Image Dehaz...

引用

IEEE International Conference on Image Processing

作者： Lan Yan Wenbo Zheng Chao Gou Fei-Yue Wang The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China School of Software Engineering Xi’an Jiaotong University Xi’an China School of Intelligent Systems Engineering Sun Yat-sen University Guangzhou China

ISBN: (数字)9781728163956

ISBN: (纸本)9781728163963

Due to its ill-posed nature, single image dehazing is a challenging problem. In this paper, we propose an end-to-end feature aggregation attention network (FAAN) for single image dehazing. It incorporates the idea of attention mechanism and residual learning and can adaptively aggregate different level features. In particular, in the proposed FANN, we design a novel block structure consisting of feature attention module, smoothed dilated convolution and local residual learning. The local residual learning allows the less useful information to be bypassed through multiple skip connections. The feature attention module is designed to assign more weight to important features. The smoothed dilated convolution is adopted to enlarge the receptive field without the negative influence of gridding artifacts. The experiments on the RESIDE dataset show that the proposed approach acquires state-of-the-art performance in both qualitative and quantitative measures.

关键词： Convolution Atmospheric modeling Kernel Image color analysis Task analysis Training

来源：评论

学校读者我要写书评

暂无评论

Algorithm design of traffic incident automatic detection based on mobile detection

Algorithm design of traffic incident automatic detection bas...

引用

IEEE International Conference on Service Operations and Logistics, and Informatics, SOLI

作者： Zi Zhang Xiaoli Lin Bin Hu Communications Commission of Guangzhou Municipality Guangzhou Communication Information Construction Investment and Operation Company Limited Guangzhou China Transportation Management General Station of Guangzhou Haizhu Guangzhou China State Key Laboratory of Intelligent Control and Management of Complex Systems Institute of Automation Chinese Academy and Sciences Beijing China

Most researches of traffic incident auto-detection are based on the data from fixed detectors, which are limited by costs and position. In order to resolve this problem, existing algorithms of traffic incident automatic detection are analyzed and compared, and an algorithm of traffic incident auto-detection are provided based on mobile-detection technology. The traffic data are grouped in 5-min intervals, analyzed by a three-layer BP neural network, and utilized for traffic incident detection. 16 traffic incidents of different locations and different levels are modeled in the simulation experiment based on VISSIM, and detection rate, false alarm rate and average detection time are adopted as indicators to evaluate the algorithm. Finally, the algorithm is proved to be effective and applicable in practice.

关键词： Three dimensional displays Bayesian methods Detectors

来源：评论

学校读者我要写书评

暂无评论

MatchMesh: Knowledge-based 3D Point Cloud Meshing Using Divide-and-conquer Deformation

MatchMesh: Knowledge-based 3D Point Cloud Meshing Using Divi...

引用

IEEE International Conference on Networking, Sensing and control

作者： Ying Tang Shengtao Sun Ben Wu Rowan University Glassboro USA State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China Institute of Smart Education Qingdao Academy of Intelligent Industries Qingdao China Rowan University Glassboro NJ USA

ISBN: (数字)9781728168555

ISBN: (纸本)9781728168562

The reconstruction of surface mesh from point cloud is compute-intensive but also very important step in the remanufacturing and personalization industries. With more 3D scanners providing lower cost and higher resolution, further detailed point clouds can be gathered without so much effort as before. In manufacturing, there are databases which contain the origin 3D design models of the products. How to utilize the design model data for swift production of related products remains a problem for remanufacturing and customization. In order to develop a knowledge-based way of handling this problem, editing or deforming an existing mesh to match the target is an effective way of easing the workload. In this paper, we introduce a divide-and -conquer process which segments the depth scan data and then find the best match in the database as its source of deformation. The segmentation is performed on 3D point level using global features extracted by 3D CNN. After that we find best match to our knowledge with the same features to acquire a fast meshing of the target object by deforming the existing parts from the match. The deformation of parts are being done sequentially. For further performance improvement, we present a deformation training method employing transfer learning on segment editing process.

关键词： Training Surface reconstruction Three-dimensional displays Databases Knowledge based systems Feature extraction Strain

来源：评论

学校读者我要写书评

暂无评论

A Collision-Free Person-Following Approach Based on Path Planning

A Collision-Free Person-Following Approach Based on Path Pla...

引用

IEEE International Conference on Mechatronics and Automation

作者： Lei Pang Zhiqiang Cao Junzhi Yu Weimin Zhang Xuechao Chen State Key Laboratory of Management and Control for Complex Systems Institute of Automation School of Artificial Intelligence University of Chinese Academy of Sciences BIC-ESAT College of Engineering Peking University Beijing Advanced Innovation Center for Intelligent Robots and Systems Beijing Institute of Technology

ISBN: (数字)9781728164168

ISBN: (纸本)9781728164175

Person following is a crucial capability for mobile robots in tasks of human assistance and cooperation. In this paper, a person-following approach based on path planning is proposed for mobile robots, which takes person following and obstacle avoidance into account in a unified framework. The proposed approach is comprised of a sensing module and a planning module. The sensing module with a camera and a 3D LiDAR is employed to locate the target person and perceive obstacles. For the planning module, the combination of global planner and local planner effectively outputs collision-free paths and control commands for a mobile robot to keep following the target person. Experiments were conducted to evaluate the performance of the proposed approach.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Hierarchical Learning from Demonstrations for Long-Horizon Tasks

Hierarchical Learning from Demonstrations for Long-Horizon T...

引用

IEEE International Conference on Robotics and Automation (ICRA)

作者： Boyao Li Jiayi Li Tao Lu Yinghao Cai Shuo Wang China Academy of Launch Vehicle Technology Beijing China State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China Center for Excellence in Brain Science and Intelligence Technology Chinese Academy of Sciences Shanghai China

Although reinforcement learning (RL) has achieved great success in robotic manipulation skills learning, it is still challenging for long-horizon tasks. Combining RL with demonstrations is an effective solution. In this paper, we propose a novel hierarchical learning from demonstrations method for long-horizon tasks, which leverages (i) object-centered segmentation of demonstrations to automatically segment the teaching trajectories into episodes. (ii) a bi-level hierarchical imitation learning method with a parallel training mechanism to train the two-level policies simultaneously. Experimental results on three challenging long-horizon tasks with sparse rewards show that our proposed method significantly outperforms state-of-art approaches in terms of both sample-efficiency and success rate. Moreover, our method is the only one which achieves satisfactory performance in tasks of multi-object stack and multi-object push&stack.

关键词： Deep learning Training Learning systems Automation Conferences Reinforcement learning Trajectory

来源：评论

学校读者我要写书评

暂无评论

Forest Representation Learning with Multiscale Contour Feature Learning for Leaf Cultivar Classification

Forest Representation Learning with Multiscale Contour Featu...

引用

IEEE International Conference on Bioinformatics and Biomedicine

作者： Wenbo Zheng Chao Gou Lan Yan School of Software Engineering Xi'an Jiaotong University Chinese Academy of Sciences School of Intelligent Systems Engineering Sun Yat-sen University Chinese Academy of Sciences The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences

ISBN: (纸本)9781728118680

Automated plant species identification system could help botanists and layman in identifying plant species rapidly. Deep learning is robust for feature extraction as it is superior in providing deeper information of images. However, deep neural networks based leaf cultivar classification model can not deeply mine the contour features of leaf images. Besides, the deep neural networks are very data-hungry due to the large complexity of the models, which means that the model's performance can decrease significantly when the size of the training data decreases. In this paper, we propose the leaf cultivar classification model based on forest representation learning and multiscale contour feature learning. Firstly, we build the multiscale contour feature learning to extract the contour features of leaf images. Then, the original input features are transformed by the data augmentation method to enhance the ability of feature expression, and the autoencoder by forest performs the data of dimensionality reduction on the features. Finally, we use the improved deep forest algorithm and autoencoder by forest to build forest representation learning for classification. The experimental results show that the proposed algorithm has higher performance than the original deep forest (gcForest) algorithm and other existing start-of-art algorithms, and has higher performance and efficiency than other deep learning algorithms.

关键词： Leaf Cultivar Classification Deep Learning Deep Forest AutoEncoder by Forest Data Augmentation contour Forest Learning plant species foliage Botanists Taxonomy dimensionality reduction algorithms cultivars HIGH-PERFORMANCE

来源：评论

学校读者我要写书评

暂无评论

Improving the Data Quality for Credit Card Fraud Detection

Improving the Data Quality for Credit Card Fraud Detection

引用

IEEE International Conference on Intelligence and Security Informatics, ISI

作者： Rongrong Jing Hu Tian Yidi Li Xingwei Zhang Xiaolong Zheng Zhu Zhang Daniel Zeng The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China Shenzhen Artificial Intelligence and Data Science Institude (Longhua) Shenzhen China

ISBN: (数字)9781728188003

ISBN: (纸本)9781728188010

Label imbalance and data missing are two major challenges in the problem of credit card fraud detection. However, existing matrix completion algorithms are generally difficult and cannot be easily applied to real-world credit card fraud detection since the scale of the normally used dataset is oversized. In this paper, we develop a spectral regularization algorithm to complete the large-scale sparse matrices, and further utilize an over-sampling algorithm to tackle the problem of the imbalance between positive and negative samples. Experimental results on a real-world dataset demonstrate that our model can outperform the state-of-the-art baseline methods. The proposed method could also be extended to other large-scale scenarios where data is missing or labels are imbalanced.

关键词： Data integrity Conferences Credit cards Sparse matrices Security Informatics

来源：评论

学校读者我要写书评

暂无评论

UNMAS: Multi-Agent Reinforcement Learning for Unshaped Cooperative Scenarios

arXiv

引用

arXiv 2022年

作者： Chai, Jiajun Li, Weifan Zhu, Yuanheng Zhao, Dongbin Ma, Zhe Sun, Kewu Ding, Jishiyu State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China School of Artificial Intelligence University of Chinese Academy of Sciences Beijing100049 China Xlab The Second Academy of CASIC Beijing100854 China

Multi-agent reinforcement learning methods such as VDN, QMIX, and QTRAN that adopt centralized training with decentralized execution (CTDE) framework have shown promising results in cooperation and competition. However, in some multi-agent scenarios, the number of agents and the size of action set actually vary over time. We call these unshaped scenarios, and the methods mentioned above fail in performing satisfyingly. In this paper, we propose a new method called Unshaped Networks for Multi-Agent systems (UNMAS) that adapts to the number and size changes in multi-agent systems. We propose the self-weighting mixing network to factorize the joint action-value. Its adaption to the change in agent number is attributed to the nonlinear mapping from each-agent Q value to the joint action-value with individual weights. Besides, in order to address the change in action set, each agent constructs an individual action-value network that is composed of two streams to evaluate the constant environment-oriented subset and the varying unit-oriented subset. We evaluate UNMAS on various StarCraft II micro-management scenarios and compare the results with several state-of-the-art MARL algorithms. The superiority of UNMAS is demonstrated by its highest winning rates especially on the most difficult scenario 3s5z-vs-3s6z. The agents learn to perform effectively cooperative behaviors while other MARL algorithms fail in. Copyright © 2022, The Authors. All rights reserved.

关键词： Multi agent systems

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：