检索结果-内蒙古大学图书馆

Dynamical causality under invisible confounders

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yan, Jinling Zhang, Shao-Wu Zhang, Chihao Huang, Weitian Shi, Jifan Chen, Luonan MOE Key Laboratory of Information Fusion Technology School of Automation Northwestern Polytechnical University Xi’an710072 China Key Laboratory of Systems Biology Shanghai Institute of Biochemistry and Cell Biology Center for Excellence in Molecular Cell Science Chinese Academy of Sciences Shanghai200031 China NCMIS CEMS RCSDS Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing100190 China School of Mathematics Sciences University of Chinese Academy of Sciences Beijing100049 China School of Computer Science and Engineering South China University of Technology Guangdong Guangzhou510006 China Guangdong Institute of Intelligence Science and Technology Hengqin Guangdong Zhuhai519031 China Research Institute of Intelligent Complex Systems Fudan University Shanghai200433 China State Key Laboratory of Medical Neurobiology MOE Frontiers Center for Brain Science Institutes of Brain Science Fudan University Shanghai200032 China Shanghai Artificial Intelligence Laboratory Shanghai200232 China Key Laboratory of Systems Health Science of Zhejiang Province School of Life Science Hangzhou Institute for Advanced Study University of Chinese Academy of Sciences Chinese Academy of Sciences Hangzhou310024 China

Causality inference is prone to spurious causal interactions, due to the substantial confounders in a complex system. While many existing methods based on the statistical methods or dynamical methods attempt to address misidentification challenges, there remains a notable lack of effective methods to infer causality, in particular in the presence of invisible/unobservable confounders. As a result, accurately inferring causation with invisible confounders remains a largely unexplored and outstanding issue in data science and AI fields. In this work, we propose a method to overcome such challenges to infer dynamical causality under invisible confounders (CIC method) and further reconstruct the invisible confounders from time-series data by developing an orthogonal decomposition theorem in a delay embedding space. The core of our CIC method lies in its ability to decompose the observed variables not in their original space but in their delay embedding space into the common and private subspaces respectively, thereby quantifying causality between those variables both theoretically and computationally. This theoretical foundation ensures the causal detection for any high-dimensional system even with only two observed variables under many invisible confounders, which is actually a long-standing problem in the field. In addition to the invisible confounder problem, such a decomposition actually makes the intertwined variables separable in the embedding space, thus also solving the non-separability problem of causal inference. Extensive validation of the CIC method is carried out using various real datasets, and the experimental results demonstrates its effectiveness to reconstruct real biological networks even with unobserved confounders. © 2024, CC BY-NC-ND.

关键词： Time series

Real-Time Marker Localization Learning for GelStereo Tactile Sensing

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Liu, Shujuan Cui, Shaowei Zhang, Chaofan Cai, Yinghao Wang, Shuo The University of Science and Technology Beijing China The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China The School of Artificial Intelligence University of Chinese Academy of Sciences Beijing100049 China Institute of Automation Chinese Academy of Sciences Beijing100190 China The Center for Excellence in Brain Science and Intelligence Technology Chinese Academy of Sciences Shanghai200031 China

Visuotactile sensing technology is becoming more popular in tactile sensing, but the effectiveness of the existing marker detection/localization methods remains to be further explored. Instead of contour-based blob detection, this paper presents a learning-based marker localization network for GelStereo visuotactile sensing called Marknet. Specifically, the Marknet presents a grid regression architecture to incorporate the distribution of the GelStereo markers. Furthermore, a marker rationality evaluator (MRE) is modelled to screen suitable prediction results. The experimental results show that the Marknet combined with MRE achieves 93.90% precision for irregular markers in contact areas, which outperforms the traditional contour-based blob detection method by a large margin of 42.32%. Meanwhile, the proposed learning-based marker localization method can achieve better real-time performance beyond the blob detection interface provided by the OpenCV library through GPU acceleration, which we believe will lead to considerable perceptual sensitivity gains in various robotic manipulation tasks. Copyright © 2022, The Authors. All rights reserved.

关键词： Machine learning

Prescriptive Manufacturing in Society 5.0: Human Autonomous Organizations and “On-Demand” Smart Services

学校读者我要写书评

暂无评论

IFAC-PapersOnLine 2024年第3期58卷 139-144页

作者： Xiao Wang Jun Huang Jing Yang Xiaoxing Wang Fei-Yue Wang Engineering Research Center of Autonomous Unmanned System Technology Ministry of Education Anhui University Hefei 230601 China Macau University of Science and Technology Macau 999078 China School of artificial Intelligence University of Chinese Academy of Sciences Beijing 100049 China Beijing SANBODY Technology Company Ltd Beijing 214000 China Qingdao Academy of Intelligent Industries Qingdao 266109 China State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing 100190 China

Society 5.0 addresses human behaviors and their consequences in social management, while emphasizing the integration of Cyber-Physical-Social Spaces to achieve a highly connected and circulated society. Prompted by the various social needs expressed on social media platforms, Prescriptive Manufacturing (PM) emerges as a proactive pre-manufacturing paradigm. This paper delves into PM within the context of Society 5.0. It highlights the development of Human Autonomous Organizations (HAOs) and the delivery of “on-demand” smart services, which are customized to meet both individual and broader societal demands. The social sensing framework of social needs across different social media platforms that enables the identification and anticipation of consumer demands in real-time is illustrated. This data-driven insight leverages artificial intelligence and robotics to facilitate customized design and on-demand production, aligning product creation with real user requirements. Furthermore, we discuss the implementation of human-centered smart services in PM, such as virtual try-on, virtual fitting, personalized retrofit, user-friendly scheduling, and proximity logistics, which not only enhance consumer experience but also optimize resource efficiency By examining the symbiosis between technological advancements and social need fulfillment, this study sheds light on how PM and HAOs can foster a more responsive, sustainable, and inclusive economy in society, ultimately contributing to a higher quality of life and sustainable development goals of Society 5.0.

关键词： Social Needs Sensing Human Autonomous Organization Customized Design personalized fine-tuning prescriptive manufacturing Human-Centric Smart Services

Traj-LLM: A New Exploration for Empowering Trajectory Prediction with Pre-trained Large Language Models

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lan, Zhengxing Li, Hongbo Liu, Lingshan Fan, Bo Lv, Yisheng Ren, Yilong Cui, Zhiyong The School of Transportation Science and Engineering Beihang University Beijing100191 China The State Key Lab of Intelligent Transportation System Beijing100191 China The Research Institute for Frontier Science Beihang University Beijing100191 China The Beijing Key Laboratory of Traffic Engineering College of Metropolitan Transportation Beijing University of Technology Beijing100124 China The State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China The School of Artificial Intelligence University of Chinese Academy of Sciences Beijing100049 China The Zhongguancun Laboratory Beijing100194 China The Beihang Hangzhou Innovation Institute Yuhang Hangzhou310023 China

Predicting the future trajectories of dynamic traffic actors is a cornerstone task in autonomous driving. Though existing notable efforts have resulted in impressive performance improvements, a gap persists in scene cognitive and understanding of the complex traffic semantics. This paper proposes Traj-LLM, the first to investigate the potential of using Large Language Models (LLMs) without explicit prompt engineering to generate future motion from agents’ past/observed trajectories and scene semantics. Traj-LLM starts with sparse context joint coding to dissect the agent and scene features into a form that LLMs understand. On this basis, we innovatively explore LLMs’ powerful comprehension abilities to capture a spectrum of high-level scene knowledge and interactive information. Emulating the human-like lane focus cognitive function and enhancing Traj-LLM’s scene comprehension, we introduce lane-aware probabilistic learning powered by the pioneering Mamba module. Finally, a multi-modal Laplace decoder is designed to achieve scene-compliant multi-modal predictions. Extensive experiments manifest that Traj-LLM, fortified by LLMs’ strong prior knowledge and understanding prowess, together with lane-aware probability learning, outstrips state-of-the-art methods across evaluation metrics. Moreover, the few-shot analysis further substantiates Traj-LLM’s performance, wherein with just 50% of the dataset, it outperforms the majority of benchmarks relying on complete data utilization. This study explores equipping the trajectory prediction task with advanced capabilities inherent in LLMs, furnishing a more universal and adaptable solution for forecasting agent motion in a new way. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantics

MixedFusion: 6D Object Pose Estimation from Decoupled RGB-Depth Features

学校读者我要写书评

暂无评论

MixedFusion: 6D Object Pose Estimation from Decoupled RGB-De...

International Conference on Pattern Recognition

作者： Hangtao Feng Lu Zhang Xu Yang Zhiyong Liu The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Center for Excellence in Brain Science and Intelligence Technology Chinese Academy of Sciences

Estimating the 6D pose of objects is an important process for intelligent systems to achieve interaction with the real-world. As the RGB-D sensors become more accessible, the fusion-based methods have prevailed, since the point clouds provide complementary geometric information with RGB values. However, due to the difference in feature space between color image and depth image, the network structures that directly perform point-to-point matching fusion do not effectively fuse the features of the two. In this paper, we propose a simple but effective approach, named MixedFusion. Different from the prior works, we argue that the spatial correspondence of color and point clouds could be decoupled and reconnected, thus enabling a more flexible fusion scheme. By performing the proposed method, more informative points can be mixed and fused with rich color features. Extensive experiments are conducted on the challenging LineMod and YCB-Video datasets, which shows that our method significantly boosts the performance without introducing extra overheads. Furthermore, when the minimum tolerance of metric narrows, the proposed approach performs better for the high-precision demands.

关键词： Measurement Three-dimensional displays Image color analysis Fuses Pose estimation Color Pattern recognition

Model Free Adaptive Control Algorithm based on ReOSELM for Autonomous Driving Vehicles

学校读者我要写书评

暂无评论

Model Free Adaptive Control Algorithm based on ReOSELM for A...

第40届中国控制会议

作者： Xiaofei Zhang Hongbin Ma Zhichao Wang Mingyu Fan Bolin Zhao School of Automation Beijing Institute of Technology State Key Laboratory of Intelligent Control and Decision of Complex Systems Beijing Institute of Technology The College of Computer Science and Artificial Intelligence Wenzhou University

Different road conditions and dynamic environment bring significant challenges to the control system of autonomous driving vehicle（ADV）. As is known, historical data collected from ADV contains valuable information about control systems,therefore, it is a promising thing to study adaptive control algorithms that have certain learning ability. In order to improve the control performance of ADV and the efficiency in data usage, in this paper, a model free adaptive control algorithm based on regularized online sequential extreme learning machine（ReOSELM） is introduced, it is difficult to analyze the algorithm based on neural network, and the system stability by improved update algorithm of ReOSELM is proved. Simulation results indicate that the proposed algorithm is effective in improving control precision when ADV is turning, and experimental results on an autonomous driving vehicle show that this algorithm is effective in real environment.

关键词： data-driven control autonomous driving vehicle model free adaptive control regularized online sequential extreme learning machine

Learning Disentangled Representation for One-Shot Progressive Face Swapping

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Li, Qi Wang, Weining Xu, Chengzhong Sun, Zhenan Yang, Ming-Hsuan State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China The School of Artificial Intelligence University of Chinese Academy of Sciences Beijing100049 China The Laboratory of Cognition and Decision Intelligence for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China The State Key Laboratory of Internet of Things for Smart City University of Macau 999078 China The Department of Computer Science and Engineering University of California MercedCA95340 United States Department of Computer Science and Engineering Yonsei University Korea Republic of

Although face swapping has attracted much attention in recent years, it remains a challenging problem. Existing methods leverage a large number of data samples to explore the intrinsic properties of face swapping without considering the semantic information of face images. Moreover, the representation of the identity information tends to be fixed, leading to suboptimal face swapping. In this paper, we present a simple yet efficient method named FaceSwapper, for one-shot face swapping based on Generative Adversarial Networks. Our method consists of a disentangled representation module and a semantic-guided fusion module. The disentangled representation module comprises an attribute encoder and an identity encoder, which aims to achieve the disentanglement of the identity and attribute information. The identity encoder is more flexible, and the attribute encoder contains more attribute details than its competitors. Benefiting from the disentangled representation, FaceSwapper can swap face images progressively. In addition, semantic information is introduced into the semantic-guided fusion module to control the swapped region and model the pose and expression more accurately. Experimental results show that our method achieves state-of-the-art results on benchmark datasets with fewer training samples. Our code is publicly available at https://***/liqi-casia/FaceSwapper. Copyright © 2022, The Authors. All rights reserved.

关键词： Generative adversarial networks

Parameter estimation survey for multi-joint robot dynamic calibration case study

学校读者我要写书评

暂无评论

science China(Information sciences) 2019年第10期62卷 102-116页

作者： Shaolin ZHANG Shuo WANG Fengshui JING Min TAN The State Key Laboratory of Management and Control for Complex Systems Institute of AutomationChinese Academy of Sciences University of Chinese Academy of Sciences CAS Center for Excellence in Brain Science and Intelligence Technology

Accurate model parameters are the basis of robot dynamics. Many linear and nonlinear models have been proposed to calibrate the inertial parameters and friction parameters of multi-joint ***, methods of choosing a model and calculating its parameters still have few summaries. This paper reviews typical linear/nonlinear models and different calculation methods for robot dynamic *** simulations, the features of different methods are analyzed, including torque error, parameter error, model adaptability, solution time, and anti-interference ability of the calibration results. Finally, an experiment performed on a six-degree-of-freedom industrial manipulator is used as an example to illustrate how to select the model for a specified robot. These comparisons and experiments provide references for the parameter calibration of multi-joint robots.

关键词： dynamic parameter calibration friction calibration robot dynamics industrial manipulator dynamic models

FusionPlanner: A Multi-task Motion Planner for Mining Trucks via Multi-sensor Fusion

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Teng, Siyu Li, Luxi Li, Yuchen Hu, Xuemin Li, Lingxi Ai, Yunfeng Chen, Long Department of Computer Science Hong Kong Baptist University Hong Kong Department of Computer Science BNU-HKBU United International College Zhuhai China The School of Artificial Intelligence Hubei University Wuhan China Waytous Inc. Haidian Distinct Beijing China Indianapolis United States School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences China

In recent years, significant achievements have been made in motion planning for intelligent vehicles. However, as a typical unstructured environment, open-pit mining attracts limited attention due to its complex operational conditions and adverse environmental factors. A comprehensive paradigm for unmanned transportation in open-pit mines is proposed in this research. Firstly, we propose a multi-task motion planning algorithm, called FusionPlanner, for autonomous mining trucks by the multi-sensor fusion method to adapt both lateral and longitudinal control tasks for unmanned transportation. Then, we develop a novel benchmark called MiningNav, which offers three validation approaches to evaluate the trustworthiness and robustness of well-trained algorithms in transportation roads of open-pit mines. Finally, we introduce the Parallel Mining Simulator (PMS), a new high-fidelity simulator specifically designed for open-pit mining scenarios. PMS enables the users to manage and control open-pit mine transportation from both the single-truck control and multi-truck scheduling perspectives. The performance of FusionPlanner is tested by MiningNav in PMS, and the empirical results demonstrate a significant reduction in the number of collisions and takeovers of our planner. We anticipate our unmanned transportation paradigm will bring mining trucks one step closer to trustworthiness and robustness in continuous round-the-clock unmanned transportation. Copyright © 2023, The Authors. All rights reserved.

关键词： Motion planning