检索结果-内蒙古大学图书馆

Variational autoencoders for deforming 3d mesh models

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Tan, Qingyang Gao, Lin Lai, Yu-Kun Xia, Shihong Beijing Key Laboratory of Mobile Computing and Pervasive Device Institute of Computing Technology Chinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences School of Computer Science and Informatics Cardiff University

3D geometric contents are becoming increasingly popular. In this paper, we study the problem of analyzing deforming 3D meshes using deep neural networks. Deforming 3D meshes are flexible to represent 3D animation sequences as well as collections of objects of the same category, allowing diverse shapes with large-scale non-linear deformations. We propose a novel framework which we call mesh variational autoencoders (mesh VAE), to explore the probabilistic latent space of 3D surfaces. The framework is easy to train, and requires very few training examples. We also propose an extended model which allows flexibly adjusting the significance of different latent variables by altering the prior distribution. Extensive experiments demonstrate that our general framework is able to learn a reasonable representation for a collection of deformable shapes, and produce competitive results for a variety of applications, including shape generation, shape interpolation, shape space embedding and shape exploration, outperforming state-ofthe-art methods. Copyright © 2017, The Authors. All rights reserved.

关键词： Deep neural networks

Mesh-based autoencoders for localized deformation component analysis

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Tan, Qingyang Gao, Lin Lai, Yu-Kun Yang, Jie Xia, Shihong Beijing Key Laboratory of Mobile Computing and Pervasive Device Institute of Computing Technology Chinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences School of Computer Science & Informatics Cardiff University

Spatially localized deformation components are very useful for shape analysis and synthesis in 3D geometry processing. Several methods have recently been developed, with an aim to extract intuitive and interpretable deformation components. However, these techniques suffer from fundamental limitations especially for meshes with noise or large-scale deformations, and may not always be able to identify important deformation components. In this paper we propose a novel mesh-based autoencoder architecture that is able to cope with meshes with irregular topology. We introduce sparse regularization in this framework, which along with convolutional operations, helps localize deformations. Our framework is capable of extracting localized deformation components from mesh data sets with large-scale deformations and is robust to noise. It also provides a nonlinear approach to reconstruction of meshes using the extracted basis, which is more effective than the current linear combination approach. Extensive experiments show that our method outperforms state-of-the-art methods in both qualitative and quantitative evaluations. Copyright © 2017, The Authors. All rights reserved.

关键词： Mesh generation

Fault-tolerant fuzzy sliding-mode control for a class of MIMO underactuated nonlinear systems

学校读者我要写书评

暂无评论

Fault-tolerant fuzzy sliding-mode control for a class of MIM...

International Conference on Systems and control (ICSC)

作者： Faten Baklouti Sinda Aloui Abdessattar Chaari Ahmed El Hajjaji Universite de Sfax Sfax TN Laboratory of Sciences and Techniques of Automatic control computer engineering (Lab-STA) University of Sfax Sfax Tunisia University of Picardie Jules Verne Lab. MIS Amiens France

ISBN: (纸本)9781509039616

This paper focuses on the synthesis of adaptive fuzzy fault tolerant controller which combines a sliding mode controller and a Proportional Integral controller (PI) for a class of MIMO underactuated nonlinear systems with external disturbances. The proposed control law treats the underactuated systems on their original non square form without squaring them by adding or eliminating variables. The uncertain system functions are predicted by fuzzy logic systems and free parameters can be updated on-line by adaptive laws based on Lyapunov theory. The overall adaptive fuzzy sliding mode scheme guarantees the global stability of the resulting closed-loop system as well as the convergence of the reference tracking error vector. The validity of the proposed approach is shown by computer simulations.

关键词： Adaptive systems Fault tolerant systems Nonlinear systems MIMO Robustness Fault tolerant control

A Retinal Blood Vessel Segmentation Approach Based on Top-hat Transformation

学校读者我要写书评

暂无评论

A Retinal Blood Vessel Segmentation Approach Based on Top-ha...

2017 2nd International Conference on Mechatronics and Information technology(ICMIT 2017)

作者： Weihua Wang Wenyuan Wu Jiao Yin Chongqing Key Laboratory of Automated Reasoning and Cognition Chongqing Institute of Green and Intelligent Technology Chinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences School of Software Engineering Chongqing University of Arts and Sciences

The analysis of retinal blood vessel plays an important role in aided diagnosis, segmenting blood from retinal fundus images is the first operation. This paper proposes a novel approach for retinal blood vessel automatic segmentation. First, the top-hat transformation is used for blood vessel enhancement. Second, using double-threshold to segment the vessels from the enhanced fundus image, the high threshold is used for main vessel segmentation, and the low threshold is used for finer vessel segmentation. Third, fusing the finer blood vessels into main blood vessels. The experiment is tested in public fundus image database STARE. The result of the experiment shows that the method obtains good visual effect.

关键词： Double threshold Blood Vessel Segmentation Fundus Image

Design of Low-cost Position Differential Positioning System Based on STM32

学校读者我要写书评

暂无评论

Design of Low-cost Position Differential Positioning System ...

Chinese Automation Congress

作者： Haijun Fu Ming Qiu Xisong Dong Gang Xiong Haonan Xiong Xiang Su Shichao Chen Zhen Shen Xiujiang Guo School of Artificial Intelligence Beijing Union University Beijing China The Cloud Computing Center Chinese Academy of Sciences Dongguan China The Qingdao Academy of Intelligent Industries Macau University of Science and Technology Qingdao China Beijing Union University Beijing China School of Electrical Engineering and Electronic Information Xihua University Chengdu China The State Key Laboratory for Management and Control of Complex Systems Chinese Academy of Sciences Beijing China The Beijing Engineering Research Center of Intelligent Systems and Technology Chinese Academy of Sciences Beijing China Department of Computer Science University of Helsinki Fin Finland Measurement Center Station PetroChina Huabei Oilfield Company Renqiu China

There exist some obstacles to realize high-precision positioning with using the position differential technique due to the limiting factors. However, with the advancement of chip performance and the establishment of multiple satellite navigation systems, position differential positioning can be realized to meet the civilian high-precision positioning demand while significantly reducing costs. This paper is to explore the engineering application and propose a new method. Through the experiments and data analysis, we have found the error correlation between two receivers in a small area and designed a set of positioning system, and programmed in the Keil μVision5 in C language. The system consists of the current mainstream low-cost STM32, UBLOX NEO 6M chip, NRF24L01 chip, GPS antenna and power supply module. Each two receivers can form a differential positioning system. At present, the system in the low-speed status is, in the urban environment, basically reaching within 0.5 - 3 m of the positioning error.

关键词： Position differential positioning Low cost Error correlation GPS STM32

Random asynchronous iterations in distributed coordination algorithms

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Chen, Yao Xia, Weiguo Cao, Ming Lü, Jinhu Department of Computer Science Southwestern University of Finance and Economics Chengdu611130 China School of Control Science and Engineering Dalian University of Technology Dalian116024 China Faculty of Science and Engineering ENTEG University of Groningen Groningen9747 AG Netherlands School of Automation Science and Electrical Engineering State Key Laboratory of Software Development Environment Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University Beijing100191 China Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing100190 China

Distributed coordination algorithms (DCA) carry out information processing processes among a group of networked agents without centralized information fusion. Though it is well known that DCA characterized by an SIA (stochastic, indecomposable, aperiodic) matrix generate consensus asymptotically via synchronous iterations, the dynamics of DCA with asynchronous iterations have not been studied extensively, especially when viewed as stochastic processes. This paper aims to show that for any given irreducible stochastic matrix, even non-SIA, the corresponding DCA lead to consensus successfully via random asynchronous iterations under a wide range of conditions on the transition probability. Particularly, the transition probability is neither required to be independent and identically distributed, nor characterized by a Markov chain. Copyright © 2018, The Authors. All rights reserved.

关键词： Stochastic systems

Secure quantum communication in the presence of phase- and polarization-dependent loss

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Li, Chenyang Curty, Marcos Xu, Feihu Bedroya, Olinka Lo, Hoi-Kwong Center for Quantum Information and Quantum Control Department of Electrical & Computer Engineering Department of Physics University of Toronto TorontoONM5S 3G4 Canada Escuela de Ingeniería de Telecomunicación Department of Signal Theory and Communications University of Vigo VigoE-36310 Spain Shanghai Branch Hefei National Laboratory for Physical Sciences at the Microscale University of Science and Technology of China Shanghai201315 China

Silicon photonics holds the promise of the miniaturization of quantum communication devices. Recently, silicon chip optical transmitters for quantum key distribution (QKD) have been built and demonstrated experimentally. Nonetheless, these silicon chips suffer substantial phase- and polarization-dependent loss (PDL) which, if unchecked, could compromise the security of QKD systems because of overestimating the secret key rate. Here, we first restore the security by regarding the single photons without phase and polarization dependence as untagged and secure qubits. Next, by using a post-selection technique, one could implement a secure QKD protocol that provides a high key generation rate even in the presence of severe phase and polarization dependent loss. Our solution is simple to realize in a practical experiment as it does not require any hardware modification. Copyright © 2018, The Authors. All rights reserved.

关键词： Quantum cryptography

Acoustic Model Compression with Knowledge Transfer

学校读者我要写书评

暂无评论

Acoustic Model Compression with Knowledge Transfer

第十四届全国人机语音通讯学术会议

作者： Jiangyan Yi Jianhua Tao Zhengqi Wen Ya Li Hao Ni National Laboratory of Pattern Recognition Institute of AutomationChinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences CAS Center for Excellence in Brain Science and Intelligence Technology Institute of AutomationChinese Academy of Sciences

Mobile devices have limited computing power and limited ***,large deep neural network(DNN) based acoustic models are not well suited for application on mobile *** order to alleviate this problem,this paper proposes to compress acoustic models by using knowledge *** approach forces a large teacher model to transfer generalized knowledge to a small student *** student model is trained with a linear interpolation of hard probabilities and soft probabilities to learn generalized knowledge from the teacher *** hard probabilities are generated from a Gaussian mixture model hidden Markov model(GMM-HMM) *** soft probabilities are computed from a teacher model(DNN or RNN).Experiments on AMI corpus show that a small student model obtains 2.4%relative WER improvement over a large teacher model with almost 7.6 times compression ratio.

关键词： model compression knowledge transfer deep neural networks automatic speech recognition

High-level activity recognition based on analysis of spatio-Temporal contexts

学校读者我要写书评

暂无评论

High-level activity recognition based on analysis of spatio-...

2017 IEEE SmartWorld Ubiquitous Intelligence and Computing, Advanced and Trusted Computed, Scalable Computing and Communications, Cloud and Big Data Computing, Internet of People and Smart City Innovation, SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI 2017

作者： Lang, Ruixiang Ye, Jian Huang, Bin Sun, Yangwei Institute of Computing Technology Chinese Academy of Sciences Beijing China School of Computer and Control Engineering University of Chinese Academy of Sciences Beijing China Beijing Key Laboratory of Mobile Computing and Pervasive Device Beijing China

ISBN: (纸本)9781538604342

High-level activity recognition (HAR) has got widely used in healthcare, recommendation, activity sensitive applications etc. Because of the complexity, diversity and individuality of high-level activity, researchers are faced with the challenge of increasing recognition accuracy under realistic conditions. In daily life, high-level activity has a strong association with spatial location since people are used to doing an activity in a specific location. Furthermore, some high-level activities usually co-occur in temporal sequence, which is supposed to be considered in activity inference. In this paper, we propose an approach of high-level activity recognition based on analysis of spatio-Temporal contexts. The approach extracts Region of Activity (ROA) from spatial locations and discovers Daily Activity Pattern (DAP) from temporal sequence of activities to support the design of classifiers of high-level activity. The experiments show that the proposed approach bring about promising performance improvements on activity recognition. The fusion of ROA and DAP gain an advantage over the existing work in accuracy. © 2017 IEEE.

关键词： Pattern recognition