检索结果-内蒙古大学图书馆

Document hashing with mixture-prior generative models

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Dong, Wei Su, Qinliang Shen, Dinghan Chen, Changyou School of Data and Computer Science Sun Yat-sen University Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China ECE Department Duke University CSE Department SUNY Buffalo United States

Hashing is promising for large-scale information retrieval tasks thanks to the efficiency of distance evaluation between binary codes. Generative hashing is often used to generate hashing codes in an unsupervised way. However, existing generative hashing methods only considered the use of simple priors, like Gaussian and Bernoulli priors, which limits these methods to further improve their performance. In this paper, two mixture-prior generative models are proposed, under the objective to produce high-quality hashing codes for documents. Specifically, a Gaussian mixture prior is first imposed onto the variational auto-encoder (VAE), followed by a separate step to cast the continuous latent representation of VAE into binary code. To avoid the performance loss caused by the separate casting, a model using a Bernoulli mixture prior is further developed, in which an endto- end training is admitted by resorting to the straight-through (ST) discrete gradient estimator. Experimental results on several benchmark datasets demonstrate that the proposed methods, especially the one using Bernoulli mixture priors, consistently outperform existing ones by a substantial margin. Copyright © 2019, The Authors. All rights reserved.

关键词： Binary codes

Residual objectness for imbalance reduction

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Chen, Joya Liu, Dong Luo, Bin Peng, Xuezheng Xu, Tong Chen, Enhong Anhui Province Key Laboratory of Big Data Analysis and Application University of Science and Technology of China CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Tencent TaiQ Team

For a long time, object detectors have suffered from extreme imbalance between foregrounds and backgrounds. While several sampling/reweighting schemes have been explored to alleviate the imbalance, they are usually heuristic and demand laborious hyper-parameters tuning, which is hard to achieve the optimality. In this paper, we first reveal that such the imbalance could be addressed in a learning-based manner. Guided by this illuminating observation, we propose a novel Residual Objectness (ResObj) mechanism that addresses the imbalance by end-to-end optimization, while no further hand-crafted sampling/reweighting is required. Specifically, by applying multiple cascaded objectness-related modules with residual connections, we formulate an elegant consecutive refinement procedure for distinguishing the foregrounds from backgrounds, thereby progressively addressing the imbalance. Extensive experiments present the effectiveness of our method, as well as its compatibility and adaptivity for both region-based and one-stage detectors, namely, the RetinaNet-ResObj, YOLOv3-ResObj and FasterRCNN-ResObj achieve relative 3.6%, 3.9%, 3.2% Average Precision (AP) improvements compared with their vanilla models on COCO, respectively. Copyright © 2019, The Authors. All rights reserved.

关键词： Object detection

Discovery of Two-dimensional Hexagonal MBene HfBO and Exploration on its Potential for Lithium-Ion Storage

学校读者我要写书评

暂无评论

Angewandte Chemie 2023年第36期135卷

作者： Nanxi Miao Prof. Yutong Gong Huaiyu Zhang Qing Shen Rui Yang Prof. Jianping Zhou Prof. Hideo Hosono Prof. Junjie Wang State Key Laboratory of Solidification Processing School of Materials Science and Engineering Department Northwestern Polytechnical University Xi'an Shaanxi 710072 P. R. China Contribution: Data curation (lead) Formal analysis (equal) Funding acquisition (equal) Software (lead) Writing - original draft (lead) Contribution: Data curation (equal) Formal analysis (equal) Funding acquisition (equal) Validation (equal) Writing - original draft (equal) Writing - review & editing (equal) Contribution: Formal analysis (equal) Software (equal) Contribution: Formal analysis (equal) Investigation (equal) Software (equal) Contribution: Investigation (equal) School of Physics & Information Technology Shaanxi Normal University Xi'an 710119 P. R. China Contribution: Formal analysis (equal) Investigation (equal) MDX Research Center for Element Strategy Tokyo Institute of Technology 4259 Nagatsuta-cho Midori-ku Yokohama Kanagawa 226-8503 Japan Contribution: Funding acquisition (equal) Investigation (equal) Writing - review & editing (equal) Contribution: Conceptualization (lead) Data curation (equal) Formal analysis (equal) Funding acquisition (equal) Supervision (lead) Writing - original draft (lead) Writing - review & editing (lead)

The practical applications of two-dimensional (2D) transition-metal borides (MBenes) have been severely hindered by the lack of accessible MBenes because of the difficulties in the selective etching of traditional ternary MAB phases with orthorhombic symmetry ( ort -MAB). Here, we discover a family of ternary hexagonal MAB ( h -MAB) phases and 2D hexagonal MBenes ( h -MBenes) by ab initio predictions and experiments. Calculations suggest that the ternary h -MAB phases are more suitable precursors for MBenes than the ort -MAB phases. Based on the prediction, we report the experimental synthesis of h -MBene HfBO by selective removal of In from h -MAB Hf 2 InB 2 . The synthesized 2D HfBO delivered a specific capacity of 420 mAh g −1 as an anode material in lithium-ion batteries, demonstrating the potential for energy-storage applications. The discovery of this h -MBene HfBO added a new member to the growing family of 2D materials and provided opportunities for a wide range of novel applications.

关键词： DFT Calculations High-Throughput Screening Lithium-Ion Batteries Two-Dimensional Materials h-MBenes

Refining HTN methods via task insertion with preferences (ArXiv version)

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Xiao, Zhanhao Wan, Hai Zhuo, Hankui Hankz Herzig, Andreas Perrussel, Laurent Chen, Peilin School of Data and Computer Science Sun Yat-sen University Guangzhou China IRIT Toulouse France University of Toulouse Toulouse France Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China Guangdong Province Key Laboratory of Big Data Analysis and Processing China

Hierarchical Task Network (HTN) planning is showing its power in real-world planning. Although domain experts have partial hierarchical domain knowledge, it is time-consuming to specify all HTN methods, leaving them incomplete. On the other hand, traditional HTN learning approaches focus only on declarative goals, omitting the hierarchical domain knowledge. In this paper, we propose a novel learning framework to refine HTN methods via task insertion with completely preserving the original methods. As it is difficult to identify incomplete methods without designating declarative goals for compound tasks, we introduce the notion of prioritized preference to capture the incompleteness possibility of methods. Specifically, the framework first computes the preferred completion profile w.r.t. the prioritized preference to refine the incomplete methods. Then it finds the minimal set of refined methods via a method substitution operation. Experimental analysis demonstrates that our approach is effective, especially in solving new HTN planning instances. Copyright © 2019, The Authors. All rights reserved.

关键词： Domain Knowledge

Waterline mapping of inland great lake with subpixel accuracy from GF-3 SAR images 6

学校读者我要写书评

暂无评论

Waterline mapping of inland great lake with subpixel accurac...

6th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2019

作者： Li, Ning Niu, Shilin Wang, Robert Wu, Lin Guo, Zhengwei College of Computer and Information Engineering Henan University Kaifeng475004 China Henan Key Laboratory of Big Data Analysis and Processing Henan University Kaifeng475004 China Henan Engineering Research Center of Intelligent Technology and Application Kaifeng475004 China Department of Space Microwave Remote Sensing System Institute of Electronics Chinese Academy of Sciences Beijing100190 China

ISBN: (纸本)9781728129129

High-accuracy waterline mapping with Synthetic Aperture Radar (SAR) images is a challenging task because of the inhomogeneities of SAR imagery caused by the speckle noise and complex terrain. This paper presents a novel method for waterline mapping with subpixel accuracy. The proposed method mainly consists of three steps. Firstly, an improved Non-Local filter is adapted to suppress the speckle noise. Secondly, a Fuzzy C-Means (FCM) clustering algorithm is used to extract the pixel-level waterline. Thirdly, along the pixel-level waterline, a novel subpixel-scale waterline extraction method based on bicubic convolution and Geometric Active Contour (GAC) model is presented to further improve the accuracy of the waterline. Without loss of generality, the Danjiangkou reservoir (DJKR), which is the largest artificial freshwater lake in Asia, is selected as a study case. SAR images from Chinese GaoFen-3 (GF-3) satellite working on different modes are processed to demonstrate the effectiveness of the proposed method. © 2019 IEEE.

关键词： Synthetic aperture radar

Multi-UAVs cooperative task assignment and path planning scheme

学校读者我要写书评

暂无评论

Journal of Physics: Conference Series 2021年第1期1856卷

作者： Xiaoyu Du Qicheng Guo Hui Li Yanyu Zhang International Joint Research Laboratory for Cooperative Vehicular Networks of Henan Henan University Kaifeng Henan 475000 China Henan Key Laboratory of Big Data Analysis and Processing Henan University Kaifeng Henan 475000 China Henan University Minsheng College Kaifeng Henan 475000 China

The task assignment and path planning problems of multiple unmanned aerial vehicles (multi-UAVs) cooperative assistance roadside units (RSUs) for data collection are optimization problems with the goal of minimizing time and energy consumption. This paper proposes a hierarchical optimization scheme for multi-UAVs collaborative assistance RSUs data collection. This solution solves the problems that the number of UAVs needs to be set in advance, the convergence speed is slow when the number of tasks increases, and it is easy to fall into a local optimal solution, and the convergence accuracy is poor. First, the solution uses the K-means algorithm to allocate tasks and group RSUs to find the right number of UAVs to perform the task. Then, this paper proposes a hybrid optimization algorithm based on bionic learning for path planning. Finally, we set up a reasonable evaluation mechanism and conducted simulation experiments. The algorithm in this paper is compared with genetic algorithm, gray wolf algorithm and whale algorithm, the results show that the total cost of the task obtained by the proposed algorithm is the lowest, the algorithm stability is better, and the convergence accuracy is the highest.

关键词：

Deploying Cationic Cellulose Nanofiber Confinement to Enable High Iodine Loadings Towards High Energy and High-Temperature Zn-I2Battery

学校读者我要写书评

暂无评论

Angewandte Chemie 2023年第5期136卷

作者： Zhenglin Li Wenwen Cao Tao Hu Yichan Hu Rong Zhang Huilin Cui Funian Mo Chaozheng Liu Chunyi Zhi Guojin Liang Institute of Technology for Carbon Neutrality Shenzhen Institute of Advanced Technology Chinese Academy of Sciences (CAS) Shenzhen 518055 China Co-Innovation Center of Efficient Processing and Utilization of Forest Resources College of Materials Science and Engineering Nanjing Forestry University Nanjing 210000 China Contribution: Data curation (lead) Investigation (lead) Methodology (lead) Writing - original draft (lead) School of Chemistry and Chemical Engineering University of Jinan Jinan 250022 China Contribution: Conceptualization (equal) Data curation (lead) Investigation (lead) Methodology (lead) School of Materials Science and Engineering Key Laboratory of Structure and Functional Regulation of Hybrid Materials Ministry of Education Anhui University Hefei 230601 China Contribution: Data curation (equal) Formal analysis (equal) Investigation (supporting) Methodology (supporting) Contribution: Conceptualization (equal) Data curation (supporting) Formal analysis (equal) Investigation (equal) Department of Materials Science and Engineering City University of Hong Kong 83 Tat Chee Avenue Kowloon Hong Kong 999077 Hong Kong Contribution: Data curation (equal) Investigation (equal) Methodology (equal) Contribution: Conceptualization (supporting) Data curation (supporting) Formal analysis (equal) Investigation (equal) Shenzhen Key Laboratory of Flexible Printed Electronics Technology Center Harbin Institute of Technology Shenzhen Guangdong 518055 China Contribution: Conceptualization (lead) Data curation (lead) Resources (lead) Supervision (equal) Contribution: Conceptualization (equal) Data curation (lead) Investigation (lead) Supervision (equal) Contribution: Supervision (lead) Visualization (lead) Writing - review & editing (lead) Contribution: Conceptualization (lead) Supervision (lead) Writing - original draft (lead) Writing - review & editing (lead)

High iodine loading and high-temperature adaptability of the iodine cathode are prerequisites to achieving high energy density at full battery level and promoting the practical application for the zinc-iodine (Zn-I 2 ) battery. However, it would aggravate the polyiodide shuttle effect when employing high iodine loading and working temperature. Here, a sustainable cationic cellulose nanofiber (cCNF) was employed to confine the active iodine species through strong physiochemical adsorption to enlarge the iodine loading and stabilize it even at high temperatures. The cCNF could accommodate dual-functionality by enlarging the iodine loading and suppressing the polyiodide shuttle effect, owing to the unique framework structure with abundant surface positive charges. As a result, the iodine cathode based on the cCNF could deliver high iodine mass loading of 14.1 mg cm −2 with a specific capacity of 182.7 mAh g −1 , high areal capacity of 2.6 mAh cm −2 , and stable cycling over 3000 cycles at 2 A g −1 , thus enabling a high energy density of 34.8 Wh kg −1 and the maximum power density of 521.2 W kg −1 at a full Zn-I 2 battery level. In addition, even at a high temperature of 60 °C, the Zn-I 2 battery could still deliver a stable cycling.

关键词： Cationic Cellulose High- Temperature Battery Iodine Shuttle-Free Zn-I2 Battery

Improved Siamese Network for Video-Based Person Re-identification

学校读者我要写书评

暂无评论

Improved Siamese Network for Video-Based Person Re-identific...

International Conference on Networking and Network Applications (NaNA)

作者： Zhongmin Wang Lanlan Cai Na Duan Lin Fan Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an University of Posts and Telecommunications Xi'an China School of Computer Science and Technology Xi'an University of Posts and Telecommunications Xi'an China

ISBN: (数字)9781728126296

ISBN: (纸本)9781728126302

In the field of human-computer interaction and visual surveillance, person re-identification is undoubtedly a key task. In this paper, we propose improved Siamese network for video-based person re-identification, which combines the verification model and the classification model, in a way that all the supervision information of the samples is fully utilized. The key frame extraction is able to select informative frames over the sequence, which removes occluded frames according to optical flow vector and measures the similarity of two adjacent frames' contents in terms of the block mutual information entropy. A convolutional network and long short-term memory joint training is used as a feature extractor to re-identify using the improved Siamese network framework. Experimental data on the iLIDS-VID and PRID-2011 datasets show that this approach is superior to existing video-based re-identification methods.

关键词： Feature extraction Optical flow Mutual information data mining Entropy Logic gates Cameras

Deep policy hashing network with listwise supervision

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Wang, Shaoying Lai, Haijiang Yang, Yifan Yin, Jian School of Electronics and Information Engineering Sun Yat-Sen University Guangzhou China Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China School of Data and Computer Science Sun Yat-Sen University Guangzhou China

Deep-networks-based hashing has become a leading approach for large-scale image retrieval, which learns a similarity-preserving network to map similar images to nearby hash codes. The pairwise and triplet losses are two widely used similarity preserving manners for deep hashing. These manners ignore the fact that hashing is a prediction task on the list of binary codes. However, learning deep hashing with listwise supervision is challenging in 1) how to obtain the rank list of whole training set when the batch size of the deep network is always small and 2) how to utilize the listwise supervision. In this paper, we present a novel deep policy hashing architecture with two systems are learned in parallel: A query network and a shared and slowly changing database network. The following three steps are repeated until convergence: 1) the database network encodes all training samples into binary codes to obtain whole rank list, 2) the query network is trained based on policy learning to maximize a reward that indicates the performance of the whole ranking list of binary codes, e.g., mean average precision (MAP), and 3) the database network is updated as the query network. Extensive evaluations on several benchmark datasets show that the proposed method brings substantial improvements over state-of-the-art hashing methods. Copyright © 2019, The Authors. All rights reserved.

关键词： Binary codes