检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Zhan, Jingtao Mao, Jiaxin Liu, Yiqun Guo, Jiafeng Zhang, Min Ma, Shaoping Department of Computer Science and Technology Institute for Artificial Intelligence Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China Beijing Key Laboratory of Big Data Management and Analysis Methods Gaoling School of Artificial Intelligence Renmin University of China Beijing100872 China CAS Key Lab of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China

Dense Retrieval (DR) has achieved state-of-the-art first-stage ranking effectiveness. However, the efficiency of most existing DR models is limited by the large memory cost of storing dense vectors and the time-consuming nearest neighbor search (NNS) in vector space. Therefore, we present RepCONC, a novel retrieval model that learns discrete Representations via CONstrained Clustering. RepCONC jointly trains dual-encoders and the Product Quantization (PQ) method to learn discrete document representations and enables fast approximate NNS with compact indexes. It models quantization as a constrained clustering process, which requires the document embeddings to be uniformly clustered around the quantization centroids and supports end-to-end optimization of the quantization method and dual-encoders. We theoretically demonstrate the importance of the uniform clustering constraint in RepCONC and derive an efficient approximate solution for constrained clustering by reducing it to an instance of the optimal transport problem. Besides constrained clustering, RepCONC further adopts a vector-based inverted file system (IVF) to support highly efficient vector search on CPUs. Extensive experiments on two popular ad-hoc retrieval benchmarks show that RepCONC achieves better ranking effectiveness than competitive vector quantization baselines under different compression ratio settings. It also substantially outperforms a wide range of existing retrieval models in terms of retrieval effectiveness, memory efficiency, and time efficiency. Copyright © 2021, The Authors. All rights reserved.

关键词： Signal encoding

来源：评论

学校读者我要写书评

暂无评论

Efficient Visual Perception of Human-Robot Walking Environments Using Semi-Supervised Learning

Efficient Visual Perception of Human-Robot Walking Environme...

引用

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

作者： Dmytro Kuzmenko Oleksii Tsepa Andrew Garrett Kurbis Alex Mihailidis Brokoslaw Laschowski Department of Mathematics National University of Kyiv-Mohyla Academy Kyiv Ukraine Department of Mathematical Methods of System Analysis Igor Sikorsky Kyiv Polytechnic Institute Kyiv Ukraine Department of Computer Science University of Toronto Toronto Canada Department of Systems Design Engineering University of Waterloo Waterloo Canada Temerty Faculty of Medicine University of Toronto Toronto Canada Institute of Biomedical Engineering University of Toronto Toronto Canada Toronto Rehabilitation Institute Toronto Canada Department of Mechanical and Industrial Engineering Robotics Institute University of Toronto Toronto Canada

Convolutional neural networks trained using supervised learning can improve visual perception for human-robot walking. These advances have been possible due to largescale datasets like ExoNet and StairNet - the largest open-source image datasets of real-world walking environments. However, these datasets require vast amounts of manually annotated data, the development of which is time consuming and labor intensive. Here we present a novel semi-supervised learning system (ExoNet-SSL) that uses over 1.2 million unlabelled images from ExoNet to improve training efficiency. We developed a deep learning model based on mobile vision transformers and trained the model using semi-supervised learning for image classification. Compared to standard supervised learning (98.4%), our ExoNet-SSL system was able to maintain high prediction accuracy (98.8%) when tested on previously unseen environments, while requiring 35% fewer labelled images during training. These results show that semi-supervised learning can improve training efficiency by leveraging large amounts of unlabelled data and minimize the size requirements for manually annotated images. Future research $\text{will}$ focus on model deployment for onboard real-time inference and control of human-robot walking.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Interpreting dense retrieval as mixture of topics

arXiv

引用

arXiv 2021年

Dense Retrieval (DR) reaches state-of-the-art results in first-stage retrieval, but little is known about the mechanisms that contribute to its success. Therefore, in this work, we conduct an interpretation study of recently proposed DR models. Specifically, we first discretize the embeddings output by the document and query encoders. Based on the discrete representations, we analyze the attribution of input tokens. Both qualitative and quantitative experiments are carried out on public test collections. Results suggest that DR models pay attention to different aspects of input and extract various high-level topic representations. Therefore, we can regard the representations learned by DR models as a mixture of high-level topics. Copyright © 2021, The Authors. All rights reserved.

关键词： Mixtures

来源：评论

学校读者我要写书评

暂无评论

Not-Quite Transcendental Functions and their Applications

arXiv

引用

arXiv 2022年

作者： Miller, Jonah M. Dolence, Joshua C. Holladay, Daniel Computational Physics and Methods Los Alamos National Laboratory P.O. Box 1663 Los AlamosNM United States Center for Theoretical Astrophysics Los Alamos National Laboratory P.O. Box 1663 Los AlamosNM United States Applied Computer Science Los Alamos National Laboratory P.O. Box 1663 Los AlamosNM United States

Transcendental functions, such as exponentials and logarithms, appear in a broad array of computational domains: from simulations in curvilinear coordinates, to interpolation, to machine learning. Unfortunately they are typically expensive to compute accurately. In this note, we argue that in many cases, the properties of the function matters more than the exact functional form. We present new functions, which are not transcendental, that can be used as drop-in replacements for the exponential and logarithm in many settings for a significant performance boost. We show that for certain applications using these functions result in no drop in the accuracy at all, as they are perfectly accurate representations of themselves, if not the original transcendental functions. Copyright © 2022, The Authors. All rights reserved.

关键词： Interpolation

来源：评论

学校读者我要写书评

暂无评论

GEA-net: Global embedded attention neural network for image classification

GEA-net: Global embedded attention neural network for image ...

引用

IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)

作者： Yajie Wang Zhiyang Wan Zhao Pei Chengcai Leng Yuli Chen School of Information Engineering Shaanxi Xueqian Normal University Xi'an China School of Computer Science Shaanxi Normal University Xi'an China Key Laboratory of Modern Teaching Technology Ministry of Education Xi'an China School of Mathematics Northwest University Xi'an China

ISBN: (纸本)9781665416597

Recently, it is generally acknowledged that the receptive field size of visual cortical neurons are regulated by the stimulus in the neuroscience community. Thus, once the global receptive field is obtained, the network performance can be greatly improved. Unfortunately, the larger receptive field method has been rarely considered in constructing CNNs. Recent studies on network design have demonstrated that the key to improving model performance is channel attention. However, they usually neglect the location information. Hence, it is difficult to capture the long-term dependency of location information. In particular, the location information is important for generating spatially selective attention blocks. Therefore, in this paper, we propose a novel attention neural network termed “GEA” by embedding global context information into channel attention. Firstly, instead of channel attention, which is transformed from feature tensor to single feature vector via 2D pooling, our method decomposes the channel attention into two 1D feature encoding processes that aggregate features along two spatial directions. In particular, the long-term dependency can be captured by using one spatial direction as well as preserving accurate location information in the opposite direction. Then, we concatenate the results of the two directions, while carry out batch normalization and use relu activation function. In addition, the cross channel soft attention is used to adaptively select different spatial scales of the information. Finally, our method is simple and can be flexibly inserted into classic networks, such as ResNet and EfficientNet, with limited computational overhead. A large number of experiments show that our method is not only conducive to the classification of COCO and ImageNet, but also demonstrates promising performance for 3D features, such as Magnetic Resonance Imaging(MRI).

关键词： Visualization Privacy Three-dimensional displays Tensors Neuroscience Fuses Neurons

来源：评论

学校读者我要写书评

暂无评论

Multi-body structure information enhancement GCN for 3D human pose estimation in videos

引用

Applied Soft Computing 2025年 180卷

作者： Zhou, Hanghang Zhang, Yumei Ma, Jinli Yang, Honghong Wu, Xiaojun School of Computer Science Shaanxi Normal University Xi'an China Key Laboratory of Modern Teaching Technology Ministry of Education Shaanxi Normal University Xi'an China Key Laboratory of Intelligent Computing and Service Technology for Folk Song Ministry of Culture and Tourism Xi'an China

To alleviate the problem of depth ambiguity in 3D human pose estimation, we propose Multi-body Structure information enhancement Graph Convolutional Network (MS-GCN), which exploits neighborhood subgraph structure to distinguish similar nodes and reduces the risk of over-smoothing of graph convolutional networks. Specifically, MS-GCN comprises the Structure Weight information enhancement GCN (SWGCN) and the Structural Semantic information excitation GCN (SSGCN), which capture global and local interactions among nodes. In the SWGCN, we propose a data-enhanced method. Here, an adaptive distance perception method is proposed to dig the multi-body structure weight information of neighborhood subgraphs, which contains the nonlinear dynamic characteristics of human motion. Then, the proposed multi-head cross-talk graph attention explores a variety of possible associations between node neighbors based on content and structure representations. In the SSGCN, we propose a Neighborhood Subgraph Encoder (NSE) to aggregate structural semantic information, which provides sufficient feature information for pose estimation. Besides, we adopt a Gated Recurrent Unit Pose Lifting (GRUPL) layer to improve the accuracy of the final result. Finally, we conduct extensive experiments on two popular benchmarks, e.g., Human3.6M and HumanEva-I, where the proposed model outperforms a number of recent state-of-the-art approaches with fewer input frames and model size. © 2025 Elsevier B.V.

关键词： 3D human pose estimation Graph convolutional networks Multi-body structure information Neighborhood subgraph encoder

来源：评论

学校读者我要写书评

暂无评论

Earf-YOLO:A Model for Recognizing Zhuang Minority Patterns Based on YOLOv3

Research Square

引用

Research Square 2021年

作者： Wang, Xin Yan, Jingke Wang, Qin Qin, Qin Wang, Jun Shi, Zhuo Li, Huanghe School of Computer Science and Information Security Guilin University of Electronic Technology Guilin541004 China School of Marine Engineering Guilin University of Electronic Technology Beihai536000 China Basic Teaching Department Guilin University of Electronic Technology Beihai536000 China

With reference to the limitations of YOLOv3 in recognizing symbols on Zhuang pattern, such as slow detection speed, unable to detect small object, and inaccurate positioning of bounding box, we propose a new model: Earf-YOLO (Efficient Attention Receptive Field You only look once) in this paper. In EarF-YOLO, we present an attention module: CBEAM (Convolution Block Efficient Attention Module) at first, which provides feature maps from channel and spatial dimensions. In CBEAM module, a local cross-channel interaction strategy without reducing dimensionality is used to improve the performance of the convolutional neural network. Besides, we put forward the SRFB (Strength Receptive Field Block) structure. During its training, more branch structures will be generated to enrich the feature space of the convolutional block. During its prediction, the multi-branched structures will be reparametrized and fused into one main branch to improve the performance of the model. Finally, we adopt some advanced training techniques to improve the detection performance. Experiments on the dataset of Zhuang patterns and the COCO dataset show that the Earf-YOLO model can effectively reduce the error of the prediction box and the ground-truth box, and decrease the calculation time. The mAP value of this model on the dataset of Zhuang patterns and on the COCO dataset reaches 82.1 (IoU=0.5) and 62.14 (IoU=0.5) respectively. © 2021, CC BY.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Predictive Scale-Bridging Simulations through Active Learning

arXiv

引用

arXiv 2022年

作者： Karra, Satish Mehana, Mohamed Lubbers, Nicholas Chen, Yu Diaw, Abdourahmane Santos, Javier E. Pachalieva, Aleksandra Pavel, Robert S. Haack, Jeffrey R. McKerns, Michael Junghans, Christoph Kang, Qinjun Livescu, Daniel Germann, Timothy C. Viswanathan, Hari S. Computational Earth Science Group Earth and Environmental Sciences Division Los Alamos National Laboratory NM87545 United States Environmental Molecular Sciences Laboratory Pacific Northwest National Laboratory RichlandWA99354 United States Information Sciences Group Computer Computational and Statistical Sciences Division Los Alamos National Laboratory Los AlamosNM87545 United States Department of Mechanics and Aerospace Engineering Southern University of Science and Technology Shenzen518055 China RadiaSoft LLC 6525 Gunpark Dr. Suite 370-411 BoulderCO80301 United States Applied Computer Science Group Computer Computational and Statistical Sciences Division Los Alamos National Laboratory Los AlamosNM87545 United States Computational Physics and Methods Los Alamos National Laboratory Los AlamosNM87545 United States Physics and Chemistry of Materials Group Theoretical Division Los Alamos National Laboratory Los AlamosNM87545 United States

Throughout computational science, there is a growing need to utilize the continual improvements in raw computational horsepower to achieve greater physical fidelity through scale-bridging over brute-force increases in the number of mesh elements. For instance, quantitative predictions of transport in nanoporous media, critical to hydrocarbon extraction from tight shale formations, are impossible without accounting for molecular-level interactions. Similarly, inertial confinement fusion simulations rely on numerical diffusion to simulate molecular effects such as non-local transport and mixing without truly accounting for molecular interactions. With these two disparate applications in mind, we develop a novel capability which uses an active learning approach to optimize the use of local fine-scale simulations for informing coarse-scale hydrodynamics. Our approach addresses three challenges: forecasting continuum coarse-scale trajectory to speculatively execute new fine-scale molecular dynamics calculations, dynamically updating coarse-scale from fine-scale calculations, and quantifying uncertainty in neural network models. © 2022, CC BY.

关键词： Molecular dynamics

来源：评论

学校读者我要写书评

暂无评论

Channel Exchanging Networks for Multimodal and Multitask Dense Image Prediction

arXiv

引用

arXiv 2021年

作者： Wang, Yikai Sun, Fuchun Huang, Wenbing He, Fengxiang Tao, Dacheng State Key Lab on Intelligent Technology and Systems Department of Computer Science and Technology Tsinghua University China Gaoling School of Artificial Intelligence Renmin University of China Beijing Key Laboratory of Big Data Management and Analysis Methods Beijing China JD Explore Academy *** Inc China

Multimodal fusion and multitask learning are two vital topics in machine learning. Despite the fruitful progress, existing methods for both problems are still brittle to the same challenge—it remains dilemmatic to integrate the common information across modalities (resp. tasks) meanwhile preserving the specific patterns of each modality (resp. task). Besides, while they are actually closely related to each other, multimodal fusion and multitask learning are rarely explored within the same methodological framework before. In this paper, we propose Channel-Exchanging-Network (CEN) which is self-adaptive, parameter-free, and more importantly, applicable for multimodal and multitask dense image prediction. At its core, CEN adaptively exchanges channels between subnetworks of different modalities. Specifically, the channel exchanging process is self-guided by individual channel importance that is measured by the magnitude of Batch-Normalization (BN) scaling factor during training. For the application of dense image prediction, the validity of CEN is tested by four different scenarios: multimodal fusion, cycle multimodal fusion, multitask learning, and multimodal multitask learning. Extensive experiments on semantic segmentation via RGB-D data and image translation through multi-domain input verify the effectiveness of CEN compared to state-of-the-art methods. Detailed ablation studies have also been carried out, which demonstrate the advantage of each component we propose. Our code is available at https://***/yikaiw/CEN. Copyright © 2021, The Authors. All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Robust Capacity Expansion Modelling for Renewable Energy Systems under Weather and Demand Uncertainty

arXiv

引用

arXiv 2025年

作者： Kebrich, Sebastian Engelhardt, Felix Franzmann, David Büsing, Christina Linßen, Jochen Heinrichs, Heidi Jülich52425 Germany RWTH Aachen University Teaching and Research Area Combinatorial Optimization Faculty of Mathematics Computer Science and Natural Sciences Aachen52074 Germany University of Siegen Chair for Energy Systems Analysis Department of Mechanical Engineering Siegen57076 Germany

Future greenhouse gas neutral energy systems will be dominated by variable renewable energy technologies. However, renewable electricity generation from wind and solar technologies, as well as electricity demand, varies with the weather. This work addresses the problem of determining optimal capacities for renewable technologies in energy systems that ensure sufficient electricity supply when dealing with multi–year time–series data. For this an iterative algorithm is proposed that starts by optimising an arbitrary starting time–series, followed by adding additional constraints and reoptimising the modified optimisation problem until sufficient energy supply is provided for all time–series, i.e. the solution is robust to weather and demand variations. This is evaluated in a computational study on a German energy system model. The results show that the iterative algorithm finds robust solutions for an increase of 2–2.5% in total annual cost for a simplified model in gurobipy and 2.9% for a model built in the open source model framework ***. Testing the feasibility for all non robust solutions showed that supply gaps occurred in at least some of the remaining years. Based on the results of this work, ensuring feasibility within an energy system model for multiple time–series boils down to two factors: 1.) ensuring sufficient back–up capacity to overcome periods of high demand combined with low electricity generation from wind and photovoltaic, and 2.) enforcing sufficient total annual electricity generation. Our proposed open source iterative algorithm is able to ensure this. For general modelling, it is recommended to (a) check for systematic effects of different years’ time–series on energy system models especially for wind, but also for photovoltaics, (b) include dark lull and cold period effects on generation and demand in time–series, and (c) assess the feasibility of energy system models using different time–*** Codes 90C10, 90C15, 90C90 Copyright

关键词： Critical path analysis

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：