检索结果-内蒙古大学图书馆

Deep Modular Co-Attention networks for visual question answering

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Yu, Zhou Yu, Jun Cui, Yuhao Tao, Dacheng Tian, Qi Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University UBTECH Sydney AI Centre School of Computer Science FEIT University of Sydney Noah's Ark Lab Huawei

Visual Question Answering (VQA) requires a finegrained and simultaneous understanding of both the visual content of images and the textual content of questions. Therefore, designing an effective 'co-attention' model to associate key words in questions with key objects in images is central to VQA performance. So far, most successful attempts at co-attention learning have been achieved by using shallow models, and deep co-attention models show little improvement over their shallow counterparts. In this paper, we propose a deep Modular Co-Attention Network (MCAN) that consists of Modular Co-Attention (MCA) layers cascaded in depth. Each MCA layer models the self-attention of questions and images, as well as the guided-attention of images jointly using a modular composition of two basic attention units. We quantitatively and qualitatively evaluate MCAN on the benchmark VQA-v2 dataset and conduct extensive ablation studies to explore the reasons behind MCAN's effectiveness. Experimental results demonstrate that MCAN significantly outperforms the previous state-ofthe- A rt. Our best single model delivers 70.63% overall accuracy on the test-dev set. Code is available at http://***/MILVLG/mcan-vqa. Copyright © 2019, The Authors. All rights reserved.

关键词： Machine learning

MobileEdge: Enhancing On-Board Vehicle Computing Units Using Mobile Edges for CAVs

学校读者我要写书评

暂无评论

MobileEdge: Enhancing On-Board Vehicle Computing Units Using...

International Conference on Parallel and Distributed systems (ICPADS)

作者： Lin Wang Qingyang Zhang Youhuizi Li Hong Zhong Weisong Shi School of Computer Science and Technology Anhui University Hefei China Key Laboratory of Complex Systems Modeling and Simulation Hangzhou Dianzi University Hangzhou China Department of Computer Science Wayne State University Detroit MI USA

ISBN: (纸本)9781728125848

As the rapid growth of connected and autonomous vehicles (CAVs) and 5G intensifies, more third-party applications are increasingly being deployed on CAVs. They not only improve user experience but also provide more helpful services, for example, enhancing public safety by recognizing criminals in real-time videos. Current CAVs prefer to process collected data on the vehicle to avoid long transmission latency and extra network cost. However, due to the limitations of the on-board vehicle computing unit (VCU) and increasing use of computing-intensive in-vehicle applications, the burden of on-board VCU has sharply increased, which may affect driving safety. In particular, for existing vehicles on the road, adding more computing devices is a challenge if not impossible due to cost concerns. Inspired by edge computing, we propose a novel platform, MobileEdge, to enhance the computing capability of the unchangeable on-board VCU, which leverages mobile devices as edge nodes, e.g., the passengers' smartphones, by offloading computing tasks to them for collaboratively computing. Moreover, MobileEdge provides the dynamic management of mobile devices, monitoring device status and interfaces for customizable task offloading strategies and eventually achieves optimal task scheduling. We build a prototype to demonstrate the designed platform and evaluate three task offloading strategies which were implemented based on the developed interfaces. The results show that MobileEdge significantly reduces the application response latency. Compared with the baseline which does not employ task offloading, the response latency is almost near real-time when more computing resources are available. In addition, the proposed shortest response latency strategy outperforms the best overall task scheduling among the three strategies.

关键词： Edge computing Vehicular data analysis Distributed computing Connected and autonomous vehicles Distributed computing TASK SCHEDULING Response Latency UNLOADING vehicle vinyl chloride-urethane copolymer Mobile devices unchangeable quasi-realtime language Public safety

Ellipticity control of terahertz high-harmonic generation in a Dirac semimetal

学校读者我要写书评

暂无评论

Physical Review B 2022年第8期106卷 L081127-L081127页

作者： Semyon Germanskiy Renato M. A. Dantas Sergey Kovalev Chris Reinhoffer Evgeny A. Mashkovich Paul H. M. van Loosdrecht Yunkun Yang Faxian Xiu Piotr Surówka Roderich Moessner Takashi Oka Zhe Wang Institute of Physics II University of Cologne 50937 Cologne Germany Max Planck Institute for the Physics of Complex Systems 01187 Dresden Germany Department of Physics University of Basel 4056 Basel Switzerland Institute of Radiation Physics Helmholtz-Zentrum Dresden-Rossendorf 01328 Dresden Germany State Key Laboratory of Surface Physics and Department of Physics Fudan University Shanghai 200232 China Shanghai Qi Zhi Institute No. 701 Yunjin Road Xuhui District Shanghai 200232 China Department of Theoretical Physics Wrocław University of Science and Technology 50-370 Wrocław Poland Institute for Theoretical Physics University of Amsterdam 1090 GL Amsterdam The Netherlands Dutch Institute for Emergent Phenomena (DIEP) University of Amsterdam 1090 GL Amsterdam The Netherlands The Institute for Solid State Physics The University of Tokyo Kashiwa Chiba 277-8581 Japan Max Planck Institute for Chemical Physics of Solids 01187 Dresden Germany Department of Physics TU Dortmund University 44227 Dortmund Germany

We report on terahertz high-harmonic generation in a Dirac semimetal as a function of the driving-pulse ellipticity and on a theoretical study of the field-driven intraband kinetics of massless Dirac fermions. Very efficient control of the third-harmonic yield and polarization state is achieved in electron-doped Cd3As2 thin films at room temperature. The observed tunability is understood as resulting from terahertz-field-driven intraband kinetics of the Dirac fermions. Our study paves the way for exploiting the nonlinear optical properties of Dirac matter for applications in signal processing and optical communications.

关键词： Dirac semimetal Boltzmann theory High-harmonic generation Terahertz techniques Terahertz time-domain spectroscopy

Physics case for quarkonium studies at the Electron Ion Collider

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Boer, Daniël Flett, Chris A. Flore, Carlo Kikola, Daniel Lansberg, Jean-Philippe Nefedov, Maxim Van Hulse, Charlotte Bhattacharya, Shohini Bor, Jelle Butenschoen, Mathias Ceccopieri, Federico Chen, Longjie Cheung, Vincent D’Alesio, Umberto Echevarria, Miguel Hatta, Yoshitaka Hyde, Charles E. Kishore, Raj Kosarzewski, Leszek Lorcé, Cédric Li, Wenliang Li, Xuan Maxia, Luca Metz, Andreas Mukherjee, Asmita Camacho, Carlos Muñoz Murgia, Francesco Nadel-Turonski, Pawel Pisano, Cristian Qiu, Jian-Wei Rajesh, Sangem Rinaldi, Matteo West, Jennifer Rittenhouse Saleev, Vladimir Santiesteban, Nathaly Setyadi, Chalis Taels, Pieter Tu, Zhoudunmin Vitev, Ivan Vogt, Ramona Watanabe, Kazuhiro Yao, Xiaojun Yedelkina, Yelyzaveta Yoshida, Shinsuke Van Swinderen Institute for Particle Physics and Gravity University of Groningen Groningen9747 AG Netherlands Université Paris-Saclay CNRS IJCLab Orsay91405 France Dipartimento di Fisica Università di Torino INFN Sezione di Torino Via P. Giuria 1 TorinoI-10125 Italy Dipartimento di Fisica Università di Cagliari INFN Sezione di Cagliari Cittadella Univ. CA MonserratoI-09042 Italy Faculty of Physics Warsaw University of Technology plac Politechniki 1 Warszawa00-661 Poland University of Alcalá Alcalá de Henares Madrid Spain Physics Department Brookhaven National Laboratory Bldg. 510A UptonNY11973 United States II. Institut für Theoretische Physik Universität Hamburg Luruper Chaussee 149 Hamburg22761 Germany Université de Liège B4000 Liège Belgium Guangdong Basic Research Center of Excellence for Structure and Fundamental Interactions of Matter Institute of Quantum Matter South China Normal University Guangzhou510006 China Guangdong-Hong Kong Joint Laboratory of Quantum Matter Guangdong Provincial Key Laboratory of Nuclear Science Southern Nuclear Science Computing Center South China Normal University Guangzhou510006 China Nuclear and Chemical Sciences Division Lawrence Livermore National Laboratory LivermoreCA94551 United States Department of Physics EHU Quantum Center University of the Basque Country UPV/EHU Apartado 644 Bilbao48080 Spain RIKEN BNL Research Center Brookhaven National Laboratory UptonNY11973 United States Department of Physics Old Dominion University NorfolkVA23529 United States Center for Frontiers in Nuclear Science Stony Brook University Stony BrookNY11794 United States Department of Physics The Ohio State University ColumbusOH43210 United States CPHT CNRS Ecole polytechnique Institut Polytechnique de Paris Palaiseau91120 France Stony Brook University Stony BrookNY11794 United States Los Alamos National Laboratory Los AlamosNM87545 United States Department of Physics Temple University Philadelp

关键词： Elementary particle interactions

Multimodal transformer with multi-view visual representation for image captioning

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Yu, Jun Li, Jing Yu, Zhou Huang, Qingming Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University Hangzhou310018 China School of Computer and Control Engineering University of Chinese Academy of Sciences Beijing101408 China

Image captioning aims to automatically generate a natural language description of a given image, and most state-ofthe-art models have adopted an encoder-decoder framework. The framework consists of a convolution neural network (CNN)-based image encoder that extracts region-based visual features from the input image, and an recurrent neural network (RNN)-based caption decoder that generates the output caption words based on the visual features with the attention mechanism. Despite the success of existing studies, current methods only model the co-attention that characterizes the inter-modal interactions while neglecting the self-attention that characterizes the intramodal interactions. Inspired by the success of the Transformer model in machine translation, here we extend it to a Multimodal Transformer (MT) model for image captioning. Compared to existing image captioning approaches, the MT model simultaneously captures intra- and inter-modal interactions in a unified attention block. Due to the in-depth modular composition of such attention blocks, the MT model can perform complex multimodal reasoning and output accurate captions. Moreover, to further improve the image captioning performance, multi-view visual features are seamlessly introduced into the MT model. We quantitatively and qualitatively evaluate our approach using the benchmark MSCOCO image captioning dataset and conduct extensive ablation studies to investigate the reasons behind its effectiveness. The experimental results show that our method significantly outperforms the previous state-of-the-art methods. With an ensemble of seven models, our solution ranks the 1st place on the real-time leaderboard of the MSCOCO image captioning challenge at the time of the writing of this paper. Copyright © 2019, The Authors. All rights reserved.

关键词： Recurrent neural networks

ActivityNet-QA: A dataset for understanding complex web videos via question answering

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Yu, Zhou Xu, Dejing Yu, Jun Yu, Ting Zhao, Zhou Zhuang, Yueting Tao, Dacheng Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University Hangzhou China College of Computer Science Zhejiang University Hangzhou China UBTECH Sydney AI Centre SIT FEIT University of Sydney Australia

Recent developments in modeling language and vision have been successfully applied to image question answering. It is both crucial and natural to extend this research direction to the video domain for video question answering (VideoQA). Compared to the image domain where large scale and fully annotated benchmark datasets exists, VideoQA datasets are limited to small scale and are automatically generated, etc. These limitations restrict their applicability in practice. Here we introduce ActivityNet-QA, a fully annotated and large scale VideoQA dataset. The dataset consists of 58,000 QA pairs on 5,800 complex web videos derived from the popular ActivityNet dataset. We present a statistical analysis of our ActivityNet-QA dataset and conduct extensive experiments on it by comparing existing VideoQA baselines. Moreover, we explore various video representation strategies to improve VideoQA performance, especially for long videos. The dataset is available at https://***/MILVLG/ activitynet-qa. Copyright © 2019, The Authors. All rights reserved.

关键词： Modeling languages

The climate opportunities and risks of improving building envelopes across 1,677 Chinese cities

学校读者我要写书评

暂无评论

Cell Reports Sustainability 2024年第12期1卷

作者： Zhang, Yufei Dang, Mengyuan Chu, Chunli Behrens, Paul Berrill, Peter Zhong, Xiaoyang Jing, Rui Lei, Nuoa Jia, Hongyuan Zhang, Lixiao Shao, Chaofeng Masanet, Eric Ju, Meiting Liu, Lirong Chen, Weiqiang Cao, Zhi College of Environmental Science and Engineering Nankai University 38 Tongyan Road Jinnan District Tianjin 300350 China Institute of Environmental Sciences (CML) Leiden University RA Leiden Leiden 2300 Netherlands Oxford Martin School University of Oxford 34 Broad Street Oxford OX1 3BD United Kingdom Sustainability Economics of Human Settlements Technische Universität Berlin Fasanenstr. 88 Berlin 10623 Germany Mercator Research Institute on Global Commons and Climate Change Berlin 10829 Germany International Institute of Applied Systems Analysis (IIASA) Schlossplatz 1 Laxenburg 2361 Austria College of Energy Xiamen University Xiamen 361005 China Energy Technologies Area Lawrence Berkeley National Laboratory Berkeley 94720 CA United States School of Civil Engineering and Architecture Chongqing University of Science and Technology Chongqing 401331 China SinBerBEST Berkeley Education Alliance for Research in Singapore Singapore 138602 Singapore State Key Joint Laboratory of Environmental Simulation and Pollution Control School of Environment Beijing Normal University Beijing 100875 China Bren School of Environmental Science and Management University of California Santa Barbara 93106 CA United States Department of Mechanical Engineering University of California Santa Barbara 93106 CA United States Centre for Environment and Sustainability University of Surrey Guildford GU2 7XH United Kingdom Institute of Urban Environment Chinese Academy of Sciences Xiamen 361021 China

The global building sector consumes approximately 30% of final energy, making it crucial for climate change mitigation and adaptation. International calls for enhancing building energy efficiencies are growing, focusing on strategies such as energy-efficient building envelopes through renovation and replacement of older structures, along with electrification and fuel switching. However, the energy-saving potential of these improvements remains uncertain due to the complex interplay of building stock characteristics and climatic conditions. Here, we diagnose the compound effects of envelope improvements and climate change on China's housing energy demand using a physics-based building energy model with fine spatial and temporal granularity, covering 1,677 sub-province-level cities. Our model shows that envelope improvements play very different roles in ameliorating climate change impacts on housing energy use across the country, highlighting the need for building climate-resilient energy supply and pursuing alternative energy efficiency strategies in less climate-resilient regions. © 2024 The Authors

关键词： building energy building envelopes climate change cooling demand heating demand stock dynamics surrogate modeling

Multimodal unified attention networks for vision-and-language interactions

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Yu, Zhou Cui, Yuhao Yu, Jun Tao, Dacheng Tian, Qi Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University China UBTECH Sydney AI Centre School of Computer Science Faculty of Engineering University of Sydney Australia Noah's Ark Lab Huawei China

Learning an effective attention mechanism for multimodal data is important in many vision-and-language tasks that require a synergic understanding of both the visual and textual contents. Existing state-of-the-art approaches use co-attention models to associate each visual object (e.g., image region) with each textual object (e.g., query word). Despite the success of these co-attention models, they only model inter-modal interactions while neglecting intra-modal interactions. Here we propose a general 'unified attention' model that simultaneously captures the intra- and inter-modal interactions of multimodal features and outputs their corresponding attended representations. By stacking such unified attention blocks in depth, we obtain the deep Multimodal Unified Attention Network (MUAN), which can seamlessly be applied to the visual question answering (VQA) and visual grounding tasks. We evaluate our MUAN models on two VQA datasets and three visual grounding datasets, and the results show that MUAN achieves top level performance on both tasks without bells and whistles. Copyright © 2019, The Authors. All rights reserved.

关键词： Deep learning

Directed momentum current induced by the PT-symmetric driving

学校读者我要写书评

暂无评论

Physical Review E 2019年第4期99卷 042201-042201页

作者： Wen-Lei Zhao Jiaozi Wang Xiaohui Wang Peiqing Tong School of Science Jiangxi University of Science and Technology Ganzhou 341000 China Department of Modern Physics University of Science and Technology of China Hefei 230026 China School of Information Engineering Jiangxi University of Science and Technology Ganzhou 341000 China Department of Physics and Institute of Theoretical Physics Nanjing Normal University Nanjing 210023 China Jiangsu Provincial Key Laboratory for Numerical Simulation of Large Scale Complex Systems Nanjing Normal University Nanjing Jiangsu 210023 China

We investigate the directed momentum current in the quantum kicked rotor model with PT-symmetric deriving potential. For the quantum nonresonance case, the values of quasienergy become complex when the strength of the imaginary part of the kicking potential exceeds a threshold value, which demonstrates the appearance of the spontaneous PT symmetry breaking. In the vicinity of the transition point, the momentum current exhibits a staircase growth with time. Each platform of the momentum current corresponds to the mean momentum of some eigenstates of the Floquet operator whose imaginary parts of the quasienergy are significantly large. Above the transition point, the momentum current increases linearly with time. Interestingly, its acceleration rate exhibits a kind of “quantized” increment with the kicking strength. We propose a modified classical acceleration mode of the kicked rotor model to explain such an intriguing phenomenon. Our theoretical prediction is in good agreement with numerical results.

关键词： Coherent control Cold atoms & matter waves Light-matter interaction Wave chaos