检索结果-内蒙古大学图书馆

Multi-modal factorized bilinear pooling with co-attention learning for visual question answering

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Yu, Zhou Yu, Jun Fan, Jianping Tao, Dacheng Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University China Department of Computer Science University of North Carolina Charlotte United States UBTECH Sydney AI Centre School of IT FEIT University of Sydney Australia

Visual question answering (VQA) is challenging because it requires a simultaneous understanding of both the visual content of images and the textual content of questions. The approaches used to represent the images and questions in a fine-grained manner and questions and to fuse these multimodal features play key roles in performance. Bilinear pooling based models have been shown to outperform traditional linear models for VQA, but their high-dimensional representations and high computational complexity may seriously limit their applicability in practice. For multimodal feature fusion, here we develop a Multi-modal Factorized Bilinear (MFB) pooling approach to efficiently and effectively combine multi-modal features, which results in superior performance for VQA compared with other bilinear pooling approaches. For fine-grained image and question representation, we develop a 'co-attention' mechanism using an end-to-end deep network architecture to jointly learn both the image and question attentions. Combining the proposed MFB approach with co-attention learning in a new network architecture provides a unified model for VQA. Our experimental results demonstrate that the single MFB with co-attention model achieves new state-of-theart performance on the real-world VQA dataset. Code available at https://***/yuzcccc/mfb. Copyright © 2017, The Authors. All rights reserved.

关键词： Network architecture

Energy efficiency analysis of heterogeneous platforms: Early experiences 7

学校读者我要写书评

暂无评论

Energy efficiency analysis of heterogeneous platforms: Early...

7th International Green and Sustainable Computing Conference, IGSC 2016

作者： Li, Youhuizi Shi, Weisong Jiang, Congfeng Zhang, Jilin Wan, Jian Key Laboratory of Complex Systems Modeling and Simulation Hangzhou Dianzi University China Mobile and Internet Systems Laboratory Wayne State University United States School of Information and Electronic Engineering Zhejiang University of Science and Technology China

ISBN: (纸本)9781509051175

Heterogeneous multi-core platforms, e.g., ARM's ***, are a promising trend to improve the performance and energy efficiency of future mobile systems. However, the immediate benefits and the challenges to take advantage of the heterogeneity are still not clear. In this paper, we present our early experiences about the energy efficiency of the two *** heterogeneous platforms: ODROID XU+E and ODROID XU3. We quantified compared them with homogeneous platforms through multiple benchmarks which include popular mobile applications and high-performance parallel benchmarks. Besides, we analyzed the scheduling impact on the energy consumption of the heterogeneous platforms and the migration cost is also discussed. Based on the results, several insights, such as fine-granularity power control and thread level parallelism, related to hardware, application and system design are derived. © 2016 IEEE.

关键词： Energy utilization

TaxiCast: Efficient Broadcasting of Multimedia Advertisements in Vehicular Ad-Hoc Networks

学校读者我要写书评

暂无评论

TaxiCast: Efficient Broadcasting of Multimedia Advertisement...

International Conference on Parallel and Distributed systems (ICPADS)

作者： Peng Liu Jia Xu Biao Xu Tingting Fu Key Laboratory of Complex Systems Modeling and Simulation Hangzhou Dianzi University Hangzhou

ISBN: (纸本)9781509053827

Traditional vision based vehicular advertising methods can only support planar advertisements. If there are obstacles blocking the line-of-sight between the advertiser and customers, the efficiency drops quickly. With the proliferation of vehicular ad-hoc networks (VANETs), advertisements can be disseminated through wireless means. However, the utility of advertisements still decay over time so that the advertiser will require fast delivery to achieve higher rewards. In this paper, we consider a taxi based multimedia advertisements broadcasting scenario where the taxis act as the advertising sources. To solve the contradiction between limited communication capacity and big data size, we propose TaxiCast which can achieve very good performance. It first applies signal strength based coding and decoding to obtain the demandings of the surrounding vehicles. Then it solves the advertisements selection process as a knapsack problem. We also consider the reward decay of advertisements and conflicts between the taxis. We conduct simulations in both fixed reward case and decayed reward case. The result shows that our scheme achieves better performance than the existing strategies.

关键词： Advertising Vehicles Public transportation Multimedia communication Ad hoc networks Digital multimedia broadcasting

The Schrödinger equation for general non-hermitian quantum system

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Xiong, Ye Tong, Peiqing Department of Physics and Institute of Theoretical Physics Nanjing Normal University Nanjing210023 China National Laboratory of Solid State Microstructures Nanjing University Nanjing210093 China Jiangsu Key Laboratory for Numerical Simulation of Large Scale Complex Systems Nanjing Normal University Nanjing210023 China

We derive a new time-dependent Schrödinger equation(TDSE) for quantum models with nonhermitian Hamiltonian. Within our theory, the TDSE is symmetric in the two Hilbert spaces spanned by the left and the right eigenstates, respectively. The physical quantities are also identical in these two spaces. Based on this TDSE, we show that exchanging two quasi-particles in a non-hermitian model can generate arbitrary geometric phase. The system can also violate the Lieb-Robinson bound in non-relativistic quantum mechanics so that an action in one place will immediately cause a change in the distance. We show that the above two surprising behaviors can also appear in anyonic model, which makes us propose that the non-hermitian single particle model may possess many common features with anyonic model. Copyright © 2017, The Authors. All rights reserved.

关键词： Quantum optics

Beyond bilinear: Generalized multimodal factorized high-order pooling for visual question answering

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Yu, Zhou Yu, Jun Xiang, Chenchao Fan, Jianping Tao, Dacheng Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University China Department of Computer Science University of North Carolina Charlotte United States UBTECH Sydney Artificial Intelligence Centre School of Information Technologies Faculty of Engineering and Information Technologies University of Sydney DarlingtonNSW2008 Australia

Visual question answering (VQA) is challenging because it requires a simultaneous understanding of both visual content of images and textual content of questions. To support the VQA task, we need to find good solutions for the following three issues: 1) fine-grained feature representations for both the image and the question;2) multi-modal feature fusion that is able to capture the complex interactions between multi-modal features;3) automatic answer prediction that is able to consider the complex correlations between multiple diverse answers for the same question. For fine-grained image and question representations, a 'co-attention' mechanism is developed by using a deep neural network architecture to jointly learn the attentions for both the image and the question, which can allow us to reduce the irrelevant features effectively and obtain more discriminative features for image and question representations. For multi-modal feature fusion, a generalized Multi-modal Factorized High-order pooling approach (MFH) is developed to achieve more effective fusion of multi-modal features by exploiting their correlations sufficiently, which can further result in superior VQA performance as compared with the state-of-the-art approaches. For answer prediction, the KL (Kullback-Leibler) divergence is used as the loss function to achieve precise characterization of the complex correlations between multiple diverse answers with same or similar meaning, which can allow us to achieve faster convergence rate and obtain slightly better accuracy on answer prediction. A deep neural network architecture is designed to integrate all these aforementioned modules into a unified model for achieving superior VQA performance. With an ensemble of our MFH models, we achieve the state-of-the-art performance on the large-scale VQA datasets and win the runner-up in VQA Challenge 2017. Copyright © 2017, The Authors. All rights reserved.

关键词： Deep neural networks

Explore K-Anycast Information Dissemination in Mobile Opportunistic Networks

学校读者我要写书评

暂无评论

Explore K-Anycast Information Dissemination in Mobile Opport...

IEEE Wireless Communications and Networking Conference

作者： Peng Liu Jia Xu Biao Xu Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University

ISBN: (纸本)9781467398152

In Mobile Opportunistic Networks, there are special cases of multicast in which destinations (except the total number) are not predetermined. E.g., a person tries to find another three players to play poker without knowing them. Achieving efficient data routing in this scenario (target any k destinations among m of them) can be very challenge due to (1) no predetermined destinations, (2) extra delay cost by the destinations collision, and (3) difficulty of the receiver quantity control. We define this as a K-Anycast problem, the goal is to route copies of the message to any k destinations with the shortest average delay. In this paper, we first propose a matching method to solve the problem in a centralized way. We then bring forward a layered hierarchical structure where nodes are organized according to their degree of activities. Based on the structure, two routing algorithms are proposed where K-Cast initializes exactly k copies of the message without replication in the middle, K-Epidemic performs epidemic routing only in a controlled range. Both algorithms will first forward copies upwards along the structure and then downwards to the destinations. Experiments on real data trace show that the proposed algorithms achieve much better delay, delivery ratio and lower forwarding numbers.

关键词： Mobile Opportunistic Network Anycast Data routing Destination collision

Grand-canonical Monte-Carlo simulation of solutions of salt mixtures: theory and implementation

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Nguyen, Toan T. VNU Key Laboratory"Multiscale Simulation of Complex Systems" Faculty of Physics VNU University of Science Vietnam National University 334 Nguyen Trai Street Thanh Xuan Hanoi School of Physics Georgia Institute of Technology 837 State Street AtlantaGA30332-0430 United States

A Grand-canonical Monte-Carlo simulation method extended to simulate a mixture of salts is presented. Due to charge neutrality requirement of electrolyte solutions, ions must be added to or removed from the system in groups. This leads to some complications compared to regular Grand Canonical simulation. Here, a recipe for simulation of electrolyte solution of salt mixture is presented. It is then implemented to simulate solution of 1:1, 2:1 and 2:2 salts or their mixtures at different concentrations using the primitive ion model. The osmotic pressures of the electrolyte solutions are calculated and shown to depend linearly on the salt concentrations within the concentration range simulated. We also show that at the same concentration of divalent anions, the presence of divalent cations make it easier to insert monovalent cations into the system. This can explain some quantitative differences observed in experiments of the MgCl2 salt mixture and MgSO4 salt mixture. Copyright © 2017, The Authors. All rights reserved.

关键词： Magnesium compounds

Touching PBSData — An exploratory study of visual analytics for public bicycle system supporting multi-touch device

学校读者我要写书评

暂无评论

Touching PBSData — An exploratory study of visual analytics...

IEEE International Conference on Progress in Informatics and Computing (PIC)

作者： Xiaoying Shi Zhenhai Yu Xinyu Qu Key Laboratory of Complex Systems Modeling and Simulation Hangzhou Dianzi University Hangzhou China Zhejiang Institute of Communications

ISBN: (纸本)9781509034857

Public bicycle sharing services are becoming popular over the past decade. The behavior of using public bicycle system (PBS) produces numerous hire records involving citizens' movement. This paper presents a visual analytics system-Touching PBSData, to help both traffic administrators and ordinary citizens visually explore and understand PBS data on multi-touch devices. Tasks and design principles are put forward firstly. Four views are designed to show different aspects of the dataset, including the geographical distribution, the spatio-temporal pattern and personal relevant information. In order to enhance the flexibility and usability of system, the organization mode of views and rich user interaction suitable for multi-touch device are further studied. By adopting real PBS dataset, the results of case studies certify that our method is able to help analyzers to find stations' main function and citizens' travel habits. Moreover, the feedback from the participants with various backgrounds is very affirmative through the user study.

关键词： Data visualization Visualization Bicycles Urban areas Organizations Switches Bars

Clustering the Stations of Bicycle Sharing System

学校读者我要写书评

暂无评论

Journal of Donghua University(English Edition) 2016年第6期33卷 968-972页

作者： SHI Xiao-ying YU Zhen-hai XU Hai-tao HUANG Bin-bin Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University Hang zhou 310018 China

Bicycle sharing system has emerged as a new mode of transportation in many big cities over the past *** the large number of bicycle stations distribute widely in the city,it is difficult to identify their unique attributes and characteristics *** to the real bicycle hire dataset in Hangzhou,China,the clustering analysis for the bicycle stations based on the temporal flow data was carried out ***,based on the spatial distribution and temporal attributes of calculated clusters,visual diagram and map were used to vividly analyze the bicycle hire behavior related to station category and study the travel rules of *** experimental results demonstrate the relation between human mobility,the time of day,day of week and the station location.

关键词： urban mining human mobility analysis visual analytics bicycle sharing system