检索结果-内蒙古大学图书馆

IEEE International Geoscience and Remote Sensing Symposium

作者： Siyuan Peng Fang Huang Jian Tao Bo Tie Jun Lu Xiaodong Zhang School of Resources & Environment University of Electronic Science and Technology of China Chengdu P.R. China Institute of Remote Sensing Big Data University of Electronic Science and Technology of China Chengdu P.R. China Texas A&M Engineering Experiment Station and High Performance Research Computing Texas A&M University College Station TX USA

Traditional 3D map reconstruction methods based on unmanned aerial (UA) always relies on multi-camera optical equipment or additional space positioning equipment. All of these restrict the UA's application scenarios to some extent. Visual simultaneous localization and mapping (VSLAM), using the camera as the only external sensor, can construct a 3D map of its spatial environment while recording its own localization. This paper proposes a technique of fast 3D map reconstruction based on UAs by using VSLAM, with parallel computing based on CUDA, to achieve fast dense reconstruction based on a UA platform. This research belongs to one of the emerging but very exciting application areas using UAs. It helps to expand the application scope of UAs to some extent. From the experiments, it is demonstrated that the methodology presented in this paper works well, and could be applied to real world applications.

关键词： Unmanned aerial Visual SLAM 3D map high performance computing Dense reconstruction

来源：评论

学校读者我要写书评

暂无评论

Scientific data Lake for high Luminosity LHC project and other data-intensive particle and astro-particle physics experiments

引用

Journal of Physics: Conference Series 2020年第1期1690卷

作者： A Alekseev A Kiryanov A Klimentov T Korchuganova V Mitsyn D Oleynik A Smirnov S Smirnov A Zarochentsev Compiler Technology Department Ivannikov Institute for System Programming of the RAS Moscow 109004 Russia Cloud technologies and Big Data analytics laboratory Plekhanov Russian University of Economics Moscow 117997 Russia ATLAS group Universidad Andres Bello Santiago 7550196 Chile Department of computing resources and technologies NRC 'Kurchatov Institute' - PNPI 188300 Gatchina Russia Physics Department Brookhaven National Laboratory Upton NY 11973 USA Laboratory of Information Technologies Joint Institute for Nuclear Research Dubna 141980 Russia Institute of Nuclear Physics and Engineering National Research Nuclear University MEPhI Moscow 115409 Russia Laboratory of Ultra-High Energy Physics Saint Petersburg State University Saint Petersburg 198504 Russia

The next phase of LHC Operations – high Luminosity LHC (HL-LHC), which is aimed at ten-fold increase in the luminosity of proton-proton collisions at the energy of 14 TeV, is expected to start operation in 2027-2028 and will deliver an unprecedented scientific data volume of multi-exabyte scale. This amount of data has to be stored and the corresponding storage system should ensure fast and reliable data delivery for processing by scientific groups distributed all over the world. The present LHC computing and data processing model will not be able to provide the required infrastructure growth even taking into account the expected hardware technology evolution. To address this challenge the new state-of-the-art computing infrastructure technologies are now being developed and are presented here. The possibilities of application of the HL-LHC distributed data handling technique for other particle and astro-particle physics experiments dealing with large-scale data volumes like DUNE, LSST, Belle-II, JUNO, SKAO etc. are also discussed.

关键词：

来源：评论

学校读者我要写书评

暂无评论

HPDedup: A Hybrid prioritized data deduplication mechanism for primary storage in the cloud 33

HPDedup: A Hybrid prioritized data deduplication mechanism f...

引用

33rd International Conference on Massive Storage Systems and Technology, MSST 2017

作者： Wu, Huijun Wang, Chen Fu, Yinjin Sakr, Sherif Zhu, Liming Lux, Kai Data CSIRO University of New South Wales Australia PLA University of Science and Technology China Science and Technology on Parallel and Distributed Laboratory State Key Laboratory of High Performance Computing State Key Lab. of High-end Server and Storage Technology Coll. of Computer Natl. Univ. of Def. Technol. China

Eliminating duplicate data in primary storage of clouds increases the cost-efficiency of cloud service providers as well as reduces the cost of users for using cloud services. Most existing primary deduplication techniques either use inline caching to exploit locality in primary workloads or use postprocessing deduplication running in system idle time to avoid the negative impact on I/O performance. However, neither of them works well in the cloud servers running multiple services or applications for the following two reasons: Firstly, the temporal locality of duplicate data writes may not exist in some primary storage workloads thus inline caching often fails to achieve good deduplication ratio. Secondly, the post-processing deduplication allows duplicate data to be written to disks, therefore does not provide the benefit of I/O deduplication and requires high peak storage capacity. This paper presents HPDedup, a Hybrid Prioritized data Deduplication mechanism to deal with the storage system shared by applications running in co-located virtual machines or containers by fusing an inline and a post-processing process for exact deduplication. In the inline deduplication phase, HPDedup gives a fingerprint caching mechanism that estimates the temporal locality of duplicates in data streams from different VMs or applications and prioritizes the cache allocation for these streams based on the estimation. HPDedup also allows different deduplication threshold for streams based on their spatial locality to reduce the disk fragmentation. The post-processing phase removes duplicates whose fingerprints are not able to be cached due to weak temporal locality from disks. The hybrid deduplication mechanism significantly reduces the amount of redundant data written to the storage system while maintaining inline data writing performance. Our experimental results show that HPDedup clearly outperforms the state-of-the-art primary storage deduplication techniques in terms of inline cac

关键词： Efficiency

来源：评论

学校读者我要写书评

暂无评论

SPT clusters with DES and HST weak lensing. I. Cluster lensing and Bayesian population modeling of multiwavelength cluster datasets

引用

Physical Review D 2024年第8期110卷 083509页

作者： S. Bocquet S. Grandis L. E. Bleem M. Klein J. J. Mohr M. Aguena A. Alarcon S. Allam S. W. Allen O. Alves A. Amon B. Ansarinejad D. Bacon M. Bayliss K. Bechtol M. R. Becker B. A. Benson G. M. Bernstein M. Brodwin D. Brooks A. Campos R. E. A. Canning J. E. Carlstrom A. Carnero Rosell M. Carrasco Kind J. Carretero R. Cawthon C. Chang R. Chen A. Choi J. Cordero M. Costanzi L. N. da Costa M. E. S. Pereira C. Davis J. DeRose S. Desai T. de Haan J. De Vicente H. T. Diehl S. Dodelson P. Doel C. Doux A. Drlica-Wagner K. Eckert J. Elvin-Poole S. Everett I. Ferrero A. Ferté A. M. Flores J. Frieman J. García-Bellido M. Gatti G. Giannini M. D. Gladders D. Gruen R. A. Gruendl I. Harrison W. G. Hartley K. Herner S. R. Hinton D. L. Hollowood W. L. Holzapfel K. Honscheid N. Huang E. M. Huff D. J. James M. Jarvis G. Khullar K. Kim R. Kraft K. Kuehn N. Kuropatkin F. Kéruzoré S. Lee P.-F. Leget N. MacCrann G. Mahler A. Mantz J. L. Marshall J. McCullough M. McDonald J. Mena-Fernández R. Miquel J. Myles A. Navarro-Alsina R. L. C. Ogando A. Palmese S. Pandey A. Pieres A. A. Plazas Malagón J. Prat M. Raveri C. L. Reichardt J. Roberson R. P. Rollins A. K. Romer C. Romero A. Roodman A. J. Ross E. S. Rykoff L. Salvati C. Sánchez E. Sanchez D. Sanchez Cid A. Saro T. Schrabback M. Schubnell L. F. Secco I. Sevilla-Noarbe K. Sharon E. Sheldon T. Shin M. Smith T. Somboonpanyakul B. Stalder A. A. Stark V. Strazzullo E. Suchyta M. E. C. Swanson G. Tarle C. To M. A. Troxel I. Tutusaus T. N. Varga A. von der Linden N. Weaverdyck J. Weller P. Wiseman B. Yanny B. Yin M. Young Y. Zhang J. Zuntz University Observatory Faculty of Physics Universität Innsbruck Institut für Astro- und Teilchenphysik Technikerstr. 25/8 6020 Innsbruck Austria Argonne National Laboratory 9700 South Cass Avenue Lemont Illinois 60439 USA Kavli Institute for Cosmological Physics University of Chicago Chicago Illinois 60637 USA Max Planck Institute for Extraterrestrial Physics Gießenbachstr. 1 85748 Garching Germany Laboratório Interinstitucional de e-Astronomia - LIneA Rua Gal. José Cristino 77 Rio de Janeiro RJ-20921-400 Brazil Fermi National Accelerator Laboratory P. O. Box 500 Batavia Illinois 60510 USA Kavli Institute for Particle Astrophysics and Cosmology Stanford University 452 Lomita Mall Stanford California 94305 USA Department of Physics Stanford University 382 Via Pueblo Mall Stanford California 94305 USA SLAC National Accelerator Laboratory 2575 Sand Hill Road Menlo Park California 94025 USA Department of Physics University of Michigan Ann Arbor Michigan 48109 USA Institute of Astronomy University of Cambridge Madingley Road Cambridge CB3 0HA United Kingdom Kavli Institute for Cosmology University of Cambridge Madingley Road Cambridge CB3 0HA United Kingdom School of Physics University of Melbourne Parkville Victoria 3010 Australia Institute of Cosmology and Gravitation University of Portsmouth Portsmouth PO1 3FX United Kingdom Department of Physics University of Cincinnati Cincinnati Ohio 45221 USA Physics Department 2320 Chamberlin Hall University of Wisconsin-Madison 1150 University Avenue Madison Wisconsin 53706-1390 USA Department of Astronomy and Astrophysics University of Chicago Chicago Illinois 60637 USA Fermi National Accelerator Laboratory Batavia Illinois 60510-0500 USA Department of Physics and Astronomy University of Pennsylvania Philadelphia Pennsylvania 19104 USA Department of Physics and Astronomy University of Missouri 5110 Rockhill Road Kansas City Missouri 64110 USA Department of Physics and Astronomy University Coll

We present a Bayesian population modeling method to analyze the abundance of galaxy clusters identified by the South Pole Telescope (SPT) with a simultaneous mass calibration using weak gravitational lensing data from the Dark Energy Survey (DES) and the Hubble Space Telescope (HST). We discuss and validate the modeling choices with a particular focus on a robust, weak-lensing-based mass calibration using DES data. For the DES Year 3 data, we report a systematic uncertainty in weak-lensing mass calibration that increases from 1% at z=0.25 to 10% at z=0.95, to which we add 2% in quadrature to account for uncertainties in the impact of baryonic effects. We implement an analysis pipeline that joins the cluster abundance likelihood with a multiobservable likelihood for the Sunyaev-Zel’dovich effect, optical richness, and weak-lensing measurements for each individual cluster. We validate that our analysis pipeline can recover unbiased cosmological constraints by analyzing mocks that closely resemble the cluster sample extracted from the SPT-SZ, SPTpol ECS, and SPTpol 500d surveys and the DES Year 3 and HST-39 weak-lensing datasets. This work represents a crucial prerequisite for the subsequent cosmological analysis of the real dataset.

关键词： Cosmological parameters Gravitational lenses Large scale structure of the Universe

来源：评论

学校读者我要写书评

暂无评论

Enabling QoE-aware Mobile Cloud Video Recording over Roadside Vehicular Networks

引用

China Communications 2016年第8期13卷 63-73页

作者： Tongtong Jiang Zhaobin Deng Weiwei Huang Guoqing Zhang Department of Computer Science Sun Yat-sen University Guangdong Province Key Laboratory of Big Data Analysis and Processing Sun Yat-sen University Collaborative Innovation Center of High Performance Computing National University of Defense Technology Institute of Computing Technology Chinese Academy of Sciences

Most of previous video recording devices in mobile vehicles commonly store captured video contents locally. With the rapid development of 4G/Wi Fi networks, there emerges a new trend to equip video recording devices with wireless interfaces to enable video uploading to the cloud for video playback in a later time point. In this paper, we propose a QoE-aware mobile cloud video recording scheme in the roadside vehicular networks, which can adaptively select the proper wireless interface and video bitrate for video uploading to the cloud. To maximize the total utility, we need to design a control strategy to carefully balance the transmission cost and the achieved QoE for users. To this purpose, we investigate the tradeoff between cost incurred by uploading through cellular networks and the achieved QoE of users. We apply the optimization framework to solve the formulated problem and design an online scheduling algorithm. We also conduct extensive trace-driven simulations and our results show that our algorithm achieves a good balance between the transmission cost and user QoE.

关键词： mobile cloud video vehicular networks quality-of-experience

来源：评论

学校读者我要写书评

暂无评论

Private model compression via knowledge distillation

arXiv

引用

arXiv 2018年

作者： Wang, Ji Bao, Weidong Sun, Lichao Zhu, Xiaomin Cao, Bokai Yu, Philip S. College of Systems Engineering National University of Defense Technology Changsha China Department of Computer Science University of Illinois at Chicago Chicago United States State Key Laboratory of High Performance Computing National University of Defense Technology Changsha China Facebook Inc. Menlo Park United States Institute for Data Science Tsinghua University Beijing China

The soaring demand for intelligent mobile applications calls for deploying powerful deep neural networks (DNNs) on mobile devices. However, the outstanding performance of DNNs notoriously relies on increasingly complex models, which in turn is associated with an increase in computational expense far surpassing mobile devices' capacity. What is worse, app service providers need to collect and utilize a large volume of users' data, which contain sensitive information, to build the sophisticated DNN models. Directly deploying these models on public mobile devices presents prohibitive privacy risk. To benefit from the on-device deep learning without the capacity and privacy concerns, we design a private model compression framework RONA. Following the knowledge distillation paradigm, we jointly use hint learning, distillation learning, and self learning to train a compact and fast neural network. The knowledge distilled from the cumbersome model is adaptively bounded and carefully perturbed to enforce differential privacy. We further propose an elegant query sample selection method to reduce the number of queries and control the privacy loss. A series of empirical evaluations as well as the implementation on an Android mobile device show that RONA can not only compress cumbersome models efficiently but also provide a strong privacy guarantee. For example, on SVHN, when a meaningful (9:83;10-6)-differential privacy is guaranteed, the compact model trained by RONA can obtain 20× compression ratio and 19× speed-up with merely 0.97% accuracy loss. Copyright © 2018, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Cross-view Asymmetric Metric Learning for Unsupervised Person Re-identification

arXiv

引用

arXiv 2017年

作者： Yu, Hong-Xing Wu, Ancong Zheng, Wei-Shi School of Data and Computer Science Sun Yat-sen University China School of Electronics and Information Technology Sun Yat-sen University China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China Collaborative Innovation Center of High Performance Computing NUDT China Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China

While metric learning is important for Person reidentification (RE-ID), a significant problem in visual surveillance for cross-view pedestrian matching, existing metric models for RE-ID are mostly based on supervised learning that requires quantities of labeled samples in all pairs of camera views for training. However, this limits their scalabilities to realistic applications, in which a large amount of data over multiple disjoint camera views is available but not labelled. To overcome the problem, we propose unsupervised asymmetric metric learning for unsupervised RE-ID. Our model aims to learn an asymmetric metric, i.e., specific projection for each view, based on asymmetric clustering on cross-view person images. Our model finds a shared space where view-specific bias is alleviated and thus better matching performance can be achieved. Extensive experiments have been conducted on a baseline and five large-scale RE-ID datasets to demonstrate the effectiveness of the proposed model. Through the comparison, we show that our model works much more suitable for unsupervised RE-ID compared to classical unsupervised metric learning models. We also compare with existing unsupervised REID methods, and our model outperforms them with notable margins. Specifically, we report the results on large-scale unlabelled RE-ID dataset, which is important but unfortunately less concerned in literatures. Copyright © 2017, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

SPT clusters with DES and HST weak lensing. II. Cosmological constraints from the abundance of massive halos

引用

Physical Review D 2024年第8期110卷 083510页

作者： S. Bocquet S. Grandis L. E. Bleem M. Klein J. J. Mohr T. Schrabback T. M. C. Abbott P. A. R. Ade M. Aguena A. Alarcon S. Allam S. W. Allen O. Alves A. Amon A. J. Anderson J. Annis B. Ansarinejad J. E. Austermann S. Avila D. Bacon M. Bayliss J. A. Beall K. Bechtol M. R. Becker A. N. Bender B. A. Benson G. M. Bernstein S. Bhargava F. Bianchini M. Brodwin D. Brooks L. Bryant A. Campos R. E. A. Canning J. E. Carlstrom A. Carnero Rosell M. Carrasco Kind J. Carretero F. J. Castander R. Cawthon C. L. Chang C. Chang P. Chaubal R. Chen H. C. Chiang A. Choi T-L. Chou R. Citron C. Corbett Moran J. Cordero M. Costanzi T. M. Crawford A. T. Crites L. N. da Costa M. E. S. Pereira C. Davis T. M. Davis J. DeRose S. Desai T. de Haan H. T. Diehl M. A. Dobbs S. Dodelson C. Doux A. Drlica-Wagner K. Eckert J. Elvin-Poole S. Everett W. Everett I. Ferrero A. Ferté A. M. Flores J. Frieman J. Gallicchio J. García-Bellido M. Gatti E. M. George G. Giannini M. D. Gladders D. Gruen R. A. Gruendl N. Gupta G. Gutierrez N. W. Halverson I. Harrison W. G. Hartley K. Herner S. R. Hinton G. P. Holder D. L. Hollowood W. L. Holzapfel K. Honscheid J. D. Hrubes N. Huang J. Hubmayr E. M. Huff D. Huterer K. D. Irwin D. J. James M. Jarvis G. Khullar K. Kim L. Knox R. Kraft E. Krause K. Kuehn N. Kuropatkin F. Kéruzoré O. Lahav A. T. Lee P.-F. Leget D. Li H. Lin A. Lowitz N. MacCrann G. Mahler A. Mantz J. L. Marshall J. McCullough M. McDonald J. J. McMahon J. Mena-Fernández F. Menanteau S. S. Meyer R. Miquel J. Montgomery J. Myles T. Natoli A. Navarro-Alsina J. P. Nibarger G. I. Noble V. Novosad R. L. C. Ogando Y. Omori S. Padin S. Pandey P. Paschos S. Patil A. Pieres A. A. Plazas Malagón A. Porredon J. Prat C. Pryke M. Raveri C. L. Reichardt J. Roberson R. P. Rollins C. Romero A. Roodman J. E. Ruhl E. S. Rykoff B. R. Saliwanchik L. Salvati C. Sánchez E. Sanchez D. Sanchez Cid A. Saro K. K. Schaffer L. F. Secco I. Sevilla-Noarbe K. Sharon E. Sheldon T. Shin C. Sievers G. Smecher M. Smith T. Somboonpanyakul M. Sommer B. Stalder A. A. Stark J. Stephen V. Straz University Observatory Faculty of Physics Universität Innsbruck Institut für Astro- und Teilchenphysik Technikerstraße 25/8 6020 Innsbruck Austria High-Energy Physics Division Argonne National Laboratory 9700 South Cass Avenue Lemont Illinois 60439 USA Kavli Institute for Cosmological Physics University of Chicago 5640 South Ellis Avenue Chicago Illinois 60637 USA Max Planck Institute for Extraterrestrial Physics Gießenbachstraße 1 85748 Garching Germany Argelander-Institut für Astronomie Auf dem Hügel 71 53121 Bonn Germany Cerro Tololo Inter-American Observatory NSF’s National Optical-Infrared Astronomy Research Laboratory Casilla 603 La Serena Chile School of Physics and Astronomy Cardiff University Cardiff CF24 3YB United Kingdom Laboratório Interinstitucional de e-Astronomia—LIneA Rua Gal. José Cristino 77 Rio de Janeiro Rio de Janeiro—20921-400 Brazil Fermi National Accelerator Laboratory P. O. Box 500 Batavia Illinois 60510 USA Kavli Institute for Particle Astrophysics and Cosmology Stanford University 452 Lomita Mall Stanford California 94305 USA Department of Physics Stanford University 382 Via Pueblo Mall Stanford California 94305 USA SLAC National Accelerator Laboratory 2575 Sand Hill Road Menlo Park California 94025 USA Department of Physics University of Michigan Ann Arbor Michigan 48109 USA Institute of Astronomy University of Cambridge Madingley Road Cambridge CB3 0HA United Kingdom Kavli Institute for Cosmology University of Cambridge Madingley Road Cambridge CB3 0HA United Kingdom School of Physics University of Melbourne Parkville Victoria 3010 Australia NIST Quantum Devices Group 325 Broadway Mailcode 817.03 Boulder Colorado 80305 USA Department of Physics University of Colorado Boulder Colorado 80309 USA Institut de Física d’Altes Energies (IFAE) The Barcelona Institute of Science and Technology Campus UAB 08193 Bellaterra (Barcelona) Spain Institute of Cosmology and Gravitation University of Portsmouth Portsmouth

We present cosmological constraints from the abundance of galaxy clusters selected via the thermal Sunyaev-Zel’dovich (SZ) effect in South Pole Telescope (SPT) data with a simultaneous mass calibration using weak gravitational lensing data from the Dark Energy Survey (DES) and the Hubble Space Telescope (HST). The cluster sample is constructed from the combined SPT-SZ, SPTpol ECS, and SPTpol 500d surveys, and comprises 1,005 confirmed clusters in the redshift range 0.25–1.78 over a total sky area of 5200 deg2. We use DES Year 3 weak-lensing data for 688 clusters with redshifts z<0.95 and HST weak-lensing data for 39 clusters with 0.6dataset, we place a 95% upper limit on the sum of neutrino masses ∑mν<0.18 eV. When additionally allowing the dark energy equation of state parameter w to vary, we obtain w=−1.45±0.31 from our cluster-based analysis. In combination with Planck data, we measure w=−1.34−0.15+0.22, or a 2.2σ difference with a cosmological constant. We use the cluster abundance to measure σ8 in five redshift bins between 0.25 and 1.8, and we find the results to be consistent with structure growth as predicted by the ΛCDM model fit to Planck primary CMB data.

关键词： Cosmological parameters Dark energy Large scale structure of the Universe Galaxy clusters

来源：评论

学校读者我要写书评

暂无评论

Uniform Information Exchange in Multi-channel Wireless Ad Hoc Networks 17

Uniform Information Exchange in Multi-channel Wireless Ad Ho...

引用

International Conference on Autonomous Agents and Multiagent Systems

作者： Dongxiao Yu Li Ning Yong Zhang Hai Jin Yuexuan Wang Francis C. M. Lau Shengzhong Feng Services Computing Technology and System Lab Big Data Technology and System Lab Clusters and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Center for High Performance Computing Shenzhen Institutes of Advanced Technology CAS Department of Computer Science The University of Hong Kong

ISBN: (纸本)9781510855076

Information exchange is a basic primitive for maintaining the smooth running of a network or a system with multiple communicating agents. Given k packets initially stored at k nodes respectively, the problem is to disseminate the k packets to the whole network with the objective of minimizing the time used. We study this problem in single-hop multi-channel networks of n nodes, and target on devising uniform distributed protocols that do not rely on any prior knowledge of network parameters, such as the network size n or the number of packet holders k. Uniform protocols have better scalability and are more suitable for implementation in reality Specifically, we propose a uniform distributed protocol that with high probability accomplishes the dissemination in O(k/F+F·log n) rounds, assuming F available channels. This protocol is asymptotically optimal when k is large (k ≥ F~2 log n), and provides the best possible linear speedup with multiple channels comparing with the results using a single channel. To the best of our knowledge, this is the first uniform protocol for information exchange in multi-channel networks.

关键词： Information exchange Multi-channel networks Distributed algorithm Uniform protocol information exchange distributed algorithms MULTICHANNEL Packet Uniform Uniforms Mobile ad hoc networks distributed protocol network parameters single-channel

来源：评论

学校读者我要写书评

暂无评论

Not just privacy: Improving performance of private deep learning in mobile cloud

arXiv

引用

arXiv 2018年

作者： Wang, Ji Zhang, Jianguo Bao, Weidong Zhu, Xiaomin Cao, Bokai Yu, Philip S. College of Systems Engineering National University of Defense Technology Changsha China Department of Computer Science University of Illinois at Chicago Chicago United States College of Systems Engineering State Key Laboratory of High Performance Computing National University of Defense Technology Changsha China Facebook Inc. Menlo Park United States Department of Computer Science University of Illinois at Chicago Chicago United States Institute for Data Science Tsinghua University Beijing China

The increasing demand for on-device deep learning services calls for a highly efficient manner to deploy deep neural networks (DNNs) on mobile devices with limited capacity. The cloud-based solution is a promising approach to enabling deep learning applications on mobile devices where the large portions of a DNN are offloaded to the cloud. However, revealing data to the cloud leads to potential privacy risk. To benefit from the cloud data center without the privacy risk, we design, evaluate, and implement a cloud-based framework Arden which partitions the DNN across mobile devices and cloud data centers. A simple data transformation is performed on the mobile device, while the resource-hungry training and the complex inference rely on the cloud data center. To protect the sensitive information, a lightweight privacy-preserving mechanism consisting of arbitrary data nullification and random noise addition is introduced, which provides strong privacy guarantee. A rigorous privacy budget analysis is given. Nonetheless, the private perturbation to the original data inevitably has a negative impact on the performance of further inference on the cloud side. To mitigate this influence, we propose a noisy training method to enhance the cloud-side network robustness to perturbed data. Through the sophisticated design, Arden can not only preserve privacy but also improve the inference performance. To validate the proposed Arden, a series of experiments based on three image datasets and a real mobile application are conducted. The experimental results demonstrate the effectiveness of Arden. Finally, we implement Arden on a demo system to verify its practicality. Copyright © 2018, The Authors. All rights reserved.

关键词： Deep neural networks

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：