Explore and consider methods, tools, and system design for video and image analysis withcloud scaling. As described in earlier articles in this series, video analytics requires a morebalanced data-centric compute architecture compared to traditional compute-centric, scalable,high-performance computing. The author examines the use of OpenCV and similar tools fordigital video analysis and methods to scale this analysis using cluster and distributed systemdesign.The use of coprocessors designed for video analytics and the new OpenVX hardwareacceleration discussed in previous articles can be applied to the computer vision (CV)examples presented in this article. This new data-centric technology for CV and video analyticsrequires the system designer to re-think application software and system design to meetdemanding requirements, such as real-time monitoring and security for large, public facilitiesand infrastructure as well as a more entertaining, interactive, and safer world.

关键词：视频分析；基本结构；云缩放；信息安全

查看摘要 索取原文[12页]

Evolutionary forces driving the IT industry toward cloud computing increase the pressureon application server management solutions to both accommodate traditional infrastructuredeployments and enable cloud-based deployments. Lightweight, standards-based, flexiblescalability are among the necessary hallmarks of an architecture that embraces this duality.Liberty collectives deliver next generation application server management for the IBM ebSphere Application Server Liberty profile both inside and outside the cloud.

关键词：剖析；发展模型；配置管理；云计算

查看摘要 索取原文[11页]

We consider distributed memory algorithms for the all-pairs shortest paths (APSP) problem. Scaling the APSP problem to high concurrencies requires both minimizing inter-processor communication as well as maximizing temporal data locality. Our 2.5D APSP algorithm, which is based on the divideand- conquer paradigm, satisfies both of these requirements: it can utilize the extra available memory to perform asymptotically less communication, and it is rich in semiring matrix multiplications, which have high temporal locality. We start by introducing a block-cyclic 2D (minimal memory) APSP algorithm. With a careful choice of block-size, this algorithm achieves known communication lower-bounds on latency and bandwidth. We extend this 2D block-cyclic algorithm to a 2.5D algorithm, which can use extra copies of data to reduce the bandwidth cost by a factor of c1=2, compared to its 2D counterpart.However, the 2.5D algorithm increases the latency cost by c1=2. We provide a tighter lower bound on latency, which dictates that the latency overhead is necessary to reduce bandwidth along the critical path of execution. Our implementation achieves impressive performance and scaling to 24,576 cores of a Cray XE6 supercomputer by utilizing well-tuned intra-node kernels within the distributed memory algorithm.

关键词：最短路径算法；矩阵乘法；关键路径

查看摘要 索取原文[18页]

This paper and poster presented a description of the current real-time SPoRT-LIS run over the southeastern CONUS to provide high-resolution, land surface initialization grids for local numerical model forecasts at NWS forecast offices. The LIS hourly output also offers a supplemental dataset to aid in situational awareness for convective initiation forecasts, assessing flood potential, and monitoring drought at fine scales. It is a goal of SPoRT and several NWS forecast offices to expand the LIS to an entire CONUS domain, so that LIS output can be utilized by NWS Western Region offices, among others. To make this expansion cleanly so as to provide high-quality land surface output, SPoRT tested new precipitation datasets in LIS as an alternative forcing dataset to the current radar+gauge Stage IV product. Similar to the Stage IV product, the NMQ product showed comparable patterns of precipitation and soil moisture distribution, but suffered from radar gaps in the intermountain West, and incorrectly set values to zero instead of missing in the data-void regions of Mexico and Canada. The other dataset tested was the next-generation GOES-R QPE algorithm, which experienced a high bias in both coverage and intensity of accumulated precipitation relative to the control (NLDAS2), Stage IV, and NMQ simulations. The resulting root zone soil moisture was substantially higher in most areas.

关键词：电子信息；信息系统；土地；压力

查看摘要 索取原文[10页]

IBM PureApplication System does not currently allow a full backup of all the storage on thesystem. Instead, what it does is permit a back up of the system configuration of PureApplicationSystem that would enable a system to be reconstructed in the case of a total PSM failure.In addition, the system can optionally also back up the virtual system images. We currentlyrecommend that you allocate about 160 GB of storage to cover the system db files and userimages, although this figure could change over time as you create new custom images and obtainnew patterns. This system and image backup is currently implemented through SSH copies, and isdone incrementally, where only deltas are recorded after the first full system backup.Since PureApplication System does not allow a full backup of all storage, users of PureApplicationSystem have to be more selective about what other parts of the user storage (besides the systemconfiguration above) is backed up. Remember that PureApplication System essentially runssystem workloads that consist of instances of patterns that are composed of virtual images, plusthe deployable artifacts (application code, such as JARs, WARs, and EARs) that are deployedinto those pattern instances. Given the automation that is either built into the system or that werecommend be added onto the system, what we contend is that the pattern instances can beentirely recreated from the patterns and the deployment automation at any time. Thus, the onlyinformation that can't be easily reconstructed would be user-specific data, such as databasecontents or the contents of specific files that hold user data. Thus, those are the only parts of theremaining data that need to be backed up. So, what we recommend is that any database that is stored on PureApplication System havean IBM Tivoli?Storage Manager agent configured on it so that it can be backed up onto a TivoliStorage Manager infrastructure. This is something that is already preconfigured in the Databaseas a Service (DBAAS) images. For other IBM DB2?images (for example, the built-in DB2HADRimages) you have to write a simple script package to install and configure the Tivoli StorageManager agent to point to the location of your existing Tivoli Storage Manager servers.

关键词：可视化系统；数据库恢复；系统和图像备份

查看摘要 索取原文[11页]

Information Security is traditionally treated in three main categories: Con dentiality, Integrity, and Availability. While much work has been done on modelling Con dentiality and Availability, aspects involving comprehensive modelling and quality of data integrity in complex systems appear to be, on a relative scale, much less well understood and implemented. Further, most work on Integrity and resultant implementations seems to have focussed more on a matters related to source authentication and transmission assurance. However, the quality of data aspect is becoming more critical for attention, given the increasing levels of automation of information fusion and data transformation in a globalised Cyberspace. In this paper, we survey the existing integrity models and identify shortcomings of these with regard to a general integrity framework encompassing the quality of data aspect. We then propose and formally model a new framework, illustrating the approach with reference to use cases built around the Secure Information ATM (SIATM) - a highly accreditable security system currently under development.

关键词：电子信息；信息系统；安全；应用

查看摘要 索取原文[41页]

IBM PureApplication System is an integrated system that combines hardware resources(compute nodes, storage, and networking) with software resources, using virtualization tomaximize the utilization of the system. Workloads (for example, transactional applications, webapplications, or data-centric applications) are supported via the use of patterns, which offerabstract views of the software components of a solution; for example, web applications, ordatabases. This article describes the various options that exist in IBM PureApplication Systemand IBM Workload Deployer for the creation of scalable middleware deployments using virtualapplication patterns.

关键词：水平和垂直缩放；可视化应用模型；手动缩放

查看摘要 索取原文[13页]

This article describes how to use the IBM?Image Construction and Composition Tool tocapture an image of a Windows?virtual machine running on a VMware ESX hypervisorexternal to IBM PureApplication?System. Then, after capturing the base image, you will seehow you can use the tool to configure your Windows image for use in virtual system and virtualapplication patterns within PureApplication System.

关键词：虚拟机；构建及合成工具；云服务提供商

查看摘要 索取原文[42页]

We consider optimal representations for representing prototypical categories in the latent deformable part model framework, with a specific emphasis on category-level retrieval tasks defined “on the fly” for a large corpus. In this
setting, it is impractical to perform an exhaustive search with a full model; we investigate methods which approximately reconstruct the score function of a novel category from a set of precomputed responses. We propose a novel
sparse reconstruction method where part classifiers are decomposed via a shared dictionary of part filters; in turn, our method can efficiently reconstruct approximate part responses on large image corpora using a sparse matrixvector product based on pre-computed filter responses instead of exhaustive convolutions. We compare our method to baseline schemes using SVD-based or nearest-category approximation and show our method is more effective at detecting novel categories. We additionally demonstrate results towards an end-to-end system for activity detection which trains a protoype category concept model from one dataset (PASCAL), learns post-hoc categories on the fly based on training data from a second dataset where labeled data are available (ImageNet), and sucessfully detects instances
in test data from a third dataset (TRECVID MED) via reconstruction with the precomputed prototype models.

关键词：模型框架原型;类级别检索;穷举搜索;稀疏重建方法;

查看摘要 索取原文[10页]

Discover the reasons for introducing lease policies into large cloud environments. This articleelaborates on one specific pattern deployment lease policy tailored for IBM PureApplication System, IBM SmartCloud Provisioning, and IBM Workload Deployer. The design andimplementation described here is a proven example of building lease policies for a cloudenvironment.

关键词：模式部署租用策略；实施；希望

查看摘要 索取原文[9页]