Brain big data processing with massively parallel computing technology: challenges and opportunities

Chen, Dan; Hu, Yangyang; Cai, Chang; Zeng, Ke; Li, Xiaoli

doi:10.1002/spe.2418

Cited by 20 publications

(15 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, the probability of the regions to be a sensitive object when a feature or a group of features presented. Massive parallel computing framework has exhibited great potentials in scientific big data analysis, and it will be exploited to figure out the object saliency related features. Moreover, the Coding Tree Unit partitions could be optimized when taken the new calculations of distortion.…”

Section: Resultsmentioning

confidence: 99%

A sensitive object‐oriented approach to big surveillance data compression for social security applications in smart cities

et al. 2016

View full text Add to dashboard Cite

Surveillance has become a fairly common practice with the global boom in "smart cities". How to efficiently store and manage the vast quantities of surveillance data is a persistent challenge in terms of analyzing social security problems. Developing data compression technology under the analytic requirements of surveillance data is the key to solving the storage problem. Criminal investigation demands the quality preservation of sensitive objects, typically pedestrians, human faces, vehicles, and license plates; however, the analytical value of surveillance data is rapidly lost as the compression ratio increases. In this paper, we propose a sensitive object-oriented regions of interest-based coding strategy for preserving the analytical value of surveillance data. In the proposed method, instead of generating a saliency map based on human visual perception, we consider saliency as a set of characteristics important for object detection and recognition. By making this modification, almost all sensitive objects necessary in a criminal investigation are assigned high saliency value rather than only one or two salient regions. Motions in the temporal domain are integrated to place emphasis on moving objects, namely moving sensitive objects, which then gain the highest saliency. Finally, a saliency-based rate control algorithm embedded in High Efficiency Video Coding is used to maintain the quality of sensitive objects in the encoded video under a fixed bitrate. Experiments were conducted on two analytical indexes: Feature similarity and object detection accuracy. The results showed that by achieving the same feature similarity and object detection accuracy, our method can save 20% and 40% bitrate over High Efficiency Video Coding, respectively, for the storage of big surveillance data. be archived. Depending on the retention policy, video streams (which are in increasingly higher resolution) can consume storage rapidly, leading to high financial pressure to scale the storage infrastructure. To relieve the financial pressure, normally a high compression ratio is adopted to make a trade-off between the retention period and the video quality. However, according to previous research 1, the reduced quality of surveillance video severely degrades the performance in video analysis tasks such as object recognition and tracking. This makes critical applications such as large crowd management in public hotspot areas very unreliable 2.Our ability to maintain video analysis performance while resolving the surveillance video storage problem is dependent on our ability to accurately and effectively account for the characteristics of the videos. In normal applications, surveillance videos are used for analytical purposes. Taking a typical criminal investigation application, for instance, the analytically sensitive objects (e.g., pedestrians and vehicles) in the videos have much higher importance than the background. Therefore, it is reasonable to divide surveillance video frames into regions of interest (ROIs) and non-ROIs where R...

show abstract

Section: Resultsmentioning

confidence: 99%

A sensitive object‐oriented approach to big surveillance data compression for social security applications in smart cities

et al. 2016

View full text Add to dashboard Cite

show abstract

“…Besides, the single speech feature is too few to reflect one's real emotion. The combination of various features from the whole body such as heart rate and facial expression is a bigger challenge because all these features will form a high-dimension and non-linear dataset, which requires massive computing resources and we should guarantee the efficiency, scalability and reliability of the system [20].…”

Section: Discussionmentioning

confidence: 99%

Deep learning and SVM‐based emotion recognition from Chinese speech for smart affective services

et al. 2017

View full text Add to dashboard Cite

SummaryEmotion recognition is challenging for understanding people and enhances human–computer interaction experiences, which contributes to the harmonious running of smart health care and other smart services. In this paper, several kinds of speech features such as Mel frequency cepstrum coefficient, pitch, and formant were extracted and combined in different ways to reflect the relationship between feature fusions and emotion recognition performance. In addition, we explored two methods, namely, support vector machine (SVM) and deep belief networks (DBNs), to classify six emotion status: anger, fear, joy, neutral status, sadness, and surprise. In the SVM‐based method, we used SVM multi‐classification algorithm to optimize the parameters of penalty factor and kernel function. With DBN, we adjusted different parameters to achieve the best performance when solving different emotions. Both gender‐dependent and gender‐independent experiments were conducted on the Chinese Academy of Sciences emotional speech database. The mean accuracy of SVM is 84.54%, and the mean accuracy of DBN is 94.6%. The experiments show that the DBN‐based approach has good potential for practical usage, and suitable feature fusions will further improve the performance of speech emotion recognition. Copyright © 2017 John Wiley & Sons, Ltd.

show abstract

“…However, instead of relying on monitoring the application performance and take reallocation decisions based upon its degradation, a more proactive scheduling is needed to improve the suitability of the resource selection. Feeding the scheduler with more fine‐grained hardware requirements, provided by the user request, such as the low‐latency interconnects (eg, Infiniband, 10 GbE) or the GPGPU blue selection, provides a better resource categorization and, consequently, will directly contribute to a more efficient execution of the application. To accomplish this, the specialized hardware must be exposed into the virtual instances, by means of a Peripheral Component Interface (PCI) pass‐through with Input‐Output Memory Management Unit (IOMMU) or Single Root I/O Virtualization techniques, and eventually managed by the cloud middleware using the underlying virtualization stack.…”

Section: Resource Provisioning In Science Cloudsmentioning

confidence: 99%

Resource provisioning in Science Clouds: Requirements and challenges

García

Fernández-del-Castillo

Fernández

et al. 2017

Softw Pract Exp

View full text Add to dashboard Cite

SummaryCloud computing has permeated into the IT industry in the last few years, and it is nowadays emerging in scientific environments. Science user communities are demanding a broad range of computing power to satisfy high-performance applications needs, such as local clusters, High Performance Computing (HPC) systems and computing grids. Different workloads need from different computational models, and the cloud is already considered as a promising paradigm.The scheduling and allocation of resources is always a challenging matter in any form of computation and clouds are not an exception. Science applications have unique features that differentiate their workloads, hence their requirements have to be taken into consideration to be fulfilled when building a Science Cloud. This paper will discuss what are the main scheduling and resource allocation challenges for any Infrastructure as a Service IaaS provider supporting scientific applications.

show abstract

Brain big data processing with massively parallel computing technology: challenges and opportunities

Cited by 20 publications

References 64 publications

A sensitive object‐oriented approach to big surveillance data compression for social security applications in smart cities

A sensitive object‐oriented approach to big surveillance data compression for social security applications in smart cities

Deep learning and SVM‐based emotion recognition from Chinese speech for smart affective services

Resource provisioning in Science Clouds: Requirements and challenges

Contact Info

Product

Resources

About