In recent years, the problem of lake eutrophication has become increasingly severe. the monitoring and control of cyanobacteria in lakes are of great significance. the information obtained by existing monitoring metho...
详细信息
Digital processing of remotely sensed image data has been great importance in recent times. this research work discusses task distribution method in parallelimageprocessing and load balancing under the circumstance ...
详细信息
MapReduce has been widely used to process large-scale data in the past decade. Among the quantity of such cloud computing applications, we pay special attention to distributed mosaic methods based on numerous drone im...
详细信息
Modern large-scale distributed computing systems, processing large volumes of data, require mature monitoring systems able to control and track in resources, networks, computing tasks, queues and other components. In ...
详细信息
As the computational cost and datasets available for deep neural network training continue to increase, there is a significant demand for fast distributed training on supercomputers. However, porting and tuning applic...
详细信息
ISBN:
(纸本)9781665410175
As the computational cost and datasets available for deep neural network training continue to increase, there is a significant demand for fast distributed training on supercomputers. However, porting and tuning applications for new advanced supercomputers requires tremendous amount of development efforts. therefore, we present software tuning best practice for a 3D-CNN model training on a new Arm CPU based supercomputer, Fugaku. We (i) tune computation in DL by a JIT translator for aarch64, (ii) optimize collective communication such as Allreduce for 6D mesh/torus network topology, (iii) tune I/O by data staging with compression and data loader with caching, and (iv) parallelize training in data and model parallelism. We apply the proposed methods to a CosmoFlow 3D-CNN model, and achieve the training in 30 minutes using 16,384 nodes consisting of 4096 data- and 4 model-parallelism. this is the fastest result of any CPU-based systems in MLPerf HPC v0.7 in the world.
the proceedings contain 68 papers. the special focus in this conference is on Human Centered Computing. the topics include: Wi-Fi Attention Network for Indoor Fingerprint Positioning;Reinforcement Learning Based Coope...
ISBN:
(纸本)9783030151263
the proceedings contain 68 papers. the special focus in this conference is on Human Centered Computing. the topics include: Wi-Fi Attention Network for Indoor Fingerprint Positioning;Reinforcement Learning Based Cooperation Transmission Policy for HetNets with CoMP Technology;capacity Estimation of Time-Triggered Ethernet Network Based on Complex Network theory;a New Communication P System Model Based on Hypergraph;Non-Orthogonal Multiple Access (NOMA) in Providing Services for High-Speed Railway and Local Users in DownLink MIMO System;research on Multi-agent distributed Supply Chain Information Collaboration Based on Cloud Environment;mobile Internet Mobile Agent System Dynamic Trust Model for Cloud Computing;Single image Super-Resolution by parallel CNN with Skip Connections and ResNet;incorporating Description Embeddings into Medical Knowledge Graphs Representation Learning;Research on the Classification and Channel Selection of Emotional EEG;A Piezoelectric MEMS Harvester Suitable Adopt a New Two-Degree-of-Freedom Structure;spiking Neural P Systems with Time Delay;energy Efficiency MapReduce Job Scheduling of Shuffle and Reduce Phases in Data Center;research on Weibo Emotion Classification Based on Context;research on Data Visualization in Different Scenarios;model Checking for Turn-Based Probability Epistemic Game Structure;depressive Emotion Recognition Based on Behavioral Data;knowledge Graph Embedding by Translation Model on Subgraph;linguistic Signatures of Impulsive Buying Consumer Based on Microblog;Research on the Model of Anomaly Detection of FMCG Based on Time Series: Illustrated by the Case of Cosmetics;research on Real-Time Low Air image Intelligence image Acquisition and processingmethods;research on Font Emotion Based on Semantic Difference Method;simulation Research on Aircraft Anti-collision Algorithm.
the general view on a situation with supercomputer calculations in Russia and in the world, including questions of education is briefly presented in the report. the technologies used in educational processes of a numb...
详细信息
ISBN:
(纸本)9781538658321
the general view on a situation with supercomputer calculations in Russia and in the world, including questions of education is briefly presented in the report. the technologies used in educational processes of a number of the Russian higher education institutions for support of teaching the parallel and distributed calculations are discussed. Some subjects of projects for performance within the research work (RW) offered students of department of Aapplied mathematics of MPEI are considered, results of the analysis and research of processes of the solution of problems of various classes on multinuclear processors are presented. the classical tasks studied by students in different disciplines, such as numerical methods, the theory of grahps, artificial intelligence, etc. act as tasks. the main objective of the conducted researches - to define efficiency of the constructed decisions, to estimate the expedient range of number of parallel processes, to compare labor input of the traditional consecutive decision and parallel calculations. Further students continue the researches with use of the MPI technology on the distributed architecture.
image clustering is one of the challenging tasks in machine learning, and has been extensively used in various applications. Recently, various deep clustering methods has been proposed. these methods take a two-stage ...
详细信息
image clustering is one of the challenging tasks in machine learning, and has been extensively used in various applications. Recently, various deep clustering methods has been proposed. these methods take a two-stage approach, feature learning and clustering, sequentially or jointly. We observe that these works usually focus on the combination of reconstruction loss and clustering loss, relatively little work has focused on improving the learning representation of the neural network for clustering. In this paper, we propose a deep convolutional embedded clustering algorithm with inception-like block (DCECI). Specifically, an inception-like block with different type of convolution filters are introduced in the symmetric deep convolutional network to preserve the local structure of convolution layers. We simultaneously minimize the reconstruction loss of the convolutional autoencoders with inception-like block and the clustering loss. Experimental results on multiple image datasets exhibit the promising performance of our proposed algorithm compared with other competitive methods.
image fusion synthesizes two or more source images having same view into a single fused image, which contains all significant details of the scene. this chapter proposes a new image fusion approach based on discrete c...
详细信息
ISBN:
(纸本)9781509036691
image fusion synthesizes two or more source images having same view into a single fused image, which contains all significant details of the scene. this chapter proposes a new image fusion approach based on discrete cosine transform (DCT) and sum-modified-laplacian (SINIL) in stationary wavelet transform (SWT) domain. Here, two source images of less information are fused to produce a more informative and better visual quality image. the performance of the proposed method is evaluated on various categories of multi-focus images and compared with some existing fusion methods. the simulation results exhibit the better performance of the proposed methods in both quantitative and qualitative measurements.
the image contains a lot of visual as well as hidden information. Both, information must be secured at the time of transmission. Withthis motivation, a scheme is proposed based on encryption in tetrolet domain. For e...
详细信息
ISBN:
(纸本)9781509036691
the image contains a lot of visual as well as hidden information. Both, information must be secured at the time of transmission. Withthis motivation, a scheme is proposed based on encryption in tetrolet domain. For encryption, an iterative based Arnold transform is used in proposed methodology. the images are highly textured, which contains the authenticity of the image. For that, decryption process is performed in this way so that maximum, the edges and textures should be recovered, effectively. the suggested method has been tested on standard images and results obtained after applying suggested method are significant. A comparison is also performed with some standard existing methods to measure the effectiveness of the suggested method.
暂无评论