Cloud contamination represents a large obstacle for mapping the earth9;s surface using remotely sensed data. therefore, cloudy pixels should be identified and eliminated before any further data processing can be ac...
详细信息
ISBN:
(纸本)9781510638570;9781510638587
Cloud contamination represents a large obstacle for mapping the earth's surface using remotely sensed data. therefore, cloudy pixels should be identified and eliminated before any further data processing can be achieved. Although several threshold, multi-temporal and machine learning applications have been developed to tackle this issue, it still remains a challenge. the main challenges are imposed by the difficulty to detect thin clouds and to separate bright clouds from bright non-cloud objects. Convolutional neural networks (CNNs) have proven to be one of the most promising methods for image classification tasks and their use is rapidly increasing in remote sensing problems. CNNs present interesting properties for image processing since they directly exploit not only the spectral information but also the spatial covariance of the data. In this work, we study the applicability of CNNs in cloud detection of Sentinel-2 imagery, a complex remote sensing problem with crucial spatial context. A patch-to-pixel CNN architecture consisting of three convolutional layers and two fully connected layers is trained on a recently available manually created public dataset. the results were evaluated both qualitatively and quantitatively through comparison with ground truth cloud masks and state-of-the art pixel-based algorithms (Fmask, Sen2Cor). It was shown that the proposed architecture even though simpler than the deep learning architectures proposed by recent literature, performs very favorably, especially in the challenging cases. Besides the evaluation of the results, feature maps where observed as an initial effort to extract the weights of the useful kernels for cloud masking applications.
Over the past few years, MCAST has been increasing its commitment towards research, yet a common limitation is the availability of processing power by our researchers. Whilst funding is available for the setting up of...
详细信息
ISBN:
(纸本)9781728127453
Over the past few years, MCAST has been increasing its commitment towards research, yet a common limitation is the availability of processing power by our researchers. Whilst funding is available for the setting up of classroom computers, there isn't enough funding for high-end processing infrastructure for research purposes. We have thus undertaken a research project to explore the possibility of repurposing existing computer systems and have them work in a cluster during out-of-office hours to address just this need. In this initial phase of our research, we are presenting our findings on a simple proof of concept using seven single-board computers benchmarked as a Beowulf and Dispy clusters. the intention is to further this research by testing up to 25 single-board computers then to migrate our setup and findings to 120 computers. Our findings in this research are very encouraging and justify the next phases of this project.
the analysis of Big data technologies was provided. An example of MapReduce paradigm application, uploading of big volumes of data, processing and analyzing of unstructured information and its distribution into the cl...
详细信息
the analysis of Big data technologies was provided. An example of MapReduce paradigm application, uploading of big volumes of data, processing and analyzing of unstructured information and its distribution into the clustered database was provided. the article summarizes the concept of "big data". Examples of methods for working with arrays of unstructured data are given. the parallel system Resilient Distributed Datasets (RDD) is organized. the class of basic database operations was realized: database con-nection, table creation, getting in line id, returning all elements of the database, update, delete and create the line. (C) 2019the Authors. Published by Elsevier B.V.
Striped variation of the Smith-Waterman algorithm is known as extremely efficient and easily adaptable for the SIMD architectures. However, the potential for improvement has not been exhausted yet. the popular Lazy-F ...
详细信息
ISBN:
(纸本)9781728146171
Striped variation of the Smith-Waterman algorithm is known as extremely efficient and easily adaptable for the SIMD architectures. However, the potential for improvement has not been exhausted yet. the popular Lazy-F loop heuristic requires additional memory access operations, and the worst-case performance of the loop could be as bad as the nonvectorized version. We demonstrate the progression of the lazy-F loop transformations that improve the loop performance, and ultimately eliminate the loop completely. Our algorithm achieves the best asymptotic performance of all scan-based SW algorithms O(n/p+log(p)), and is very efficient in practice.
the growing scale of applications encoded to Boolean Satisfiability (SAT) problems imposes the need for accelerating SAT simplifications or preprocessing. parallel SAT preprocessing has been an open challenge for many...
详细信息
ISBN:
(纸本)9783030174620;9783030174613
the growing scale of applications encoded to Boolean Satisfiability (SAT) problems imposes the need for accelerating SAT simplifications or preprocessing. parallel SAT preprocessing has been an open challenge for many years. therefore, we propose novel parallelalgorithms for variable and subsumption elimination targeting Graphics processing Units (GPUs). Benchmarks show that the algorithms achieve an acceleration of 66x over a state-of-the-art SAT simplifier (SatELite). Regarding SAT solving, we have conducted a thorough evaluation, combining both our GPU algorithms and SatELite with MiniSat to solve the simplified problems. In addition, we have studied the impact of the algorithms on the solvability of problems with Lingeling. We conclude that our algorithms have a considerable impact on the solvability of SAT problems.
Network packet processingarchitectures use heterogeneous processors as accelerators to speed-up classic application domain tasks. Our platform compiles applications to bytecodes for a generalized packet processing ma...
详细信息
ISBN:
(纸本)9781538694039
Network packet processingarchitectures use heterogeneous processors as accelerators to speed-up classic application domain tasks. Our platform compiles applications to bytecodes for a generalized packet processing machine, then uses microcoded interpreters running in parallel to trigger accelerators as needed. To make the system effective requires helping users debug apps, which includes tracking runtime exceptions. Exception tracking is complicated when a system-thrown exception is detected on an accelerator and the current binary form is far removed from the original high-level language source or associated assembly code. We tackle this problem by (1) instrumenting the compiler and a low-level bytecode tool, (2) reporting exceptions withthe interpreter, (3) creating a specialized tool to collate the higher-level program forms withthe lower-level bytecode forms. this functionality provides data needed for post-mortem program analysis.
the greatest common divisor (GCD) is used for numerous applications in number theory, modular arithmetic, encryption algorithms such as RSA, random number generation, and solving linear Diophantine equations. High-per...
详细信息
ISBN:
(数字)9781728186559
ISBN:
(纸本)9781728186566
the greatest common divisor (GCD) is used for numerous applications in number theory, modular arithmetic, encryption algorithms such as RSA, random number generation, and solving linear Diophantine equations. High-performance algorithms, which are efficiently and accurately find the GCD of two integers and n (>2) integers, are needed in the modern world for many applications in science and engineering. parallel hardware and parallel programming solve such emerging challenges that are not possible in a serial world. Modern desktop and laptop computers are equipped with multicore processors with shared memory architecture. In this paper, we develop novel efficient parallel n integers GCD algorithms for multicore shared memory architecture. the brute force, divide-and-conquer, linear recursive and finding minimum first techniques are adopted in our novel algorithms to reduce the size and the complexity of the n integers GCD problem. Various working models of OpenMP, such as the thread-centric, loop-centric and task-centric models are utilized, which promised a more natural way of exploiting and expressing regular and irregular algorithms. A comprehensive performance analysis applies to prove the efficiency of the proposed algorithms.
the proceedings contain 232 papers. the topics discussed include: convolutional neural network-based approach for citrus diseases recognition;using temporal conceptual graphs and neural networks for big data-based att...
ISBN:
(纸本)9781728143286
the proceedings contain 232 papers. the topics discussed include: convolutional neural network-based approach for citrus diseases recognition;using temporal conceptual graphs and neural networks for big data-based attack scenarios reconstruction;taxi demand prediction with LSTM-based combination model;counting attention based on classification confidence for visual question answering;an optimization method of WebP images lossy compression algorithm for FPGAs;self-adaptive address mapping mechanism for access pattern awareness on DRAM;a high-performance self-learning antelopes migration algorithm (SAMA) for global optimization;and using resource use data and system logs for HPC system error propagation and recovery diagnosis.
Numerous architectures for medium-voltage power electronic systems are investigated. To find out the architectures suitable for the medium-voltage photovoltaic power electronic systems, this paper gives a systematic s...
详细信息
ISBN:
(数字)9781728153018
ISBN:
(纸本)9781728153025
Numerous architectures for medium-voltage power electronic systems are investigated. To find out the architectures suitable for the medium-voltage photovoltaic power electronic systems, this paper gives a systematic study on medium-voltage power electronic systems (MVPESs) architectures for large-scale PV plants. A synthesis method of MVPESs is firstly introduced. It is composed of two steps. First, the power conversion chains are synthesized by traversing all possible designs of the multi-stage converters. Based on the power conversion chains, topologies are obtained by cascade connection or parallel connection. A data structure representing the MVPESs is also introduced that can help to synthesize the MVPESs using computer. the advantages and disadvantages of these MVPESs are also discussed. Two typical MVPES architectures are compared with efficiency and cost.
Phonocardiogram known as PCG plays a significant role in the early diagnosis of cardiac abnormalities. Phonocardiogram can be used as initial diagnostics tool in remote applications due to its simplicity and cost effe...
详细信息
暂无评论