Many data processing problems are successfully solved by artificial neural networks (ANN) possessing the property of a universal approximator. However, in case when the number of data patterns available is small, ANN ...
详细信息
ISBN:
(纸本)9783319993164;9783319993157
Many data processing problems are successfully solved by artificial neural networks (ANN) possessing the property of a universal approximator. However, in case when the number of data patterns available is small, ANN may tend to overtrain and not to generalize well enough. An alternative is use of such a biologically inspired cognitive architecture as fuzzy networks, or Adaptive Neuro-Fuzzy Inference Systems (ANFIS), based on the notions of fuzzy logics and often used in control systems. Like conventional ANN, ANFIS can be also trained by example with error backpropagation algorithm. In this study, we demonstrate use of neuro-fuzzy networks to solve a classification problem for high-dimensional, highly variable and noisy data of chemical sensors. the results are compared to those obtained by a multi-layer perceptron ANN and by linear regression.
Nowadays, GPUs are known as one of the most important, most remarkable, and perhaps most popular computing platforms. In recent years, GPUs have increasingly been considered as co-processors and accelerators. Along wi...
详细信息
ISBN:
(纸本)9781728150758
Nowadays, GPUs are known as one of the most important, most remarkable, and perhaps most popular computing platforms. In recent years, GPUs have increasingly been considered as co-processors and accelerators. Along with growing technology, Graphics processing Units (GPUs) with more advanced features and capabilities are manufactured and launched by the world's largest commercial companies. Unified memory is one of these new features introduced on the latest generations of Nvidia GPUs which allows programmers to write a program considering the uniform memory shared between CPU and GPU. this feature makes programming considerably easier. the present study introduces this new feature and its attributes. In addition, a model is proposed to predict the execution time of applications if using unified memory style programming based on the information of non-unified style implementation. the proposed model can predict the execution time of a kernel with an average accuracy of 87.60%.
this article describes an approach to parallelizing of data mining algorithms in logical programming framework, for distributed data processing in cluster. As an example Naive Bayes algorithm implementation in Prolog ...
详细信息
ISBN:
(纸本)9783030308599;9783030308582
this article describes an approach to parallelizing of data mining algorithms in logical programming framework, for distributed data processing in cluster. As an example Naive Bayes algorithm implementation in Prolog framework, its conversion into parallel type and execution on cluster with MPI system are described.
the proceedings contain 18 papers. the special focus in this conference is on High Performance Computing in Computational Science. the topics include: A Scheduling theory Framework for GPU Tasks Efficient Execution;A ...
ISBN:
(纸本)9783030159955
the proceedings contain 18 papers. the special focus in this conference is on High Performance Computing in Computational Science. the topics include: A Scheduling theory Framework for GPU Tasks Efficient Execution;A Timer-Augmented Cost Function for Load Balanced DSMC;Accelerating Scientific Applications on Heterogeneous Systems with HybridOMP;a New parallel Benchmark for Performance Evaluation and Energy Consumption;bigger Buffer k-d Trees on Multi-Many-Core Systems;a parallel Generator of Non-Hermitian Matrices Computed from Given Spectra;LRMalloc: A Modern and Competitive Lock-Free Dynamic Memory Allocator;towards a Strategy for Performance Prediction on Heterogeneous architectures;Dynamic Configuration of CUDA Runtime Variables for CDP-Based Divide-and-Conquer algorithms;Design, Implementation and Performance Analysis of a CFD Task-Based Application for Heterogeneous CPU/GPU Resources;Optimizing Packed String Matching on AVX2 Platform;A GPU-Based Metaheuristic for Workflow Scheduling on Clouds;a Systematic Mapping on High-Performance Computing for Protein Structure Prediction;performance Evaluation of Deep Learning Frameworks over Different architectures;non-uniform Domain Decomposition for Heterogeneous Accelerated processing Units;performance Evaluation of Two Load Balancing algorithms for Hybrid Clusters.
this paper proposes the use of parallel approximate sorting networks in Non-Binary LDPC decoders and quantifies their impact in terms of performance and complexity. Motivated by the messages truncation concept of the ...
详细信息
ISBN:
(数字)9781728160443
ISBN:
(纸本)9781728160450
this paper proposes the use of parallel approximate sorting networks in Non-Binary LDPC decoders and quantifies their impact in terms of performance and complexity. Motivated by the messages truncation concept of the Extended Min Sum (EMS) algorithm and the reduction it induces on the decoding complexity, we seek for an approximate parallel sorting strategy for the Elementary Check Node (ECN) Unit processing in the EMS algorithm. We quantify the impact of parallel approximate sorting on both error correction decoding performance and hardware complexity. Bubble check sorting methods are compared to proposed parallel approximate sorting techniques in terms of BER vs. noise plots. Furthermore hardware synthesis results are provided to demonstrate the effects of parallel sorting networks on area and throughput. In certain cases of practical interest, a modest degradation of 0.2 dB in coding gain is imposed; however due to proposed parallel approximate sorters, the latency of ECN can be reduced by up to a factor of two or be logarithmic to the input sequence length, in certain architectures. To better illustrate the impact of the proposed techniques, a complete NB-LDPC decoder has been implemented and used for evaluation; it is found that a substantial reduction in the number of clock cycles required per decoding iteration is achieved by the proposed techniques, more than 50% in certain cases.
the success of Deep Learning (DL) algorithms in computer vision tasks have created an on-going demand of dedicated hardware architecturesthat could keep up withthe their required computation and memory complexities....
详细信息
ISBN:
(纸本)9781450371896
the success of Deep Learning (DL) algorithms in computer vision tasks have created an on-going demand of dedicated hardware architecturesthat could keep up withthe their required computation and memory complexities. this task is particularly challenging when embedded smart camera platforms have constrained resources such as power consumption, processing Element (PE) and communication. this article describes a heterogeneous system embedding an FPGA and a GPU for executing CNN inference for computer vision applications. the built system addresses some challenges of embedded CNN such as task and data partitioning, and workload balancing. the selected heterogeneous platform embeds an Nvidia (R) Jetson TX2 for the CPU-GPU side and an Intel Altera (R) Cyclone10GX for the FPGA side interconnected by PCIe Gen2 with a MIPI-CSI camera for prototyping. this test environment will be used as a support for future work on a methodology for optimized model partitioning.
In this paper, information indicators for solving statistical problems at various stages of analyzing psychological test data are analyzed: in the process of detecting and removing from the sample incorrectly formulat...
详细信息
ISBN:
(纸本)9783319993164;9783319993157
In this paper, information indicators for solving statistical problems at various stages of analyzing psychological test data are analyzed: in the process of detecting and removing from the sample incorrectly formulated test tasks, for grouping participants according to similarity indicators of their answers in test tasks and highlighting participants with unique characteristics. the proposed methodology is illustrated by the data of a psychological test aimed at identifying a person's ability to spatial imagination, the formation of associative links between objects and the solution of logical problems.
this paper presents comprehensive analysis of main SIMD-processing features and computational characteristics of three high performance architectures: two NVIDIA GPU architectures (of Pascal and Volta generations) and...
详细信息
Pedestrian evacuation is one of the most crucial research fields in safety management. Staircase is a main escape route when pedestrians evacuate from the multi-story building. According to different priorities about ...
详细信息
ISBN:
(纸本)9783319945897;9783319945880
Pedestrian evacuation is one of the most crucial research fields in safety management. Staircase is a main escape route when pedestrians evacuate from the multi-story building. According to different priorities about evacuation, it is necessary to make a rational strategy that reducing the casualties and property damage effectively. the article makes an attempt to establish a extended network model, which is used to make further applications in more complex architectures to analyze the evacuation. the results of our study could provide individual instructions for risk avoidance and effective strategies for safety management.
Withthe rapid development of deep learning (DL), various convolution neural network (CNN) models have been developed. Moreover, to execute different DL workloads efficiently, many accelerators have been proposed. To ...
详细信息
暂无评论