Search results for: iterative computation

Authors: Muhammad A. Alsubaie

Abstract:

An iterative learning control framework designed in state feedback structure suffers a lack in investigating load disturbance considerations. The presented work discusses the controller previously designed, highlights the disturbance problem, finds new conditions using singular value principle to assure safe operation conditions with error convergence and reference tracking under the influence of load disturbance. It is known that periodic disturbances can be represented by a delay model in a positive feedback loop acting on the system input. This model can be manipulated by isolating the delay model and finding a controller for the overall system around the delay model to remedy the periodic disturbances using the small signal theorem. The overall system is the base for control design and load disturbance investigation. The major finding of this work is the load disturbance condition found which clearly sets safe operation condition under the influence of load disturbances such that the error tends to nearly zero as the system keeps operating trial after trial.

Keywords: iterative learning control, singular values, state feedback, load disturbance

Procedia PDF Downloads 134

Authors: Iuliia Zarubiieva, Joyun Tseng, Vishwesh Kulkarni

Abstract:

Recent researches has focused on nucleic acids as a substrate for designing biomolecular circuits for in situ monitoring and control. A common approach is to express them by a set of idealised abstract chemical reaction networks (ACRNs). Here, we present new results on how abstract chemical reactions, viz., catalysis, annihilation and degradation, can be used to implement circuit that accurately computes logarithm function using the method of Arithmetic-Geometric Mean (AGM), which has not been previously used in conjunction with ACRNs.

Keywords: chemical reaction networks, ratio computation, stability, robustness

Procedia PDF Downloads 127

Authors: Abdullah Bal, Sevdenur Bal

Abstract:

This paper purposes a new type of hardware application for training of cellular neural networks (CNN) using optical joint transform correlation (JTC) architecture for image feature extraction. CNNs require much more computation during the training stage compare to test process. Since optoelectronic hardware applications offer possibility of parallel high speed processing capability for 2D data processing applications, CNN training algorithm can be realized using Fourier optics technique. JTC employs lens and CCD cameras with laser beam that realize 2D matrix multiplication and summation in the light speed. Therefore, in the each iteration of training, JTC carries more computation burden inherently and the rest of mathematical computation realized digitally. The bipolar data is encoded by phase and summation of correlation operations is realized using multi-object input joint images. Overlapping properties of JTC are then utilized for summation of two cross-correlations which provide less computation possibility for training stage. Phase-only JTC does not require data rearrangement, electronic pre-calculation and strict system alignment. The proposed system can be incorporated simultaneously with various optical image processing or optical pattern recognition techniques just in the same optical system.

Keywords: CNN training, image processing, joint transform correlation, optoelectronic hardware

Procedia PDF Downloads 469

Authors: N. Dahraoui, M. Boulakroune, D. Benatia

Abstract:

In this paper, the improvement by deconvolution of the depth resolution in Secondary Ion Mass Spectrometry (SIMS) analysis is considered. Indeed, we have developed a new Tikhonov-Miller deconvolution algorithm where a priori model of the solution is included. This is a denoisy and pre-deconvoluted signal obtained from: firstly, by the application of wavelet shrinkage algorithm, secondly by the introduction of the obtained denoisy signal in an iterative deconvolution algorithm. In particular, we have focused the light on the effect of the iterations number on the evolution of the deconvoluted signals. The SIMS profiles are multilayers of Boron in Silicon matrix.

Keywords: DRF, in-depth resolution, multiresolution deconvolution, SIMS, wavelet shrinkage

Procedia PDF Downloads 380

Authors: Apurva Saxena, Sunita Gond

Abstract:

Demand of computation, data storage in large amount are rapidly increases day by day. Cloud computing technology fulfill the demand of today’s computation but this will lead to high power consumption in cloud data centers. Initiative for Green IT try to reduce power consumption and its adverse environmental impacts. Paper also focus on various green computing techniques, proposed models and efficient way to make cloud greener.

Keywords: virtualization, cloud computing, green computing, data center

Procedia PDF Downloads 514

Authors: Naser Alajmi, Ali Alobaidly, Mubarak Alhajri, Salem Salamah, Muhammad Alsubaie

Abstract:

Iterative Learning Control (ILC) known to be a controlling tool to overcome periodic disturbances for repetitive systems. This technique is required to let the error signal tends to zero as the number of operation increases. The learning process that lies within this context is strongly dependent on the initial input which if selected properly tends to let the learning process be more effective compared to the case where a system starts from blind. ILC uses previous recorded execution data to update the following execution/trial input such that a reference trajectory is followed to a high accuracy. Error convergence in ILC is generally highly dependent on the input applied to a plant for trial $1$, thus a good choice of initial starting input signal would make learning faster and as a consequence the error tends to zero faster as well. In the work presented within, an upper limit based on the Singular Values Principle (SV) is derived for the initial input signal applied at trial $1$ such that the system follow the reference in less number of trials without responding aggressively or exceeding the working envelope where a system is required to move within in a robot arm, for example. Simulation results presented illustrate the theory introduced within this paper.

Keywords: initial input, iterative learning control, maximum input, singular values

Procedia PDF Downloads 205

Authors: Umar Yusuf Batsari

Abstract:

Let E be a uniformly smooth and uniformly convex real Banach space and C be a nonempty, closed and convex subset of E. Let $V= \{S_i : C\to C, ~i=1, 2, 3\cdots N\}$ be a convex set of relatively nonexpansive mappings containing identity. In this paper, an iterative sequence obtained from CQ algorithm was shown to have strongly converge to a point $\hat{x}$ which is a common fixed point of relatively nonexpansive mappings in V and also solve the system of equilibrium problems in E. The result improve some existing results in the literature.

Keywords: relatively nonexpansive mappings, strong convergence, equilibrium problems, uniformly smooth space, uniformly convex space, convex set, kadec klee property

Procedia PDF Downloads 386

Authors: Jie Yang, Jian-Min Meng

Abstract:

A GPU-based floating object detection scheme is presented in this paper which is designed for floating mine detection tasks. This system uses contrast and motion information to eliminate as many false positives as possible while avoiding false negatives. The GPU computation platform is deployed to allow detecting objects in real-time. From the experimental results, it is shown that with certain configuration, the GPU-based scheme can speed up the computation up to one thousand times compared to the CPU-based scheme.

Keywords: object detection, GPU, motion estimation, parallel processing

Procedia PDF Downloads 438

Authors: Aditi Viswanathan, Shree Ranjani, Aruna Govada

Abstract:

With data sizes constantly expanding, and with classical machine learning algorithms that analyze such data requiring larger and larger amounts of computation time and storage space, the need to distribute computation and memory requirements among several computers has become apparent. Although substantial work has been done in developing distributed binary SVM algorithms and multi-class SVM algorithms individually, the field of multi-class distributed SVMs remains largely unexplored. This research seeks to develop an algorithm that implements the Support Vector Machine over a multi-class data set and is efficient in a distributed environment. For this, we recursively choose the best binary split of a set of classes using a greedy technique. Much like the divide and conquer approach. Our algorithm has shown better computation time during the testing phase than the traditional sequential SVM methods (One vs. One, One vs. Rest) and out-performs them as the size of the data set grows. This approach also classifies the data with higher accuracy than the traditional multi-class algorithms.

Keywords: distributed algorithm, MapReduce, multi-class, support vector machine

Procedia PDF Downloads 360

Authors: Chae Hoon Park, Jong Kang Park, Jong Tae Kim

Abstract:

Electrical characteristics of capacitive touchscreen need to be accurately analyzed to result in better performance for multi-channel capacitance sensing. In this paper, we extracted the panel resistances and capacitances of the touchscreen by comparing measurement data and model data. By employing a lumped RC model for driver-to-receiver paths in touchscreen, we estimated resistance and capacitance values according to the physical lengths of channel paths which are proportional to the RC model. As a result, we obtained the model having 95.54% accuracy of the measurement data.

Keywords: electrical characteristics of capacitive touchscreen, iterative extraction, lumped RC model, physical lengths of channel paths

Procedia PDF Downloads 306

Authors: Wang Xingbo

Abstract:

Computation of determinant in the form |I-X| is primary and fundamental because it can help to compute many other determinants. This article puts forward a time-reducible approach to compute determinant |I-X|. The approach is derived from the Newton’s identity and its time complexity is no more than that to compute the eigenvalues of the square matrix X. Mathematical deductions and numerical example are presented in detail for the approach. By comparison with classical approaches the new approach is proved to be superior to the classical ones and it can naturally reduce the computational time with the improvement of efficiency to compute eigenvalues of the square matrix.

Keywords: algorithm, determinant, computation, eigenvalue, time complexity

Procedia PDF Downloads 382

Authors: Anteneh Getachew Gebrie, Rabian Wangkeeree

Abstract:

The purpose of this paper is to introduce iterative algorithm solving split system of minimization problem given as a task of finding a common minimizer point of finite family of proper, lower semicontinuous convex functions and whose image under a bounded linear operator is also common minimizer point of another finite family of proper, lower semicontinuous convex functions. We obtain strong convergence of the sequence generated by our algorithm under some suitable conditions on the parameters. The iterative schemes are developed with a way of selecting the step sizes such that the information of operator norm is not necessary. Some applications and numerical experiment is given to analyse the efficiency of our algorithm.

Keywords: Hilbert Space, minimization problems, Moreau-Yosida approximate, split feasibility problem

Procedia PDF Downloads 102

Authors: Sandeep Singh

Abstract:

In this article, two classes of iterative schemes are proposed for approximating solutions of nonlinear systems of equations whose orders of convergence are six and eight respectively. Sixth order scheme requires the evaluation of two vector-functions, two first Fr'echet derivatives and three matrices inversion per iteration. This three-step sixth-order method is further extended to eighth-order method which requires one more step and the evaluation of one extra vector-function. Moreover, computational efficiency is compared with some other recently published methods in which we found, our methods are more efficient than existing numerical methods for higher and medium size nonlinear system of equations. Numerical tests are performed to validate the proposed schemes.

Keywords: Nonlinear systems, Computational complexity, order of convergence, Jarratt-type scheme

Procedia PDF Downloads 102

Authors: Royan Dawud Aldian, Endah Purwanti, Soegianto Soelistiono

Abstract:

In this research we have been developed an automatic investigation to classify normal children voice or autistic by using modern computation technology that is computation based on artificial neural network. The superiority of this computation technology is its capability on processing and saving data. In this research, digital voice features are gotten from the coefficient of linear-predictive coding with auto-correlation method and have been transformed in frequency domain using fast fourier transform, which used as input of artificial neural network in back-propagation method so that will make the difference between normal children and autistic automatically. The result of back-propagation method shows that successful classification capability for normal children voice experiment data is 100% whereas, for autistic children voice experiment data is 100%. The success rate using back-propagation classification system for the entire test data is 100%.

Keywords: autism, artificial neural network, backpropagation, linier predictive coding, fast fourier transform

Procedia PDF Downloads 416

Authors: John Pomerat, Aviv Segev

Abstract:

Recently, deep learning has had many theoretical breakthroughs. For deep learning to be successful in the industry, however, there need to be practical algorithms capable of handling many real-world hiccups preventing the immediate application of a learning algorithm. Although AI promises to revolutionize the healthcare industry, getting access to patient data in order to train learning algorithms has not been easy. One proposed solution to this is data- sharing. In this paper, we propose an alternative protocol, based on multi-party computation, to train deep learning models while maintaining both the privacy and security of training data. We examine three methods of training neural networks in this way: Transfer learning, average ensemble learning, and series network learning. We compare these methods to the equivalent model obtained through data-sharing across two different experiments. Additionally, we address the security concerns of this protocol. While the motivating example is healthcare, our findings regarding multi-party computation of neural network training are purely theoretical and have use-cases outside the domain of healthcare.

Keywords: neural network aggregation, multi-party computation, transfer learning, average ensemble learning

Procedia PDF Downloads 120

Authors: Dalvinder Kaur Mangal

Abstract:

For the past decade, noise corrupted output measurements have been a fundamental research problem to be investigated. On the other hand, the estimation of the parameters for linear dynamic systems when also the input is affected by noise is recognized as more difficult problem which only recently has received increasing attention. Representations where errors or measurement noises/disturbances are present on both the inputs and outputs are usually called errors-in-variables (EIV) models. These disturbances may also have additive effects which are also considered in this paper. Parameter estimation of false EIV problem using equation error, output error and iterative prefiltering identification schemes with and without additive uncertainty, when only the output observation is corrupted by noise has been dealt in this paper. The comparative study of these three schemes has also been carried out.

Keywords: errors-in-variable (EIV), false EIV, equation error, output error, iterative prefiltering, Gaussian noise

Procedia PDF Downloads 450

Authors: Paschal A. Ochang, Emmanuel C. Oji

Abstract:

The Duffing’s Equation is a second order system that is very important because they are fundamental to the behaviour of higher order systems and they have applications in almost all fields of science and engineering. In the biological area, it is useful in plant stem dependence and natural frequency and model of the Brain Crash Analysis (BCA). In Engineering, it is useful in the study of Damping indoor construction and Traffic lights and to the meteorologist it is used in the prediction of weather conditions. However, most Problems in real life that occur are non-linear in nature and may not have analytical solutions except approximations or simulations, so trying to find an exact explicit solution may in general be complicated and sometimes impossible. Therefore we aim to find out if it is possible to obtain one analytical fixed point to the non-linear ordinary equation using fixed point analytical method. We started by exposing the scope of the Duffing’s equation and other related works on it. With a major focus on the fixed point and fixed point iterative scheme, we tried different iterative schemes on the Duffing’s Equation. We were able to identify that one can only see the fixed points to a Damped Duffing’s Equation and not to the Undamped Duffing’s Equation. This is because the cubic nonlinearity term is the determining factor to the Duffing’s Equation. We finally came to the results where we identified the stability of an equation that is damped, forced and second order in nature. Generally, in this research, we approximate the solution of Duffing’s Equation by converting it to a system of First and Second Order Ordinary Differential Equation and using Fixed Point Iterative approach. This approach shows that for different versions of Duffing’s Equations (damped), we find fixed points, therefore the order of computations and running time of applied software in all fields using the Duffing’s equation will be reduced.

Keywords: damping, Duffing's equation, fixed point analysis, second order differential, stability analysis

Procedia PDF Downloads 250

Authors: Sheng-Po Tseng, Che-Hua Yang

Abstract:

Motivation and Objective: Ultrasonic guided waves become an important tool for nondestructive evaluation of structures and components. Guided waves are used for the purpose of identifying defects or evaluating material properties in a nondestructive way. While guided waves are applied for evaluating material properties, instead of knowing the properties directly, preliminary signals such as time domain signals or frequency domain spectra are first revealed. With the measured ultrasound data, inversion calculation can be further employed to obtain the desired mechanical properties. Methods: This research is development of high speed inversion calculation technique for obtaining full-field mechanical properties from the quantitative laser ultrasound visualization system (QLUVS). The quantitative laser ultrasound visualization system (QLUVS) employs a mirror-controlled scanning pulsed laser to generate guided acoustic waves traveling in a two-dimensional target. Guided waves are detected with a piezoelectric transducer located at a fixed location. With a gyro-scanning of the generation source, the QLUVS has the advantage of fast, full-field, and quantitative inspection. Results and Discussions: This research introduces two important tools to improve the computation efficiency. Firstly, graphic procession unit (GPU) with large amount of cores are introduced. Furthermore, combining the CPU and GPU cores, parallel procession scheme is developed for the inversion of full-field mechanical properties based on the QLUVS data. The newly developed inversion scheme is applied to investigate the computation efficiency for single-layered and double-layered plate-like samples. The computation efficiency is shown to be 80 times faster than unparalleled computation scheme. Conclusions: This research demonstrates a high-speed inversion technique for the characterization of full-field material properties based on quantitative laser ultrasound visualization system. Significant computation efficiency is shown, however not reaching the limit yet. Further improvement can be reached by improving the parallel computation. Utilizing the development of the full-field mechanical property inspection technology, full-field mechanical property measured by non-destructive, high-speed and high-precision measurements can be obtained in qualitative and quantitative results. The developed high speed computation scheme is ready for applications where full-field mechanical properties are needed in a nondestructive and nearly real-time way.

Keywords: guided waves, material characterization, nondestructive evaluation, parallel processing

Procedia PDF Downloads 166

Authors: Vinai K. Singh

Abstract:

For reducing impulsive noise without degrading image contours, median filtering is a powerful tool. In multiband images as for example colour images or vector fields obtained by optic flow computation, a vector median filter can be used. Vector median filters are defined on the basis of a suitable distance, the best performing distance being the Euclidean. Euclidean distance is evaluated by using the Euclidean norms which is quite demanding from the point of view of computation given that a square root is required. In this paper an optimal piece-wise linear approximation of the Euclidean norm is presented which is applied to vector median filtering.

Keywords: euclidean norm, quasi euclidean norm, vector median filtering, applied mathematics

Procedia PDF Downloads 429

Authors: Ebrahim Alrashed, Yousef Rafique

Abstract:

Mobile devices have drastically changed the way we do things on the move. They are being extremely relied on to perform tasks that are analogous to desktop computer capability. There has been a rapid increase of computational power on these devices; however, battery technology is still the bottleneck of evolution. The primary modern approach day approach to tackle this issue is offloading computation to the cloud, proving to be latency expensive and requiring high network bandwidth. In this paper, we explore efforts to perform barter-based mobile-to-mobile offloading. We present define a protocol and present an architecture to facilitate the development of such a system. We further highlight the deployment and security challenges.

Keywords: computational offloading, power conservation, cloud, sandboxing

Procedia PDF Downloads 355

Authors: Peter M. Maurer Maurer

Abstract:

DGL is a context free grammar-based tool for generating random data. Many types of simulator input data require some computation to be placed in the proper format. For example, it might be necessary to generate ordered triples in which the third element is the sum of the first two elements, or it might be necessary to generate random numbers in some sorted order. Although DGL is universal in computational power, generating these types of data is extremely difficult. To overcome this problem, we have enhanced DGL to include features that permit direct computation within the structure of a context free grammar. The features have been implemented as special types of productions, preserving the context free flavor of DGL specifications.

Keywords: DGL, Enhanced Context Free Grammars, Programming Constructs, Random Data Generation

Procedia PDF Downloads 113

Authors: Jing Zhang, Daniel Nikovski

Abstract:

We propose an approximation algorithm called LINKUMP to compute the Pan Matrix Profile (PMP) under the unnormalized l∞ distance (useful for value-based similarity search) using double-ended queue and linear interpolation. The algorithm has comparable time/space complexities as the state-of-the-art algorithm for typical PMP computation under the normalized l₂ distance (useful for shape-based similarity search). We validate its efficiency and effectiveness through extensive numerical experiments and a real-world anomaly detection application.

Keywords: pan matrix profile, unnormalized euclidean distance, double-ended queue, discord discovery, anomaly detection

Procedia PDF Downloads 206

Authors: Rafat Alshorman, Fadi Awawdeh

Abstract:

By employing a simplified bilinear method, a class of generalized fifth-order KdV (gfKdV) equations which arise in nonlinear lattice, plasma physics and ocean dynamics are investigated. With the aid of symbolic computation, both solitary wave solutions and multiple-soliton solutions are obtained. These new exact solutions will extend previous results and help us explain the properties of nonlinear solitary waves in many physical models in shallow water. Parametric analysis is carried out in order to illustrate that the soliton amplitude, width and velocity are affected by the coefficient parameters in the equation.

Keywords: multiple soliton solutions, fifth-order evolution equations, Cole-Hopf transformation, Hirota bilinear method

Procedia PDF Downloads 286

Authors: F. Jiménez, R. Jódar, M. Martín, G. Sánchez, G. Sciavicco

Abstract:

Abstract—Attribute or feature selection is one of the basic strategies to improve the performances of data classification tasks, and, at the same time, to reduce the complexity of classifiers, and it is a particularly fundamental one when the number of attributes is relatively high. Its application to unsupervised classification is restricted to a limited number of experiments in the literature. Evolutionary computation has already proven itself to be a very effective choice to consistently reduce the number of attributes towards a better classification rate and a simpler semantic interpretation of the inferred classifiers. We present a feature selection wrapper model composed by a multi-objective evolutionary algorithm, the clustering method Expectation-Maximization (EM), and the classifier C4.5 for the unsupervised classification of data extracted from a psychological test named BASC-II (Behavior Assessment System for Children - II ed.) with two objectives: Maximizing the likelihood of the clustering model and maximizing the accuracy of the obtained classifier. We present a methodology to integrate feature selection for unsupervised classification, model evaluation, decision making (to choose the most satisfactory model according to a a posteriori process in a multi-objective context), and testing. We compare the performance of the classifier obtained by the multi-objective evolutionary algorithms ENORA and NSGA-II, and the best solution is then validated by the psychologists that collected the data.

Keywords: evolutionary computation, feature selection, classification, clustering

Procedia PDF Downloads 332

Authors: Guenther Schuh, Michael Riesener, Frederic Diels

Abstract:

Nowadays, manufacturing companies are faced with the challenge of meeting heterogeneous customer requirements in short product life cycles with a variety of product functions. So far, some of the functional requirements remain unknown until late stages of the product development. A way to handle these uncertainties is the highly iterative product development (HIP) approach. By structuring the development project as a highly iterative process, this method provides customer oriented and marketable products. There are first approaches for combined, hybrid models comprising deterministic-normative methods like the Stage-Gate process and empirical-adaptive development methods like SCRUM on a project management level. However, almost unconsidered is the question, which development scopes can preferably be realized with either empirical-adaptive or deterministic-normative approaches. In this context, a development scope constitutes a self-contained section of the overall development objective. Therefore, this paper focuses on a methodology that deals with the uncertainty of requirements within the early development stages and the corresponding selection of the most appropriate development approach. For this purpose, internal influencing factors like a company’s technology ability, the prototype manufacturability and the potential solution space as well as external factors like the market accuracy, relevance and volatility will be analyzed and combined into an Agile-Indicator. The Agile-Indicator is derived in three steps. First of all, it is necessary to rate each internal and external factor in terms of the importance for the overall development task. Secondly, each requirement has to be evaluated for every single internal and external factor appropriate to their suitability for empirical-adaptive development. Finally, the total sums of internal and external side are composed in the Agile-Indicator. Thus, the Agile-Indicator constitutes a company-specific and application-related criterion, on which the allocation of empirical-adaptive and deterministic-normative development scopes can be made. In a last step, this indicator will be used for a specific clustering of development scopes by application of the fuzzy c-means (FCM) clustering algorithm. The FCM-method determines sub-clusters within functional clusters based on the empirical-adaptive environmental impact of the Agile-Indicator. By means of the methodology presented in this paper, it is possible to classify requirements, which are uncertainly carried out by the market, into empirical-adaptive or deterministic-normative development scopes.

Keywords: agile, highly iterative development, agile-indicator, product development

Procedia PDF Downloads 212

Authors: Jagroop Singh Sidhu, Buta Singh

Abstract:

A low computational deblocking filter including three frequency related modes (smooth mode, intermediate mode, and non-smooth mode for low-frequency, mid-frequency, and high frequency regions, respectively) is proposed. The suggested approach requires zero additions, zero subtractions, zero multiplications (for intermediate region), no divisions (for non-smooth region) and no comparison. The suggested method thus keeps the computation lower and thus suitable for image coding systems based on blocks. Comparison of average number of operations for smooth, non-smooth, intermediate (per pixel vector for each block) using filter suggested by Chen and the proposed method filter suggests that the proposed filter keeps the computation lower and is thus suitable for fast processing algorithms.

Keywords: blocking artifacts, computational complexity, non-smooth, intermediate, smooth

Procedia PDF Downloads 430

Authors: Belgherbi Aicha, Hadjidj Ismahen, Bessaid Abdelhafid

Abstract:

Interpretation of medical images benefits from anatomical and physiological priors to optimize computer-aided diagnosis applications. Segmentation of liver, spleen and kidneys is regarded as a major primary step in the computer-aided diagnosis of abdominal organ diseases. In this paper, a semi-automated method for medical image data is presented for the abdominal organ segmentation data using mathematical morphology. Our proposed method is based on hierarchical segmentation and watershed algorithm. In our approach, a powerful technique has been designed to suppress over-segmentation based on mosaic image and on the computation of the watershed transform. Our algorithm is currency in two parts. In the first, we seek to improve the quality of the gradient-mosaic image. In this step, we propose a method for improving the gradient-mosaic image by applying the anisotropic diffusion filter followed by the morphological filters. Thereafter, we proceed to the hierarchical segmentation of the liver, spleen and kidney. To validate the segmentation technique proposed, we have tested it on several images. Our segmentation approach is evaluated by comparing our results with the manual segmentation performed by an expert. The experimental results are described in the last part of this work.

Keywords: anisotropic diffusion filter, CT images, morphological filter, mosaic image, simultaneous organ segmentation, the watershed algorithm

Procedia PDF Downloads 407

Authors: Sergey Kopysov, Nikita Nedozhogin, Leonid Tonkov

Abstract:

The article presents a parallel iterative solver for large sparse linear systems which can be used on a heterogeneous platform. Traditionally, the problem of solving linear systems does not scale well on multi-CPU/multi-GPUs clusters. For example, most of the attempts to implement the classical conjugate gradient method were at best counted in the same amount of time as the problem was enlarged. The paper proposes the pipelined variant of the conjugate gradient method (PCG), a formulation that is potentially better suited for hybrid CPU/GPU computing since it requires only one synchronization point per one iteration instead of two for standard CG. The standard and pipelined CG methods need the vector entries generated by the current GPU and other GPUs for matrix-vector products. So the communication between GPUs becomes a major performance bottleneck on multi GPU cluster. The article presents an approach to minimize the communications between parallel parts of algorithms. Additionally, computation and communication can be overlapped to reduce the impact of data exchange. Using the pipelined version of the CG method with one synchronization point, the possibility of asynchronous calculations and communications, load balancing between the CPU and GPU for solving the large linear systems allows for scalability. The algorithm is implemented with the combined use of technologies: MPI, OpenMP, and CUDA. We show that almost optimum speed up on 8-CPU/2GPU may be reached (relatively to a one GPU execution). The parallelized solver achieves a speedup of up to 5.49 times on 16 NVIDIA Tesla GPUs, as compared to one GPU.

Keywords: conjugate gradient, GPU, parallel programming, pipelined algorithm

Procedia PDF Downloads 124

Authors: Claudeline D. Cellan

Abstract:

Development is multifaceted, and efforts to hasten growth in all these facets have been gaining traction in recent years. Thus, producing a composite index that is reflective of these multidimensional impacts captures the interests of policymakers. The problem lies in going through a mixture of theoretical, methodological and empirical decisions and complexities which, when done carelessly, can lead to inconsistent and unreliable results. This study looks into index computation from a different and less complex perspective. Borrowing the idea of archetypes or ‘pure types’, archetypal analysis looks for points in the convex hull of the multivariate data set that captures as much information in the data as possible. The archetypes or 'pure types' are estimated such that they are convex combinations of all the observations, which in turn are convex combinations of the archetypes. This ensures that the archetypes are realistically observable, therefore achievable. In the sense of composite indices, we look for the best among these archetypes and use this as a benchmark for index computation. Its straightforward and simplistic approach does away with aggregation and substitutability problems which are commonly encountered in index computation. As an example of the application of archetypal analysis in index construction, the country data for the Human Development Index (HDI 2017) of the United Nations Development Programme (UNDP) is used. The goal of this exercise is not to replicate the result of the UNDP-computed HDI, but to illustrate the usability of archetypal analysis in index construction. Here best is defined in the context of life, education and gross national income sub-indices. Results show that the HDI from the archetypal analysis has a linear relationship with the UNDP-computed HDI.

Keywords: archetypes, composite index, convex combination, development

Procedia PDF Downloads 86

Authors: Zeyu Zhou, Xiaojun Bi

Abstract:

Due to the limited environmental parameters and the limited resolution of the sensor, the universal existence of the mixed pixels in the process of remote sensing images restricts the spatial resolution of the remote sensing images. Sub-pixel mapping technology can effectively improve the spatial resolution. As the bilinear interpolation algorithm inevitably produces the edge blur effect, which leads to the inaccurate sub-pixel mapping results. In order to avoid the edge blur effect that affects the sub-pixel mapping results in the interpolation process, this paper presents a new edge-directed interpolation algorithm which uses the covariance adaptive interpolation algorithm on the edge of the low-resolution image and uses bilinear interpolation algorithm in the low-resolution image smooth area. By using the edge-directed interpolation algorithm, the super-resolution of the image with low resolution is obtained, and we get the percentage of each sub-pixel under a certain type of high-resolution image. Then we rely on the probability value as a soft attribute estimate and carry out sub-pixel scale under the ‘hard classification’. Finally, we get the result of sub-pixel mapping. Through the experiment, we compare the algorithm and the bilinear algorithm given in this paper to the results of the sub-pixel mapping method. It is found that the sub-pixel mapping method based on the edge-directed interpolation algorithm has better edge effect and higher mapping accuracy. The results of the paper meet our original intention of the question. At the same time, the method does not require iterative computation and training of samples, making it easier to implement.

Keywords: remote sensing images, sub-pixel mapping, bilinear interpolation, edge-directed interpolation

Procedia PDF Downloads 189