Search results for: successive approximation register.

Authors: Yongxin Yuan

Abstract:

In this paper, we first give the representation of the general solution of the following inverse eigenvalue problem (IEP): Given X ∈ Rn×p and a diagonal matrix Λ ∈ Rp×p, find nontrivial real-valued symmetric arrow-head matrices A and B such that AXΛ = BX. We then consider an optimal approximation problem: Given real-valued symmetric arrow-head matrices A, ˜ B˜ ∈ Rn×n, find (A, ˆ Bˆ) ∈ SE such that Aˆ − A˜2 + Bˆ − B˜2 = min(A,B)∈SE (A−A˜2 +B −B˜2), where SE is the solution set of IEP. We show that the optimal approximation solution (A, ˆ Bˆ) is unique and derive an explicit formula for it.

Keywords: Partially prescribed spectral information, symmetric arrow-head matrix, inverse problem, optimal approximation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1745

Authors: P. T. Ahamed Seyd, V. I. Thajudin Ahamed, Jeevamma Jacob, Paul Joseph K

Abstract:

Diabetes mellitus (DM) is frequently characterized by autonomic nervous dysfunction. Analysis of heart rate variability (HRV) has become a popular noninvasive tool for assessing the activities of autonomic nervous system (ANS). In this paper, changes in ANS activity are quantified by means of frequency and time domain analysis of R-R interval variability. Electrocardiograms (ECG) of 16 patients suffering from DM and of 16 healthy volunteers were recorded. Frequency domain analysis of extracted normal to normal interval (NN interval) data indicates significant difference in very low frequency (VLF) power, low frequency (LF) power and high frequency (HF) power, between the DM patients and control group. Time domain measures, standard deviation of NN interval (SDNN), root mean square of successive NN interval differences (RMSSD), successive NN intervals differing more than 50 ms (NN50 Count), percentage value of NN50 count (pNN50), HRV triangular index and triangular interpolation of NN intervals (TINN) also show significant difference between the DM patients and control group.

Keywords: Autonomic nervous system, diabetes mellitus, frequency domain and time domain analysis, heart rate variability.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 3044

Authors: M-J. Huang, C-J. Huang, L-C. Chen

Abstract:

In a previously developed fast vortex method, the diffusion of the vortex sheet induced at the solid wall by the no-slip boundary conditions was modeled according to the approximation solution of Koumoutsakos and converted into discrete blobs in the vicinity of the wall. This scheme had been successfully applied to a simulation of the flow induced with an impulsively initiated circular cylinder. In this work, further modifications on this vortex method are attempted, including replacing the approximation solution by the boundary-element-method solution, incorporating a new algorithm for handling the over-weak vortex blobs, and diffusing the vortex sheet circulation in a new way suitable for high-curvature solid bodies. The accuracy is thus largely improved. The predictions of lift and drag coefficients for a uniform flow past a NASA airfoil agree well with the existing literature.

Keywords: Resurrected core-spreading vortex method, Boundaryelement method, Vortex sheet, Over-weak vortex blobs.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1367

Authors: Li Long, Thomas Ortlepp

Abstract:

The transport properties of carriers in polycrystalline silicon film affect the performance of polycrystalline silicon-based devices. They depend strongly on the grain structure, grain boundary trap properties and doping concentration, which in turn are determined by the film deposition and processing conditions. Based on the properties of charge carriers, phonons, grain boundaries and their interactions, the thermoelectric properties of polycrystalline silicon are analyzed with the relaxation time approximation of the Boltzmann transport equation. With this approach, thermal conductivity, electrical conductivity and Seebeck coefficient as a function of grain size, trap properties and doping concentration can be determined. Experiment on heavily doped polycrystalline silicon is carried out and measurement results are compared with the model.

Keywords: Conductivity, polycrystalline silicon, relaxation time approximation, Seebeck coefficient, thermoelectric property.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 158

Authors: Edlira Donefski, Tina Donefski, Lorenc Ekonomi

Abstract:

Edgeworth Approximation, Bootstrap and Monte Carlo Simulations have a considerable impact on the achieving certain results related to different problems taken into study. In our paper, we have treated a financial case related to the effect that have the components of a Cash-Flow of one of the most successful businesses in the world, as the financial activity, operational activity and investing activity to the cash and cash equivalents at the end of the three-months period. To have a better view of this case we have created a Vector Autoregression model, and after that we have generated the impulse responses in the terms of Asymptotic Analysis (Edgeworth Approximation), Monte Carlo Simulations and Residual Bootstrap based on the standard errors of every series created. The generated results consisted of the common tendencies for the three methods applied, that consequently verified the advantage of the three methods in the optimization of the model that contains many variants.

Keywords: Autoregression, Bootstrap, Edgeworth Expansion, Monte Carlo Method.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 527

Authors: Anam Rafique, Syed Sohail Ahmed

Abstract:

One of the essential components of much of DSP application is noise cancellation. Changes in real time signals are quite rapid and swift. In noise cancellation, a reference signal which is an approximation of noise signal (that corrupts the original information signal) is obtained and then subtracted from the noise bearing signal to obtain a noise free signal. This approximation of noise signal is obtained through adaptive filters which are self adjusting. As the changes in real time signals are abrupt, this needs adaptive algorithm that converges fast and is stable. Least mean square (LMS) and normalized LMS (NLMS) are two widely used algorithms because of their plainness in calculations and implementation. But their convergence rates are small. Adaptive averaging filters (AFA) are also used because they have high convergence, but they are less stable. This paper provides the comparative study of LMS and Normalized NLMS, AFA and new enhanced average adaptive (Average NLMS-ANLMS) filters for noise cancelling application using speech signals.

Keywords: AFA, ANLMS, LMS, NLMS.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1892

Authors: V. T. S. Dao, T. G. Etoh, C. Vo Le, H. D. Nguyen, K. Takehara, T. Akino, K. Nishi

Abstract:

We present an explicit expression to estimate driving voltage attenuation through RC networks representation of an ultrahigh- speed image sensor. Elmore delay metric for a fundamental RC chain is employed as the first-order approximation. By application of dimensional analysis to SPICE simulation data, we found a simple expression that significantly improves the accuracy of the approximation. Estimation error of the resultant expression for uniform RC networks is less than 2%. Similarly, another simple closed-form model to estimate 50 % delay through fundamental RC networks is also derived with sufficient accuracy. The framework of this analysis can be extended to address delay or attenuation issues of other VLSI structures.

Keywords: Dimensional Analysis, Elmore model, RC network, Signal Attenuation, Ultra-High-Speed Image Sensor.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1379

Authors: Y. Z. Wu, Z. Dong, S. K. You

Abstract:

Global approximation using metamodel for complex mathematical function or computer model over a large variable domain is often needed in sensibility analysis, computer simulation, optimal control, and global design optimization of complex, multiphysics systems. To overcome the limitations of the existing response surface (RS), surrogate or metamodel modeling methods for complex models over large variable domain, a new adaptive and regressive RS modeling method using quadratic functions and local area model improvement schemes is introduced. The method applies an iterative and Latin hypercube sampling based RS update process, divides the entire domain of design variables into multiple cells, identifies rougher cells with large modeling error, and further divides these cells along the roughest dimension direction. A small number of additional sampling points from the original, expensive model are added over the small and isolated rough cells to improve the RS model locally until the model accuracy criteria are satisfied. The method then combines local RS cells to regenerate the global RS model with satisfactory accuracy. An effective RS cells sorting algorithm is also introduced to improve the efficiency of model evaluation. Benchmark tests are presented and use of the new metamodeling method to replace complex hybrid electrical vehicle powertrain performance model in vehicle design optimization and optimal control are discussed.

Keywords: Global approximation, polynomial response surface, domain decomposition, domain combination, multiphysics modeling, hybrid powertrain optimization

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1867

Authors: Wanhyun Cho, Soonja Kang, Sangkyoon Kim, Soonyoung Park

Abstract:

We present probabilistic multinomial Dirichlet classification model for multidimensional data and Gaussian process priors. Here, we have considered efficient computational method that can be used to obtain the approximate posteriors for latent variables and parameters needed to define the multiclass Gaussian process classification model. We first investigated the process of inducing a posterior distribution for various parameters and latent function by using the variational Bayesian approximations and important sampling method, and next we derived a predictive distribution of latent function needed to classify new samples. The proposed model is applied to classify the synthetic multivariate dataset in order to verify the performance of our model. Experiment result shows that our model is more accurate than the other approximation methods.

Keywords: Multinomial dirichlet classification model, Gaussian process priors, variational Bayesian approximation, Importance sampling, approximate posterior distribution, Marginal likelihood evidence.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1566

Authors: Abdullah Y. Al-Hossain

Abstract:

This paper considers inference under progressive type II censoring with a compound Rayleigh failure time distribution. The maximum likelihood (ML), and Bayes methods are used for estimating the unknown parameters as well as some lifetime parameters, namely reliability and hazard functions. We obtained Bayes estimators using the conjugate priors for two shape and scale parameters. When the two parameters are unknown, the closed-form expressions of the Bayes estimators cannot be obtained. We use Lindley.s approximation to compute the Bayes estimates. Another Bayes estimator has been obtained based on continuous-discrete joint prior for the unknown parameters. An example with the real data is discussed to illustrate the proposed method. Finally, we made comparisons between these estimators and the maximum likelihood estimators using a Monte Carlo simulation study.

Keywords: Progressive type II censoring, compound Rayleigh failure time distribution, maximum likelihood estimation, Bayes estimation, Lindley's approximation method, Monte Carlo simulation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2347

Authors: Jianguo Tang, Kun She, William Zhu

Abstract:

Covering-based rough sets is an extension of rough sets and it is based on a covering instead of a partition of the universe. Therefore it is more powerful in describing some practical problems than rough sets. However, by extending the rough sets, covering-based rough sets can increase the roughness of each model in recognizing objects. How to obtain better approximations from the models of a covering-based rough sets is an important issue. In this paper, two concepts, determinate elements and indeterminate elements in a universe, are proposed and given precise definitions respectively. This research makes a reasonable refinement of the covering-element from a new viewpoint. And the refinement may generate better approximations of covering-based rough sets models. To prove the theory above, it is applied to eight major coveringbased rough sets models which are adapted from other literature. The result is, in all these models, the lower approximation increases effectively. Correspondingly, in all models, the upper approximation decreases with exceptions of two models in some special situations. Therefore, the roughness of recognizing objects is reduced. This research provides a new approach to the study and application of covering-based rough sets.

Keywords: Determinate element, indeterminate element, refinementof covering-element, refinement of covering, covering-basedrough sets.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1280

Authors: Gabriela R. Fernandes, Renato F. Denadai, Guido J. Denipotti

Abstract:

In this work, are discussed two formulations of the boundary element method - BEM to perform linear bending analysis of plates reinforced by beams. Both formulations are based on the Kirchhoff's hypothesis and they are obtained from the reciprocity theorem applied to zoned plates, where each sub-region defines a beam or a slab. In the first model the problem values are defined along the interfaces and the external boundary. Then, in order to reduce the number of degrees of freedom kinematics hypothesis are assumed along the beam cross section, leading to a second formulation where the collocation points are defined along the beam skeleton, instead of being placed on interfaces. On these formulations no approximation of the generalized forces along the interface is required. Moreover, compatibility and equilibrium conditions along the interface are automatically imposed by the integral equation. Thus, these formulations require less approximation and the total number of the degree s of freedom is reduced. In the numerical examples are discussed the differences between these two BEM formulations, comparing as well the results to a well-known finite element code.

Keywords: Boundary elements, Building floor structures, Platebending.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1624

Authors: Min Kyung An

Abstract:

In this paper, we study the data collection problem in Wireless Sensor Networks (WSNs) adopting the two interference models: The graph model and the more realistic physical interference model known as Signal-to-Interference-Noise-Ratio (SINR). The main issue of the problem is to compute schedules with the minimum number of timeslots, that is, to compute the minimum latency schedules, such that data from every node can be collected without any collision or interference to a sink node. While existing works studied the problem with unit-sized and unbounded-sized message models, we investigate the problem with the bounded-sized message model, and introduce a constant factor approximation algorithm. To the best known of our knowledge, our result is the first result of the data collection problem with bounded-sized model in both interference models.

Keywords: Data collection, collision-free, interference-free, physical interference model, SINR, approximation, bounded-sized message model, wireless sensor networks, WSN.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1188

Authors: J. Zamastil, V. Patkos

Abstract:

Recent progress in calculation of the one-loop selfenergy of the electron bound in the Coulomb field is summarized. The relativistic multipole expansion is introduced. This expansion is based on a single assumption: except for the part of the time component of the electron four-momentum corresponding to the electron rest mass, the exchange of four-momentum between the virtual electron and photon can be treated perturbatively. For non Sstates and normalized difference n3
En −
E1 of the S-states this itself yields very accurate results after taking the method to the third order. For the ground state the perturbation treatment of the electron virtual states with very high three-momentum is to be avoided. For these states one can always rearrange the pertinent expression in such a way that free-particle approximation is allowed. Combination of the relativistic multipole expansion and free-particle approximation yields very accurate result after taking the method to the ninth order. These results are in very good agreement with the previous results obtained by the partial wave expansion and definitely exclude the possibility that the uncertainity in determination of the proton radius comes from the uncertainity in the calculation of the one-loop selfenergy.

Keywords: Hydrogen-like atoms, self-energy.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1644

Authors: Salah Alrabeei, Mohammad Yousuf

Abstract:

The goal of option pricing theory is to help the investors to manage their money, enhance returns and control their financial future by theoretically valuing their options. However, most of the option pricing models have no analytical solution. Furthermore, not all the numerical methods are efficient to solve these models because they have nonsmoothing payoffs or discontinuous derivatives at the exercise price. In this paper, we solve the American option under jump diffusion models by using efficient time-dependent numerical methods. several techniques are integrated to reduced the overcome the computational complexity. Fast Fourier Transform (FFT) algorithm is used as a matrix-vector multiplication solver, which reduces the complexity from O(M2) into O(M logM). Partial fraction decomposition technique is applied to rational approximation schemes to overcome the complexity of inverting polynomial of matrices. The proposed method is easy to implement on serial or parallel versions. Numerical results are presented to prove the accuracy and efficiency of the proposed method.

Keywords: Integral differential equations, American options, jump–diffusion model, rational approximation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 497

Authors: Julien Romieux, Fabio Verdicchio

Abstract:

Wireless Sensor Networks (WSNs), which sense environmental data with battery-powered nodes, require multi-hop communication. This power-demanding task adds an extra workload that is unfairly distributed across the network. As a result, nodes run out of battery at different times: this requires an impractical individual node maintenance scheme. Therefore we investigate a new Cooperative Sensing approach that extends the WSN operational life and allows a more practical network maintenance scheme (where all nodes deplete their batteries almost at the same time). We propose a novel cooperative algorithm that derives a piecewise representation of the sensed signal while controlling approximation accuracy. Simulations show that our algorithm increases WSN operational life and spreads communication workload evenly. Results convey a counterintuitive conclusion: distributing workload fairly amongst nodes may not decrease the network power consumption and yet extend the WSN operational life. This is achieved as our cooperative approach decreases the workload of the most burdened cluster in the network.

Keywords: Cooperative signal processing, power management, signal representation, signal approximation, wireless sensor networks.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1736

Authors: N. H. Adenan, M. S. M. Noorani

Abstract:

River flow prediction is an essential to ensure proper management of water resources can be optimally distribute water to consumers. This study presents an analysis and prediction by using nonlinear prediction method involving monthly river flow data in Tanjung Tualang from 1976 to 2006. Nonlinear prediction method involves the reconstruction of phase space and local linear approximation approach. The phase space reconstruction involves the reconstruction of one-dimensional (the observed 287 months of data) in a multidimensional phase space to reveal the dynamics of the system. Revenue of phase space reconstruction is used to predict the next 72 months. A comparison of prediction performance based on correlation coefficient (CC) and root mean square error (RMSE) have been employed to compare prediction performance for nonlinear prediction method, ARIMA and SVM. Prediction performance comparisons show the prediction results using nonlinear prediction method is better than ARIMA and SVM. Therefore, the result of this study could be used to develop an efficient water management system to optimize the allocation water resources.

Keywords: River flow, nonlinear prediction method, phase space, local linear approximation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2307

Authors: Petter Eklund, Jonathan Sjölund, Sandra Eriksson, Mats Leijon

Abstract:

The spoke type rotor can be used to obtain magnetic flux concentration in permanent magnet machines. This allows the air gap magnetic flux density to exceed the remanent flux density of the permanent magnets but gives problems with leakage fluxes in the magnetic circuit. The end leakage flux of one spoke type permanent magnet rotor design is studied through measurements and finite element simulations. The measurements are performed in the end regions of a 12 kW prototype generator for a vertical axis wind turbine. The simulations are made using three dimensional finite elements to calculate the magnetic field distribution in the end regions of the machine. Also two dimensional finite element simulations are performed and the impact of the two dimensional approximation is studied. It is found that the magnetic leakage flux in the end regions of the machine is equal to about 20% of the flux in the permanent magnets. The overestimation of the performance by the two dimensional approximation is quantified and a curve-fitted expression for its behavior is suggested.

Keywords: End effects, end leakage flux, permanent magnet machine, spoke type rotor.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1027

Authors: Kittisak Kerdprasop, Nittaya Kerdprasop, Pairote Sattayatham

Abstract:

Data stream analysis is the process of computing various summaries and derived values from large amounts of data which are continuously generated at a rapid rate. The nature of a stream does not allow a revisit on each data element. Furthermore, data processing must be fast to produce timely analysis results. These requirements impose constraints on the design of the algorithms to balance correctness against timely responses. Several techniques have been proposed over the past few years to address these challenges. These techniques can be categorized as either dataoriented or task-oriented. The data-oriented approach analyzes a subset of data or a smaller transformed representation, whereas taskoriented scheme solves the problem directly via approximation techniques. We propose a hybrid approach to tackle the data stream analysis problem. The data stream has been both statistically transformed to a smaller size and computationally approximated its characteristics. We adopt a Monte Carlo method in the approximation step. The data reduction has been performed horizontally and vertically through our EMR sampling method. The proposed method is analyzed by a series of experiments. We apply our algorithm on clustering and classification tasks to evaluate the utility of our approach.

Keywords: Data Stream, Monte Carlo, Sampling, DensityEstimation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1382

Authors: N. H. Adenan, M. S. M. Noorani

Abstract:

River flow prediction is an essential tool to ensure proper management of water resources and the optimal distribution of water to consumers. This study presents an analysis and prediction by using nonlinear prediction method with monthly river flow data for Tanjung Tualang from 1976 to 2006. Nonlinear prediction method involves the reconstruction of phase space and local linear approximation approach. The reconstruction of phase space involves the reconstruction of one-dimension (the observed 287 months of data) in a multidimensional phase space to reveal the dynamics of the system. The revenue of phase space reconstruction is used to predict the next 72 months. A comparison of prediction performance based on correlation coefficient (CC) and root mean square error (RMSE) was employed to compare prediction performance for the nonlinear prediction method, ARIMA and SVM. Prediction performance comparisons show that the prediction results using the nonlinear prediction method are better than ARIMA and SVM. Therefore, the results of this study could be used to develop an efficient water management system to optimize the allocation of water resources.

Keywords: River flow, nonlinear prediction method, phase space, local linear approximation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1916

Authors: Jiann-Ming Wu, Hsiao-Chang Chen, Chun-Chang Wu, Pei-Hsun Hsu

Abstract:

This work explores blind image deconvolution by recursive function approximation based on supervised learning of neural networks, under the assumption that a degraded image is linear convolution of an original source image through a linear shift-invariant (LSI) blurring matrix. Supervised learning of neural networks of radial basis functions (RBF) is employed to construct an embedded recursive function within a blurring image, try to extract non-deterministic component of an original source image, and use them to estimate hyper parameters of a linear image degradation model. Based on the estimated blurring matrix, reconstruction of an original source image from a blurred image is further resolved by an annealed Hopfield neural network. By numerical simulations, the proposed novel method is shown effective for faithful estimation of an unknown blurring matrix and restoration of an original source image.

Keywords: Blind image deconvolution, linear shift-invariant(LSI), linear image degradation model, radial basis functions (rbf), recursive function, annealed Hopfield neural networks.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2006

Authors: Arturo Aquino, Manuel Emilio Gegundez, Diego Marin

Abstract:

In this paper, a new automated methodology to detect the optic disc (OD) automatically in retinal images from patients with risk of being affected by Diabetic Retinopathy (DR) and Macular Edema (ME) is presented. The detection procedure comprises two independent methodologies. On one hand, a location methodology obtains a pixel that belongs to the OD using image contrast analysis and structure filtering techniques and, on the other hand, a boundary segmentation methodology estimates a circular approximation of the OD boundary by applying mathematical morphology, edge detection techniques and the Circular Hough Transform. The methodologies were tested on a set of 1200 images composed of 229 retinographies from patients affected by DR with risk of ME, 431 with DR and no risk of ME and 540 images of healthy retinas. The location methodology obtained 98.83% success rate, whereas the OD boundary segmentation methodology obtained good circular OD boundary approximation in 94.58% of cases. The average computational time measured over the total set was 1.67 seconds for OD location and 5.78 seconds for OD boundary segmentation.

Keywords: Diabetic retinopathy, macular edema, optic disc, automated detection, automated segmentation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2744

Authors: Li Long, Thomas Ortlepp

Abstract:

A theoretical model for the optimization of thermopile sensor performance is developed for thermoelectric-based infrared radiation detection. It is shown that the performance of polycrystalline silicon film thermopile sensor can be optimized according to the thermoelectric quality factor, sensor layer structure factor and sensor layout shape factor. Based on the properties of electrons, phonons, grain boundaries and their interactions, the thermoelectric quality factor of polycrystalline silicon is analyzed with the relaxation time approximation of Boltzmann transport equation. The model includes the effects of grain structure, grain boundary trap properties and doping concentration. The layer structure factor of sensor is analyzed with respect to infrared absorption coefficient. The effect of layout design is characterized with the shape factor, which is calculated for different sensor designs. Double layer polycrystalline silicon thermopile infrared sensors on suspended support membrane have been designed and fabricated with a CMOS-compatible process. The theoretical approach is confirmed with measurement results.

Keywords: Polycrystalline silicon film, relaxation time approximation, specific detectivity, thermal conductivity, thermopile infrared sensor.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 153

Authors: Apoorva Vinod, Archana Mathur, Snehanshu Saha

Abstract:

Though there exists a plethora of Activation Functions (AFs) used in single and multiple hidden layer Neural Networks (NN), their behavior always raised curiosity, whether used in combination or singly. The popular AFs – Sigmoid, ReLU, and Tanh – have performed prominently well for shallow and deep architectures. Most of the time, AFs are used singly in multi-layered NN, and, to the best of our knowledge, their performance is never studied and analyzed deeply when used in combination. In this manuscript, we experiment on multi-layered NN architecture (both on shallow and deep architectures; Convolutional NN and VGG16) and investigate how well the network responds to using two different AFs (Sigmoid-Tanh, Tanh-ReLU, ReLU-Sigmoid) used alternately against a traditional, single (Sigmoid-Sigmoid, Tanh-Tanh, ReLU-ReLU) combination. Our results show that on using two different AFs, the network achieves better accuracy, substantially lower loss, and faster convergence on 4 computer vision (CV) and 15 Non-CV (NCV) datasets. When using different AFs, not only was the accuracy greater by 6-7%, but we also accomplished convergence twice as fast. We present a case study to investigate the probability of networks suffering vanishing and exploding gradients when using two different AFs. Additionally, we theoretically showed that a composition of two or more AFs satisfies Universal Approximation Theorem (UAT).

Keywords: Activation Function, Universal Approximation function, Neural Networks, convergence.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 91

Authors: Wei Zhang, Yao-Yu Li, Yi-Fan Zhu, Qun Li, Wei-Ping Wang

Abstract:

It-s known that incorporating prior knowledge into support vector regression (SVR) can help to improve the approximation performance. Most of researches are concerned with the incorporation of knowledge in the form of numerical relationships. Little work, however, has been done to incorporate the prior knowledge on the structural relationships among the variables (referred as to Structural Prior Knowledge, SPK). This paper explores the incorporation of SPK in SVR by constructing appropriate admissible support vector kernel (SV kernel) based on the properties of reproducing kernel (R.K). Three-levels specifications of SPK are studied with the corresponding sub-levels of prior knowledge that can be considered for the method. These include Hierarchical SPK (HSPK), Interactional SPK (ISPK) consisting of independence, global and local interaction, Functional SPK (FSPK) composed of exterior-FSPK and interior-FSPK. A convenient tool for describing the SPK, namely Description Matrix of SPK is introduced. Subsequently, a new SVR, namely Motivated Support Vector Regression (MSVR) whose structure is motivated in part by SPK, is proposed. Synthetic examples show that it is possible to incorporate a wide variety of SPK and helpful to improve the approximation performance in complex cases. The benefits of MSVR are finally shown on a real-life military application, Air-toground battle simulation, which shows great potential for MSVR to the complex military applications.

Keywords: admissible support vector kernel, reproducing kernel, structural prior knowledge, motivated support vector regression

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1579

Authors: Wei Zhang, Yao-Yu Li, Yi-Fan Zhu, Qun Li, Wei-Ping Wang

Abstract:

It-s known that incorporating prior knowledge into support vector regression (SVR) can help to improve the approximation performance. Most of researches are concerned with the incorporation of knowledge in form of numerical relationships. Little work, however, has been done to incorporate the prior knowledge on the structural relationships among the variables (referred as to Structural Prior Knowledge, SPK). This paper explores the incorporation of SPK in SVR by constructing appropriate admissible support vector kernel (SV kernel) based on the properties of reproducing kernel (R.K). Three-levels specifications of SPK are studies with the corresponding sub-levels of prior knowledge that can be considered for the method. These include Hierarchical SPK (HSPK), Interactional SPK (ISPK) consisting of independence, global and local interaction, Functional SPK (FSPK) composed of exterior-FSPK and interior-FSPK. A convenient tool for describing the SPK, namely Description Matrix of SPK is introduced. Subsequently, a new SVR, namely Motivated Support Vector Regression (MSVR) whose structure is motivated in part by SPK, is proposed. Synthetic examples show that it is possible to incorporate a wide variety of SPK and helpful to improve the approximation performance in complex cases. The benefits of MSVR are finally shown on a real-life military application, Air-toground battle simulation, which shows great potential for MSVR to the complex military applications.

Keywords: admissible support vector kernel, reproducing kernel, structural prior knowledge, motivated support vector regression

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1358

Authors: J. P. Dubois, H. M. Chiu

Abstract:

In this paper, we study statistical multiplexing of VBR video in ATM networks. ATM promises to provide high speed realtime multi-point to central video transmission for telemedicine applications in rural hospitals and in emergency medical services. Video coders are known to produce variable bit rate (VBR) signals and the effects of aggregating these VBR signals need to be determined in order to design a telemedicine network infrastructure capable of carrying these signals. We first model the VBR video signal and simulate it using a generic continuous-data autoregressive (AR) scheme. We carry out the queueing analysis by the Fluid Approximation Model (FAM) and the Markov Modulated Poisson Process (MMPP). The study has shown a trade off: multiplexing VBR signals reduces burstiness and improves resource utilization, however, the buffer size needs to be increased with an associated economic cost. We also show that the MMPP model and the Fluid Approximation model fit best, respectively, the cell region and the burst region. Therefore, a hybrid MMPP and FAM completely characterizes the overall performance of the ATM statistical multiplexer. The ramifications of this technology are clear: speed, reliability (lower loss rate and jitter), and increased capacity in video transmission for telemedicine. With migration to full IP-based networks still a long way to achieving both high speed and high quality of service, the proposed ATM architecture will remain of significant use for telemedicine.

Keywords: ATM, multiplexing, queueing, telemedicine, VBR.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1703

Authors: Adrian Perreau de Pinninck, David Dupplaw, Spyros Kotoulas, Ronny Siebes

Abstract:

Web services are pieces of software that can be invoked via a standardized protocol. They can be combined via formalized taskflow languages. The Open Knowledge system is a fully distributed system using P2P technology, that allows users to publish the setaskflows, and programmers to register their web services or publish implementations of them, for the roles described in these workflows.Besides this, the system offers the functionality to select a peer that could coordinate such an interaction model and inform web services when it is their 'turn'. In this paper we describe the architecture and implementation of the Open Knowledge Kernel which provides the core functionality of the Open Knowledge system.

Keywords: Architecture, P2P, Web Services, Semantic Web

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1357

Authors: Min Kyung An

Abstract:

In Wireless Sensor Networks which consist of tiny wireless sensor nodes with limited battery power, one of the most fundamental applications is data aggregation which collects nearby environmental conditions and aggregates the data to a designated destination, called a sink node. Important issues concerning the data aggregation are time efficiency and energy consumption due to its limited energy, and therefore, the related problem, named Minimum Latency Aggregation Scheduling (MLAS), has been the focus of many researchers. Its objective is to compute the minimum latency schedule, that is, to compute a schedule with the minimum number of timeslots, such that the sink node can receive the aggregated data from all the other nodes without any collision or interference. For the problem, the two interference models, the graph model and the more realistic physical interference model known as Signal-to-Interference-Noise-Ratio (SINR), have been adopted with different power models, uniform-power and non-uniform power (with power control or without power control), and different antenna models, omni-directional antenna and directional antenna models. In this survey article, as the problem has proven to be NP-hard, we present and compare several state-of-the-art approximation algorithms in various models on the basis of latency as its performance measure.

Keywords: Data aggregation, convergecast, gathering, approximation, interference, omni-directional, directional.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 761

Authors: Rubén Bãnos, José Arcos, Oscar Bautista, Federico Méndez

Abstract:

The Oscillatory electroosmotic flow (OEOF) in power law fluids through a microchannel is studied numerically. A time-dependent external electric field (AC) is suddenly imposed at the ends of the microchannel which induces the fluid motion. The continuity and momentum equations in the x and y direction for the flow field were simplified in the limit of the lubrication approximation theory (LAT), and then solved using a numerical scheme. The solution of the electric potential is based on the Debye-H¨uckel approximation which suggest that the surface potential is small,say, smaller than 0.025V and for a symmetric (z : z) electrolyte. Our results suggest that the velocity profiles across the channel-width are controlled by the following dimensionless parameters: the angular Reynolds number, Reω, the electrokinetic parameter, ¯κ, defined as the ratio of the characteristic length scale to the Debye length, the parameter λ which represents the ratio of the Helmholtz-Smoluchowski velocity to the characteristic length scale and the flow behavior index, n. Also, the results reveal that the velocity profiles become more and more non-uniform across the channel-width as the Reω and ¯κ are increased, so oscillatory OEOF can be really useful in micro-fluidic devices such as micro-mixers.

Keywords: Oscillatory electroosmotic flow, Non-Newtonian fluids, power-law model, low zeta potentials.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 818