This paper presents a comparison of results obtained from neural network training by backpropagation and particle swarm optimization (PSO) algorithms. The neural network model has been developed for field strength pre...
详细信息
ISBN:
(纸本)9789537044107
This paper presents a comparison of results obtained from neural network training by backpropagation and particle swarm optimization (PSO) algorithms. The neural network model has been developed for field strength prediction in indoor environments. It has been already shown for neural networks as powerful tool in RF propagation prediction. It is very important to choose proper algorithm for training a neural network, so we compared BP training algorithms: gradientdescent method and Levenberg- Marquardt algorithm with PSO algorithm. PSO algorithm has been shown as powerful method for global optimization in several applications. A floor of university building in Dubrovnik has been used as case for simulation and measurement of signal strength. The results show that the neural network weights converge faster with PSO than with standard BP algorithms.
Determining fuzzy measure from data is an important topic in some practical applications. Some computing techniques are adopted, such as particle swarm optimization (PSO) and gradient descent algorithm (GD), to identi...
详细信息
Determining fuzzy measure from data is an important topic in some practical applications. Some computing techniques are adopted, such as particle swarm optimization (PSO) and gradient descent algorithm (GD), to identify fuzzy measure. However, there exist some limitations. In this paper, we design a hybrid algorithm called CDPSO, through introducing GD to PSO for the first time. This algorithm has the advantages of GD and PSO, and avoids the disadvantages of them. Theoretical analysis and experimental results verify this, and show that GDPSO is effective and efficient.
This article addresses the problem of blind identification of a non-minimum phase system from only third- and fourth-order cumulants of the output noisy observations of the system. Nonlinear optimization algorithms, n...
详细信息
This article addresses the problem of blind identification of a non-minimum phase system from only third- and fourth-order cumulants of the output noisy observations of the system. Nonlinear optimization algorithms, namely the gradientdescent, the Gauss-Newton and the Newton-Raphson algorithms, are proposed for estimating the parameters of the moving average models. A relationship between third- and fourth-order cumulants of the noisy system output and the parameters of the model is exploited to build a set of non-linear equations that is solved by means of the three non-linear optimization algorithms cited above. Simulation results demonstrate the performance of the proposed algorithms.
As a popular and competent kernel function in Kernel based machine learning techniques, conventional Gaussian kernel has unified kernel width with each of basis functions, which make impliedly a basic assumption: the ...
详细信息
ISBN:
(纸本)9781424421138
As a popular and competent kernel function in Kernel based machine learning techniques, conventional Gaussian kernel has unified kernel width with each of basis functions, which make impliedly a basic assumption: the response signal represents below certain frequency and the noise represents above such certain frequency. However, in many case, this assumption does not hold. To overcome this limitation, a novel adaptive spherical Gaussian kernel is utilized for nonlinear regression, and the stagewise optimization algorithm for maximizing Bayesian evidence in sparse Bayesian learning framework is proposed for model selection. Extensive empirical study shows its effectiveness and flexibility of model on representing regression problem with higher levels of sparsity and higher performance than classical RVM. The attractive ability of this approach is to automatically choose the right kernel widths locally fitting RVs from the training dataset;which could keep right level smoothing at each scale of signal.
In this paper, we propose two kinds of orthogonalization gradient linear discriminant analysis (OGLDA) algorithms to improve the performance of traditional gradient LDA (GLDA) for undersampled problems in face recogni...
详细信息
ISBN:
(纸本)9781424435296
In this paper, we propose two kinds of orthogonalization gradient linear discriminant analysis (OGLDA) algorithms to improve the performance of traditional gradient LDA (GLDA) for undersampled problems in face recognition tasks. In the OGLDA1, the orthogonalization procedure is performed on the discriminant vectors to reduce the redundancy among the discriminant features obtained by GLDA. Thus, all obtained discriminative features can equally contribute to classification performance, which significantly improves the performance of GLDA algorithm for face recognition. In the OGLDA2, the discriminant vectors are resolved one by one in each iterative procedure which overcomes the drawbacks of high computational cost and numerical instability existing in the OGLDA1 algorithm. Since the orthogonalization procedure is applied in the proposed OGLDA methods, the computational stability is improved greatly. The effectiveness of the proposed methods is verified in the experiments on the standard face image databases.
In this paper,we propose two kinds of orthogonalization gradient linear discriminant analysis (OGLDA) algorithms to improve the performance of traditional gradient LDA (GLDA) for undersampled problems in face recognit...
详细信息
In this paper,we propose two kinds of orthogonalization gradient linear discriminant analysis (OGLDA) algorithms to improve the performance of traditional gradient LDA (GLDA) for undersampled problems in face recognition *** the OGLDA1,the orthogonalization procedure is performed on the discriminant vectors to reduce the redundancy among the discriminant features obtained by ***,all obtained discriminative features can equally contribute to classification performance,which significantly improves the performance of GLDA algorithm for face *** the OGLDA2,the discriminant vectors are resolved one by one in each iterative procedure which overcomes the drawbacks of high computational cost and numerical instability existing in the OGLDA1 *** the orthogonalization procedure is applied in the proposed OGLDA methods,the computational stability is improved *** effectiveness of the proposed methods is verified in the experiments on the standard face image databases.
A laboratory demonstration of two novel tactical beam control methods for correcting the effects of strong turbulence including Beacon Anisoplanatism, and the combined effects of Beacon Anisoplanatism and Thermal Bloo...
详细信息
ISBN:
(纸本)9780819468567
A laboratory demonstration of two novel tactical beam control methods for correcting the effects of strong turbulence including Beacon Anisoplanatism, and the combined effects of Beacon Anisoplanatism and Thermal Blooming, respectively, were performed in SAIC's Tactical Beam Control Test-Bed. Both systems were tested with ratio of aperture diameter to Fried parameter, D/r(0), of up to 7, and ratio of beam spot size at the target to isoplanatic angle, theta(B)/theta(O), of up to 10. The first method was implemented in a Wavefront-based Stochastic Parallel gradient Decent (WSPGD) adaptive optics (AO) system, which uses an off-axis wavefront sensor (WFS) to provide feedback for a multi-dithering beam control algorithm. The second method was implemented in a Hybrid WSPGD AO system, which incorporates the WSPGD AO system with a conventional Phase Conjugate (PC) AO, system. The Hybrid system uses an on-axis WFS to generate initial deformable mirror commands and an off-axis WFS to generate additional commands that account for the high frequency phase components removed from the wavefront of a laser return by Beacon Anisoplanatism. We developed a low speed PC-based WSPGD controller, implemented designs of the WSPGD and Hybrid WSPGD AO systems in SAIC's Test-Bed, and tested both AO systems in static and dynamic turbulence over a wide range of turbulence conditions. A target-plane tracker was used to stabilize the line-of-sight in the AO corrected beam. Test results show that the WSPGD AO system efficiently compensates the effects of Beacon Anisoplanatism for both static and dynamic turbulence, providing a mean performance gain of 1.8 averaged over multiple turbulent realizations. We also found in testing that the Hybrid WSPGD system efficiently compensates for Beacon Anisoplanatism in the presence of Thermal Blooming - providing improved compensation for both Thermal Blooming and turbulence. In the presence of strong Beacon Anisoplanatism with theta B/theta(O) of up to 10, t
For a non-cooperative target, a laser beacon is created by illuminating the target with a beacon beam. When a beacon beam propagates though deep turbulence, turbulence spreads the beam. A conventional phase conjugate ...
详细信息
ISBN:
(纸本)9780819468598
For a non-cooperative target, a laser beacon is created by illuminating the target with a beacon beam. When a beacon beam propagates though deep turbulence, turbulence spreads the beam. A conventional phase conjugate adaptive optics (AO) system is not efficient in the presence of Beacon Anisoplanatism when the beacon beam spot size at the target includes many isoplanatic patch sizes. We introduce a concept of the wavefront-based stochastic parallel gradient decent (WSPGD) AO system, which uses an off-axis wavefront sensor to provide feedback for the beam control algorithm. This concept is based on the finding that the phase aberrations of laser return from the target contain information about beam spot size at the target, and that correction of a limited number of low-order Zernike modes increases on-axis intensity and power in the bucket at the target. We evaluated the WSPGD AO system performance in simulation for two tactical engagement scenarios in the presence of strong turbulence. We found that that the WSPGD AO system can efficiently compensate the effects of strong turbulence including Beacon Anisoplanatism, even when the beam spot size at the target includes up to 20 isoplanatic patch sizes and the isoplanatic angle is by a factor of 2.6 less than the diffraction limit. The Strehl ratio gain for this scenario is 1.6 - 2.5, and the maximum Strehl ratio is achieved after 15-20 iterations. A laboratory demonstration performed under a separate program confirmed our theoretical predictions.
An effective approach is developed to establish affine Takagi-Sugeno (T-S) fuzzy model for a given nonlinear system from its input-output data. Firstly, the fuzzy c-regression model (FCRM) clustering technique is appl...
详细信息
An effective approach is developed to establish affine Takagi-Sugeno (T-S) fuzzy model for a given nonlinear system from its input-output data. Firstly, the fuzzy c-regression model (FCRM) clustering technique is applied to partition the product space of the given input-output data into hyper-plan-shaped clusters. Each cluster is essentially a basis of the fuzzy rule that describes the system behaviour, and the number of clusters is just the number of fuzzy rules. Particularly, a novel cluster validity criterion for FCRM is set up to choose the appropriate number of clusters (rules). Once the number of clusters is determined, the consequent parameters of each IF-THEN rule are directly obtained from the functional cluster representatives (affine linear functions). The antecedent fuzzy sets of each IF-THEN fuzzy rule are acquired by projecting the fuzzy partitions matrix U onto the axes of individual antecedent variable to obtain point-wise defined fuzzy sets and to approximate these point-wise defined fuzzy sets by normal bell-shaped membership functions. Additionally, a check and repartition algorithm is suggested to prevent the inappropriate premise structure where separate regions of data shared the same regression model. Finally, the gradient descent algorithm is included to adjust the fuzzy model precisely. An affine T-S fuzzy model with compact IF-THEN rules could thus be generated systematically. Several simulation examples are provided to demonstrate the accuracy and effectiveness of the affine T-S fuzzy modelling algorithm.
A concept of a Hybrid Wavefront-based Stochastic Parallel gradient Decent (WSPGD) Adaptive Optics (AO) system for correcting the combined effects of Beacon Anisoplanatism and Thermal Blooming is introduced. This syste...
详细信息
ISBN:
(纸本)9780819468598
A concept of a Hybrid Wavefront-based Stochastic Parallel gradient Decent (WSPGD) Adaptive Optics (AO) system for correcting the combined effects of Beacon Anisoplanatism and Thermal Blooming is introduced. This system integrates a conventional phase conjugate (PC) AO system with a WSPGD AO system. It uses on-axis wavefront measurements of a laser return from an extended beacon to generate initial deformable mirror (DM) commands. Since high frequency phase components are removed from the wavefront of a laser return by a low-pass filter effect of an extended beacon, the system also uses off-axis wavefront measurements to provide feedback for a multi-dithering beam control algorithm in order to generate additional DM commands that account for those missing high frequency phase components. Performance of the Hybrid WSPGD AO system was evaluated in simulation using a wave optics code. Numerical analysis was performed for two tactical scenarios that included ranges of L = 2 km and L = 20 km, ratio of aperture diameter to Fried parameter, D/r(0), of up to 15, ratio of beam spot size at the target to isoplanatic angle, theta(B)/theta(0), of UP to 40, and general distortion number characterizing the strength of Thermal Blooming, N-d = 50, 75, and 100. A line-of-sight in the corrected beam was stabilized using a target-plane tracker. The simulation results reveal that the Hybrid WSPGD AO system can efficiently correct the effects of Beacon Anisoplanatism and Thermal Blooming, providing improved compensation of Thermal Blooming in the presence of strong turbulence. Simulation results also indicate that the Hybrid WSPGD AO system outperforms a conventional PC AO system, increasing the Strehl ratio by up to 300% in less than 50 iterations. A follow-on laboratory demonstration performed under a separate program confirmed our theoretical predictions.
暂无评论