Mohamed A. El-Gamal

Optimization and Characterization of Electromagnetically Coupled Patch Antennas using RBF Neural Networks

A new neural network model is presented in this paper. It utilizes radial basis functions neural network. The model solves the problem of the electromagnetically coupled microstrip patch antennas. At a specific resonance frequency, the proposed model predicts the optimum geometrical dimensions of both the patch and feeding microstrip line. Moreover, it provides the important characteristics of the optimum design. These characteristics include the impedance bandwidth, gain, and radiation efficiency. The proposed neural network model is very accurate and extremely faster than the classical approach.

Ensembles of Neural Networks for Fault Diagnosis in Analog Circuits

A new neural network-based analog fault diagnosis strategy is introduced. Ensemble of neural networks is constructed and trained for efficient and accurate fault classification of the circuit under test (CUT). In the testing phase, the outputs of the individual ensemble members are combined to isolate the actual CUT fault. Prominent techniques for producing the ensemble are utilized, analyzed and compared. The created ensemble exhibit high classification accuracy even if the CUT has overlapping fault classes which cannot be isolated using a unitary neural network. Each neural classifier of the ensemble focuses on a particular region in the CUT measurement space. As a result, significantly better generalization performance is achieved by the ensemble as compared to any of its individual neural nets. Moreover, the selection of the proper architecture of the neural classifiers is simplified. Experimental results demonstrate the superior performance of the developed approach.

Neural Network vs. Linear Models for Stock Market Sectors Forecasting

The majority of work on forecasting the stock market has focused on individual stocks or stock indexes. In this study we consider the problem of forecasting stock sectors (or industries). We have found no study that considers this problem. Stock sectors are indexes that group several stocks covering a specific sector in the economy, for example the banking sector, the retail sector, etc. It is important for investment allocation purposes to know where each sector is going. In this study we apply linear models, such as Box-Jenkins methodology and multiple regression, as well as neural networks on the sector forecasting problem. As it turns out neural networks yielded the best forecasting performance.

A new approach for the selection of test points for fault diagnosis

A new criterion and an efficient algorithm for the selection of test points in multifrequency fault diagnosis of linear circuits are presented. The proposed criterion exploits the biquadratic nature of the response in terms of circuit parameters instead of the common use of first order sensitivities. Accordingly it is capable of handling catastrophic faults. Employing the proposed criterion, an efficient two-phase fault diagnosis algorithm is introduced. The first phase selects a set of test points and characterizes the response for possible faults. This is done without the simulation of a preselected set of faults. The second phase efficiently isolates on-line actual faults using test points without any computation. A test example is presented to demonstrate the effectiveness of the proposed criterion and algorithm.

Artificial neural network modeling of plasmonic transmission lines

In this paper, new models based on an artificial neural network (ANN) are developed to predict the propagation characteristics of plasmonic nanostrip and coupled nanostrips transmission lines. The trained ANNs are capable of providing the required propagation characteristics with good accuracy and almost instantaneously. The nonlinear mapping performed by the trained ANNs is written as closed-form expressions, which facilitate the direct use of the results obtained in this research. The propagation characteristics of the investigated transmission lines include the effective refractive index and the characteristic impedance. The time needed to simulate 1000 different versions of the transmission line structure is about 48 h, using a full-wave electromagnetic solver compared to 3 s using the developed ANN model.

Estimating the Location of a Metal Strip Using Radial Basis Function Neural Networks

In this paper a neural network model is presented for solving the two-dimensional inverse scattering problem of a metal strip. The proposed model estimates the location of the strip based on the back-scattering field measurements at a few distant points. A radial basis functions neural network is adopted. Besides its remarkably fast response, the proposed model is capable of estimating the location of the strip with very high accuracy.

Analog Fault Diagnosis Using Conic Optimization and Ellipsoidal Classifiers

This paper introduces a new fault diagnosis strategy for analog circuits based on conic optimization and ellipsoidal classifiers. Ellipsoidal classifiers are trained for efficient and accurate fault classification of the circuit under test (CUT). In the testing phase, the output of the ellipsoidal classifiers is used to isolate the actual CUT fault. The constructed classifiers exhibit high classification rate with competitive computational complexity even if the CUT has overlapping faults. Experimental results demonstrate the superior performance of the ellipsoidal classifiers in analog fault diagnosis.

AUTOMATIC CIRCUIT TUNING VIA UNSUPERVISED LEARNING PARADIGMS

This work describes a novel technique for automating the post-fabrication circuit tuning process. A training set that characterizes the behavior of the circuit under test is first constructed. The data in this set consists of input measurement vectors with no output attributes, and is clustered via unsupervised learning algorithm in order to explore its underlying structure and correlations. The generated clusters are labeled and utilized in circuit tuning by calculating the value(s) of the tuning parameter(s). Three prominent and fundamentally different unsupervised learning algorithms, namely, the self-organizing map, the Gaussian mixture model, and the fuzzy C-means algorithm are employed and their performance is compared. The experimental results demonstrate that the proposed technique provides a robust and efficient circuit tuning approach.

Automatic circuit tuning using unsupervised learning procedures

This work describes a novel technique for automating the post-fabrication circuit tuning process. A training set that characterizes the behavior of the circuit under test is first constructed. The data in this set consists of input measurement vectors with no output attributes, and is clustered via unsupervised learning algorithm in order to explore its underlying structure and correlations. The generated clusters are efficiently labeled and directly utilized in circuit tuning by calculating the value(s) of the tuning parameter(s). Three prominent and fundamentally different unsupervised learning algorithms, namely, the self-organizing map, the Gaussian mixture model, and the fuzzy C-means algorithm are tried and their performance is compared. Experimental results demonstrate that the proposed technique provides a robust and efficient circuit tuning approach