Cascade Feed-forward Neural Network Research Articles

Damage detection of plate-like structures using intelligent surrogate model

Cracks in plate-like structures are some of the main reasons for destruction of the entire structure. In this study, a novel two-stage methodology is proposed for damage detection of flexural plates using an optimized artificial neural network. In the first stage, location of damages in plates is investigated using curvature-moment and curvature-moment derivative concepts. After detecting the damaged areas, the equations for damage severity detection are solved via Bat Algorithm (BA). In the second stage, in order to efficiently reduce the computational cost of model updating during the optimization process of damage severity detection, multiple damage location assurance criterion index based on the frequency change vector of structures are evaluated using properly trained cascade feed-forward neural network (CFNN) as a surrogate model. In order to achieve the most generalized neural network as a surrogate model, its structure is optimized using binary version of BA. To validate this proposed solution method, two examples are presented. The results indicate that after determining the damage location based on curvature-moment derivative concept, the proposed solution method for damage severity detection leads to significant reduction of computational time compared with direct finite element method. Furthermore, integrating BA with the efficient approximation mechanism of finite element model, maintains the acceptable accuracy of damage severity detection.

Strength optimisation of mortar with CNTs and nanoclays

This study explores the use of carbon nanotubes (CNTs) and nanoclays in cement mortars. The paper presents modelling and optimisation of the compressive and flexural strengths of cement mortars containing CNTs and the compressive, tensile and flexural strengths of cement mortars including nanoclays. The CNT ratio, CNT aspect ratio and functionalisation effects are optimised, as are the montmorillonite nanoclay ratio and temperature effect. Mechanical strengths are modelled using two cascade feed-forward neural networks (NNs), designated CSNN-CNTs and CSNN-CLY for CNTs and nanoclays, respectively. The trained NN CNT and nanoclay models both successfully replicated experiments with significant accuracy. Inputs to the NN models were different mix combinations, optimised using a genetic algorithm to achieve optimal strength results. The optimisation process for mortars with CNTs revealed increases of 20·96% and 54·27% for compressive and flexural strengths, respectively. Similarly, optimisation of the mortar with nanoclays achieved increases of 63·7%, 199% and 133% for compressive, tensile and flexural strengths, respectively. These results demonstrate that the NN integrated genetic algorithm based optimisation is an effective way of determining the best recipe for CNT and nanoclay mortars for optimum strength.

A Unique Approach to Epilepsy Classification from EEG Signals Using Dimensionality Reduction and Neural Networks

Characterized by recurrent and rapid seizures, epilepsy is a great threat to the livelihood of the human beings. Abnormal transient behaviour of neurons in the cortical regions of the brain leads to a seizure which characterizes epilepsy. The physical and mental activities of the patient are totally dampened with this epileptic seizure. A significant clinical tool for the study, analysis and diagnosis of the epilepsy is electroencephalogram (EEG). To detect such seizures, EEG signals aids greatly to the clinical experts and it is used as an important tool for the analysis of brain disorders, especially epilepsy. In this paper, the high dimensional EEG data are reduced to a low dimension by incorporating techniques such as Fuzzy Mutual Information (FMI), Independent Component Analysis (ICA), Linear Graph Embedding (LGE), Linear Discriminant Analysis (LDA) and Variational Bayesian Matrix Factorization (VBMF). After employing them as dimensionality reduction techniques, the Neural Networks (NN) such as Cascaded Feed Forward Neural Network (CFFNN), Time Delay Neural Network (TDNN) and Generalized Regression Neural Network (GRNN) are used as Post Classifiers for the Classification of Epilepsy Risk Levels from EEG signals. The bench mark parameters used here are Performance Index (PI), Quality Values (QV), Time Delay, Accuracy, Specificity and Sensitivity.

Constructive algorithm for fully connected cascade feedforward neural networks

In this paper, a novel constructive algorithm, named fast cascade neural network (FCNN), is proposed to design the fully connected cascade feedforward neural network (FCCFNN). First, a modified index, based on the orthogonal least square method, is derived to select new hidden units from candidate pools. Each hidden unit leads to the maximal reduction of the sum of squared errors. Secondly, the input weights and biases of hidden units are randomly generated and remain unchanged during the learning process. The weights, which connect the input and hidden units with the output units, are calculated after all necessary units have been added. Thirdly, the convergence of FCNN is guaranteed in theory. Finally, the performance of FCNN is evaluated on some artificial and real-world benchmark problems. Simulation results show that the proposed FCNN algorithm has better generalization performance and faster learning speed than some existing algorithms.

Automatic segmentation of colon in 3D CT images and removal of opacified fluid using cascade feed forward neural network.

Purpose. Colon segmentation is an essential step in the development of computer-aided diagnosis systems based on computed tomography (CT) images. The requirement for the detection of the polyps which lie on the walls of the colon is much needed in the field of medical imaging for diagnosis of colorectal cancer. Methods. The proposed work is focused on designing an efficient automatic colon segmentation algorithm from abdominal slices consisting of colons, partial volume effect, bowels, and lungs. The challenge lies in determining the exact colon enhanced with partial volume effect of the slice. In this work, adaptive thresholding technique is proposed for the segmentation of air packets, machine learning based cascade feed forward neural network enhanced with boundary detection algorithms are used which differentiate the segments of the lung and the fluids which are sediment at the side wall of colon and by rejecting bowels based on the slice difference removal method. The proposed neural network method is trained with Bayesian regulation algorithm to determine the partial volume effect. Results. Experiment was conducted on CT database images which results in 98% accuracy and minimal error rate. Conclusions. The main contribution of this work is the exploitation of neural network algorithm for removal of opacified fluid to attain desired colon segmentation result.

Comparison of Different Neural Network Architectures for Classification of Feature Transformed Data for Face Recognition

In this paper neural network classifier is applied on transformed shape features for face recognition. Classification by neural networks to a large extent depends on the neural network architecture. We have investigated three different neural network architectures for classification namely-Feed Forward Neural Network, Cascade Feed Forward Neural Network and Radial Basis Function Neural Network and tested their performance for three sets of feature extracted data. For feature extraction we convert the 2-D gray level face images into their respective depth maps or physical shape which are subsequently transformed by three different methods to get three separate data sets,namely-Coiflet Packet , Radon Transform and Fourier Mellin Transform to compute energy for feature extraction. After feature extraction each of the training classes are optimally separated using linear discriminant analysis. The neural network classifiers have been tested on each of the three sets of feature extracted data and a comparative analysis has been done on the results obtained. The proposed algorithms have been tested on the ORL database, widely used for face recognition experiments. General Terms Pattern Recognition, Computer Vision, Artificial Intellegence,Image Processing

Neural Network Guided Spatial Fault Resilience in Array Processors

Computing with large die-size graphical processors, that involves huge arrays of identical structures, in the late CMOS era is abounding with challenges due to spatial non-idealities arising from chip-to-chip and within-chip variation of MOSFET threshold voltage. In this paper, we propose a software-framework using machine learning for in-situ prediction and correction of computation corrupted due to threshold voltage variation of transistors. Semi-supervised training is imparted to a fully connected cascade feed-forward (FCCFF) neural network (NN). This FCCFF-NN then creates an accurate spatial map of faulty processing elements (PE), which are avoided in computing. Besides correcting spatial faults, any transient errors (such as single-event upsets) are also tracked and corrected if the number of affected PEs is large enough to cause noticeable computing errors. For experimental validation, we consider a 256 × 256 PE array. Each PE is comprised of add-accumulate-multiply (AAM) block with three 8-bit registers (two for inputs and a third for storing the computed result). One thousand instances of this processor array are created and PEs in each instance are randomly perturbed with threshold voltage variation. Common image processing operations such as low pass filtering and edge enhancement are performed on each of these 1,000 instances. A fraction of these images (about 10 %) is used to train the NN for spatial non-idealities. Based on this training, the NN is able to accurately predict the spatial extremities in 95 % of all the remaining 90 % of the cases. The proposed NN based error tolerance produces superior quality processed images whose degradation is no longer visually perceptible.

A New Iris Detection Method based on Cascaded Neural Network

Iris recognition is one of the most reliable and applicable methods for a person's identification. The most complex and important phase of recognition is iris segmentation of an input eye image that affects iris recognition successful rate significantly. Due to missed parameters in noisy images, main error occurs in the performance of classic localization. Artificial neural networks (ANN) are appropriate substitutes for classic methods because of their flexibility on noisy images. In this paper, we use feedforward neural network (FFNN) for the improvement of iris localization accuracy. We apply two methods in order to reduce neural network error: first, designing one neural network for each output neuron .Second, using cascaded feedforward neural network (CFFNN). Then, we examine proposed methods on different datasets which cause remarkable reduction of localization error.

Optimization of the core configuration design using a hybrid artificial intelligence algorithm for research reactors

To successfully carry out material irradiation experiments and radioisotope productions, a high thermal neutron flux at irradiation box over a desired life time of a core configuration is needed. On the other hand, reactor safety and operational constraints must be preserved during core configuration selection. Two main objectives and two safety and operational constraints are suggested to optimize reactor core configuration design. Suggested parameters and conditions are considered as two separate fitness functions composed of two main objectives and two penalty functions. This is a constrained and combinatorial type of a multi-objective optimization problem. In this paper, a fast and effective hybrid artificial intelligence algorithm is introduced and developed to reach a Pareto optimal set. The hybrid algorithm is composed of a fast and elitist multi-objective genetic algorithm (GA) and a fast fitness function evaluating system based on the cascade feed forward artificial neural networks (ANNs). A specific GA representation of core configuration and also special GA operators are introduced and used to overcome the combinatorial constraints of this optimization problem. A software package (Core Pattern Calculator 1) is developed to prepare and reform required data for ANNs training and also to revise the optimization results. Some practical test parameters and conditions are suggested to adjust main parameters of the hybrid algorithm. Results show that introduced ANNs can be trained and estimate selected core parameters of a research reactor very quickly. It improves effectively optimization process. Final optimization results show that a uniform and dense diversity of Pareto fronts are gained over a wide range of fitness function values. To take a more careful selection of Pareto optimal solutions, a revision system is introduced and used. The revision of gained Pareto optimal set is performed by using developed software package. Also some secondary operational and safety terms are suggested to help for final trade-off. Results show that the selected benchmark case study is dominated by gained Pareto fronts according to the main objectives while safety and operational constraints are preserved.