In this work, a FeedForwardBackPropagation Neural Network based mammogram classification system was proposed for detection of breast cancer. The biological features as well as the visual content features were used for the classification of mammograms.K-Means clustering process was applied on the extracted features and the extracted features are combined to obtain better classification performance. The proposed system with K-Means clustering has given higher performance and confirmed that the performance of our proposed technique is better than the other
Various methods have been used to identify a person through the fingerprint image. In this work to increase the accuracy of the result a new approach have been implemented in which each sample is divided into four parts and then feedforwardbackpropagation neural network have been applied to identify a person. Section II describes about the pattern matching and its methods, in section III detailed methodology is shown, section IV contains data preparation while experiment, result and discussions are shown in the section V. Section VI contains conclusion and references are shown in the last section.
Backpropagation calculation is utilized as a part of feedforward ANNs. First term, “feedforward” defines how this neural network works and recalls patterns and the term “backpropagation” defines how this kind of neural networks are accomplished. The network obtains inputs by neurons in the input layer, and the output of the network is given by the neurons on an output layer. There may be one or additional intermediate hidden layers.
Imaging (MRI) is a medical imaging technique used to visualize the internal structure of the body and provide high quality images. MRI contains useful and fine information which is used to improve diagnosis accuracy. This automatic segmentation algorithm gives shape, size and location of the tumor more accurately. It gives a less demanding approach to analyze the tumor and encourages specialists to design the careful methodology. Information of images can be obtained by Statistical analysis, where the possibility of tumor is highest by using mean, entropy and GLCM. Bounding Box technique used to locating tumor areas. Easy to extract features like size and location. Using bounding box method less number of MRI images required for training and also extracting features from each blocks.Which helps to fast detection of abnormal areas. Block matching technique with FFBP used to classification. FeedforwardBackpropagation neural network is a multi-level error feed-back network that reduces error rate. We have tested the proposed approach using a benchmark dataset of MRI brain images. The experimental results show that our approach achieved was 98 % classification accuracy achieved by ANN.
Feedforwardbackpropagation neural network are used for training and testing. Training algorithm is Levenberg- Marquardt, performance is measure by mean square error, and data division is random. Results are marked as bold for the best performance. The results show that the best classification accuracy is 99.998% for two hidden layers. Fig. 6, Fig 7, Fig. 8 respectively show performance plot, regression plot and training state graph for highest performance.
Parametric Autoregressive (AR) modeling, AR spectral analysis and power differences between four frequency bands. In another work the neural network has been used to classify EEG data by using autoregressive with maximum likelihood pre-process for epileptic seizure detection. The purpose of the work was to investigate the use of autoregressive (AR) model by using maximum likelihood estimation (MLE) also interpretation and performance of this method to extract classifiable features from human electroencephalogram (EEG) by using Artificial Neural Networks. Since the voltages recorded on an electro-encephalograph are the result of many processes that occur simultaneously in the brain, only events that involve larger areas of the brain, such as epileptic seizures, can be readily identified on the EEG recording. For this reason, mental disorders generally cannot be diagnosed from the electro-encephalograph. In this paper back- propagationfeedforward neural network have been used to identify the status of mental retarded patient [3-14].
Feedforwardback-propagation neural networks were developed using the JOONE toolset [8] which is an object based neural network framework with a graphical user interface. EasyNN-plus [9] was used to validate the output from Joone. The neural network architecture took the input data from an Excel spreadsheet entering the input layer containing 122 neurons, the data progressed to a hidden layer containing 10 neurons, before it finally reached the output layer which contained a single neuron. The output value was in the range zero to one and was passed into an Excel spreadsheet, all three layers utilised the sigmoid activation function [10]. The Teacher layer trained the network by presenting it with complete examples, including whether the example was a terrorist or not (this is known as supervised learning). The training was then presented graphically via a Root Mean Square Error chart (RMSE) [11] examples of which are presented within the DScent Final Report [1].
the patch dimensions i.e. length (L) and width (W) of the patch. The different variants of backpropagation training algorithm of MLFFBP-ANN (Multilayer feedforwardbackpropagation Artificial Neural Network) and RBF-ANN (Radial basis function Artificial Neural Network) has been used to implement the network model. The results obtained from artificial neural network when compared with simulation results, found satisfactory and also it is concluded that RBF network is more accurate and fast as compared to different variants of backpropagation training algorithms of MLPFFBP. The ANNs results are more in agreement with the simulation findings. Neural network based estimation has the usual advantage of very fast and simultaneous response of all the outputs.
Investigation of soil properties like Cation Exchange Capacity (CEC) plays important roles in study of environmental reaserches as the spatial and temporal variability of this property have been led to development of indirect methods in estimation of this soil characteristic. Pedotransfer functions (PTFs) provide an alternative by estimating soil parameters from more readily available soil data. 70 soil samples were collected from different horizons of 15 soil profiles located in the Ziaran region, Qazvin province, Iran. Then, multivariate regression and neural network model (feed-forwardbackpropagation network) were employed to develop a pedotransfer function for predicting soil parameter using easily measurable characteristics of clay and organic carbon. The performance of the multivariate regression and neural network model was evaluated using a test data set. In order to evaluate the models, root mean square error (RMSE) was used. The value of RMSE and R 2 derived by ANN model for CEC were 0.47 and 0.94 respectively, while these parameters for multivariate regression model were 0.65 and 0.88 respectively. Results showed that artificial neural network with seven neurons in hidden layer had better performance in predicting soil cation exchange capacity than multivariate regression.
dimension of the rainfall pattern, thus, provides better forecasting accuracy [47]. Geetha and Selvaraj (2011) developed a backpropagation neural network model for rainfall prediction in Chennai, India. The mean monthly rainfall was predicted by them using that model. The model can perform well both in training and independent periods [48]. El-shafie et al. (2011) have tried to use neural network and regression technique for rainfall-runoff prediction and finally they concluded that the feedforwardbackpropagation ANN can describe the behavior of rainfall-runoff relation more accurately than the classical regression model [49]. In another research integrated artificial neural network-fuzzy logic-wavelet model is employed to predict Long term rainfall by Afshin et al. (2011). The results of the integrated model showed superior results when compared to the two year forecasts to predict the six-month and annual periods. As a result of the root mean squared error, predicting the two-year and annual periods is 6.22 and 7.11, respectively. However, the predicted six months shows 13.15 [50].
In this study, the availability of ANN models, which is developed as an alternative technique against to em- pirical methods such as Penman, were examined. Penman needs complex calculations and too many parameters belongs to study area, but ANN can obtain more significant results with less parameters than empirical equations such as Penman. ANN is an artificial technique which developed by using properties of neural system such as definition, prediction and reproduce of knowledge [12]. It is possible to come across too many algorithm types which developed within ANN. In this study, three different ANN algorithm; Levenberg-Marquardt algorithm which is one of the feedforwardbackpropagation algorithm [13], Radial Basis Neural Network (RBNN) [14] and Generalized Regression Neural Network (GRNN) [15] were used. Details of method were presented by Okkan and Dalkılıç [16] [17].
The input signal propagates through the network layer-by- layer. The signal-flow of such a network with two hidden layer is shown in Figure 1. Multi layer feedforwardbackpropagation algorithm is used to train the network and tests the performance of the network. MLP networks are typically used in supervised learning problems. This means that there is a training set of input-output pairs and the network must learn to model the dependency between them. Multilayer Perceptron (MLP) network is a popular learning algorithm in a sense that neural network knows the desired output and adjusting of weight coefficients is done in such way, that the calculated and desired outputs are as close as possible. This paper is organized as follows. In next section, we introduce some related background including some basic concepts of backpropagation algorithm with learning, training data set and testing data set determination and backpropagation network generation using matlab toolbox. In the following section, we explain results and performance evaluation of multilayer perceptron (MLP) using backpropagation algorithm. Finally, we have some conclusions with future work.
Abstract- In this paper, for region segmentation some methods are used to segment the image. The main steps of texture processing technique such as filterization, feature extraction, segmentation, classification. For filterization, the Gabor/Gabor Wavelet filters are to be used and then some useful informations are to be extracted from the filtered images through spatial smoothing method. Further step is to clustering the feature vectors through K Means and Fuzzy C- mean clustering. Then we compare between Gabor and Gabor Wavelet. And also compare between K means and Fuzzy c mean clustering. Clustering is an unsupervised classification technique. For better result, supervised classification technique have been used which is feedforwardbackpropagation Neural network. In this paper, the comparison of segmentation results generated using unsupervised and supervised approach is presented. Supervised approach using FeedForwardBackPropagation algorithm shows promising result compared to unsupervised technique.
The annealing process is one of the important operations in production of cold rolled steel sheets, which significantly influences the final product quality of cold rolling mills. In this process, cold rolled coils are heated slowly to a desired temperature and then cooled. Modelling of annealing process (prediction of heating and cooling time and trend prediction of coil core temperature) is a very sophisticated and expensive work. Modelling of annealing process can be done by using of thermal models. In this paper, Modelling of steel annealing process is proposed by using data mining techniques. The main advantages of modelling with data mining techniques are: high speed in data processing, acceptable accuracy in obtained results and simplicity in using of this method. In this paper, after comparison of results of some data mining techniques, feedforwardbackpropagation neural network is applied for annealing process modelling. A good correlation between results of this method and results of thermal models has been obtained.
Abstract— Face recognition is one of the most relevant applications of image analysis. It’s an efficient task (true challenge) to build an automated system with equal human ability to face recognised. Face is a complex 3D visual model and developing a computational model for face recognition is a difficult task. The paper presents a methodology for face recognition based on information theory approach of coding and decoding the face image. Proposed methodology is combination of two stages – Feature extraction using principle component analysis and recognition using the feedforwardbackpropagation Neural Network. The proposed method has been tested on Oracle Research Laboratory (ORL) face database containing 400 images (40 classes). A recognition score for the test lot is calculated by considering almost all the variants of feature extraction. The test results gave a recognition rate of 99.50%.
An extensive experimental study was conducted by Lee and Li (2001) to investigate the effect of machining parameters such as the electrode materials, electrode polarity, open circuit voltage, peak current, pulse duration, pulse interval and flushing on the machining characteristics, such as MRR, surface finish and relative tool wear in EDM of tungsten carbide. They observed that the MRR generally decreases with the increase of open circuit voltage. For low current setting, the MRR increases with increase in peak current, but becomes constant when machining at higher values of peak current. The surface roughness increases with increasing peak current. Increase in pulse duration results in increase in MRR. Lee and Li (2003) studied surface integrity of EDMed surface of tungsten carbide. They found that the surface roughness is a function of two main parameters, peak current and pulse duration, both of which were settings of the power supply. High peak current and/or long pulse duration produces a rough surface. At high peak current and pulse duration abundance of micro- cracks was observed. Saha et al. (2008) developed a second order multi-variable regression model and a feedforwardback-propagation neural network to correlate the input process parameters, such as pulse on time, pulse-off time, peak current and capacitance with the performance measures namely cutting speed and surface roughness while doing WEDM of tungsten carbide-cobalt composite material. Increase in both peak current and capacitance led to increase of cutting speed and surface roughness within the range of investigation. Chen at al. (2010) optimized the WEDM for pure tungsten using an approach that integrates Taguchi’s parameter design method, back-propagation neural network, genetic algorithm and engineering optimization concepts. Through ANOVA, the percentage of contribution to the WEDM process, the pulse on time is the most significant controlled factor affecting the cutting speed and surface roughness. Several other researchers (Puertas et al., 2004; Kanagarajan et al., 2006; Kung et al., 2007; etc.) investigated the performance of EDM in processing of tungsten carbide.
Abstract: In this paper we investigate the use of the feed-forwardbackpropagation neural networks (FFBPNN) for automatic speech recognition of Arabic letters with their four vowels (Fatha, dhamma, Kasra, Soukoun). This investigation will constitute a basically step for the recognition of continuous Speech. Features were extracted from recorded corpus by using a variety of conventional methods such as Linear Predictive Codes (LPC), Perceptual Linear Prediction (PLP), Relative Spectral Perceptual Linear Prediction (RASTA-PLP), Mel Frequency Cepstral Coefficients (MFCC), Continuous Wavelet Transform (CWT), etc. Here, several hybrid methods have been used too. Since the extracted features have large dimensionalities they were reduced by conserving the most discriminatory information with the Principal Component Analysis (PCA) technique. The recognition performance has been improved particularly when we use the PLP method followed by PCA technique.
Based on the results reported in the literature [1to9] for Iris recognition, gopikrishnan m etal [10] studied Hamming distance coupled with Neural Network based iris recognition techniques. Perfect recognition on a set of 150 eye images has been achieved through this approach ; Further, Tests on another set of 801 images resulted in false accept and false reject rates of 0.0005% and 0.187% respectively, providing the reliability and accuracy of the biometric technology. In a subsequent paper same authors provided [11- 12] results of iris recognition performed on a reduced size template, applying Hamming distance, Feedforwardbackpropagation, Cascade forwardbackpropagation, Elman forwardbackpropagation and perceptron. It has been established that the method suggested applying perceptron, using hardlim training function and learnp learning function, provides the best accuracy in respect of iris recognition with no major additional computational complexity. This paper uses the CASIA iris image database collected by Institute of Automation, Chinese Academy of Sciences [13]. In this paper,
Design optimization of frame structure using approximations based on feed-forwardback-propagation neural network was explored in this paper. Implementation of NN has successfully conducted in this project with proper training of optimized data. Three cases have been investigated, and results show neural network performs well to predict the optimization criteria. Some adjustment of learning rate, epoch number as well as number of neurons and hidden layers has been done in order to increase the efficiency of NN performance. Mean square error analysis reveals that higher learning rate applied to particular situation, will yield better performance of NN scheme obtained.
In this work a feedforwardBackPropagation Neural Network (BPNN) will be used to predict the fractures dip angle for the third well using the image log and other geological log data of the two other wells nearby. The new method can save costs and time in drilling and production operations. It can reduce the risk of drilling operation and post fracturing job. It can be also used in real time logging operation and many other benefits that in next chapters will be mentioned.