Rice false smut is one of the most dangerous diseases in rice at the ripening phase caused by Ustilaginoidea Virens. It is one of the most important grain diseases in rice production worldwide. Its epidemics not only lead to yield loss but also reduce grain quality because of multiple mycotoxins generated by the causative pathogen. The pathogen infects developing spikelets and specifically converts individual grain into rice false smut ball. Rice false smut balls seem to be randomly formed in some grains on a panicle of a plant in the paddy field. In this study, we suggest a novel approach for the detection of rice false smut based on faster R-CNN. The process of faster R-CNN comprises regional proposal generation and object detection. The both tasks are done in same convolutional network. Because of such design it is faster for object detection. The faster R-CNN is able to detect the RFS using rectangular labelling from on-field images. The proposed approach is the initial steps to make a prototype for the automatic detection of RFS.

/pdf/rice-false-smut-detection-based-on-faster-r-cnn-4vxc1geibm.pdf

Rice false smut detection based on faster R-CNN

Person identification based on voice biometric using deep neural network

This research aims to analyze the effect of feature selection on the accuracy of music genre classification using support vector machine with radial basis function kernel as a classifier. In this research, the music dataset from Spotify is used, which is one of the best-selling music streaming platforms today. The selected feature is metadata because it is considered to have simpler processing than audio feature extraction. The music contained in the Spotify dataset also has complete metadata so that the metadata feature can be used properly. At the feature selection stage, some features are combined in different combination groups (FC1, FC2, FC3, FC4). The classification results prove each feature combination has an accuracy result that has a significant difference, where the best accuracy is 80% and the lowest is 67%. Where the combination of FC1 and FC2 features produces the same accuracy of 80%, but because FC2 has a smaller number of features, so the FC2 combination is recommended because with fewer features, so logically the computing time is shorter.

Effect of Feature Selection on The Accuracy of Music Genre Classification using SVM Classifier

Sharing of information is one of the essential characteristic that makes the quality of businesses getting better. With the advent of information and communication technology (ICT), electronic information sharing is increasingly required in order to support decision making. Ministry of Higher Education and Scientific Research (MOHESR) in Iraq has given public universities authorities in making decisions. However, not all Iraqi public universities are able to make own decisions due to low experiences and insufficient resources, such as information. The lack of electronic information sharing between universities and MOHESR is considered as a huge gap in order to increase the universities’ information. Four characteristics known as electronic information sharing, technological, organizational and environmental are studied to understand how they could influence universities to make better decision making.

Electronic Information Sharing to Improve Decision Making in Public Universities

Beaufort and Vigenere are fast and strong substitution encryption methods against brute force attacks if they have good key quality. But both of these methods by default can not produce a good avalanche effect. Avalanche effect is a measuring tool that can determine the strength of encryption from differential attacks. This research proposes hybrid techniques of Beaufort and Vigenere algorithms to optimize the security of text message encryption by modifying key generators. With this method, the encryption process can be done by input text messages and keys. The key is processed with a key generator to produce two new keys that are used for encryption using the Beaufort and Vigenere hybrid methods. Based on the test results, the Avalanche effect value rose significantly compared to the standard method, and the value of the avalanche effect was more stable and close to ideal compared to the previous method. The time required is also relatively very fast for the encryption or decryption process, experimental result shows it takes less than 0.1 seconds for more than 43 thousand of characters in the text message.

https://iopscience.iop.org/article/10.1088/1742-6596/1577/1/012032/pdf

Securing Text Messages using the Beaufort-Vigenere Hybrid Method

Abnormal head movements play a crucial role in diagnoisis of varity diseases. Moreover, different studies considered with these type of information. In addition, the gestures based mainly on head movement which can be employed in many applications such as using head-nodding or shaking to feedback content-related feedback, detect and interpret the emotion, gaze orientation, focus of attention, driver assistance system and so on. In this paper, a new method proposed to detect and classify the flopping head movements as normal or abnormal based on Convolution Neural Network CNN in the term of special sense interaction and behavioral studies with this movement. The proposed system based on deep learning employing the Convolution Neural Network CNN as most promising approach to deal with lighting condition (illumination change) and distortion and noise. Normal Abnormal Head Movement Dataset (NAHM) static images dataset gathered and used in proposed system implementation. This dataset provided the various image conditions and subjects to prevent the overfitting and under fitting problem that appears with publically datasets. The proposed frame work presents robust learning ability based on accuracy and lose functions which achieved training loss: 0.0106, training accuracy: 0.9980, validation loss: 0.0968 and validation accuracy: 0.9831.

Abnormal head movement classification using deep neural network DNN

The development of tracking and surveillance devices makes extracting useful information efficiently. Head tracking is an efficient method to obtain then analyze trajectory data and make a decision based on the spatiotemporal information of videos. Many applications are based on head tracking such as diseases some diagnosis,  the gestures languages, and drowsiness detection and so on. Abnormal head movement detection can be achieved using spatial information based on a single image (one frame) at a time without considering the temporal information over time. In this paper, a new method based on multi-images is proposed to track head in order to detect abnormal head movement depending on spatiotemporal Feature using Deep Neural Network DNN that employed the 3-Dimensional Convolution Neural Networks 3D CNN. The proposed method extracts the spatial information as well as the temporal information available in a video then analysis this information to make the decision based on time series (sequences of frames); these time series provides the tracking to the head overtime to make the decision. The new dataset created and gathered to implement with the proposed system and called Normal Abnormal Head Movement Dataset (NAHM) video dataset. The new dataset provides different subjects with different conditions that give more efficiency in the implementation of the proposed system. The accuracy of the training set that achieves by the proposed system reach to 88% and of validation set reaches to 86%. The values of loss function reach to 0.3 for the training set and 0.4 for the validation set.

/pdf/detecting-abnormal-movement-of-driver-s-head-based-on-5a1wps0hov.pdf

Detecting abnormal movement of driver's head based on spatial-temporal features of video using deep neural network DNN

/pdf/cryptographic-system-based-on-unicode-1ntkpyhmrj.pdf

Cryptographic system based on Unicode

In this paper, we present an neural network approach to solve a set of linear algebraic equations. A Black-Box NeuroIdentifier network has been developed to optimize an convergence until getting the optimal solutions. BlackBox approach called also (input-output description) which is used when no information is available about the system except its input and output. The input-output description of a system gives a mathematical relationship between the input and output of the system. In developing this description, the knowledge of the internal structure of a system may be assumed to be unavailable; the only access to the system is by means of the input and output terminals. Our approach called Black Box Neural Network(BBNN) which provided an optimal solution and a faster way to solve the linear systems; and considered the best with large systems ( the number of equations and parameters very large) and conceded the first step to apply this approach with nonlinear algebraic equations which needs difficult computes and long time relatively. : صخلملا يزناً ًعبنطصلاا ًجصعنا زًًٍنا واذخزسبث خٍطخنا خٌشجدنا دلادبعًنا خًظنأ محن حذٌذخ خقٌشط ىٌذقر ىر ثحجنا ازى ًف ذشزقًنا وبظننا واذخزسا ىرً . دبخشخًناً دلاخذًنا شٍغ وبظننا نع دبيٌهعي يأ شفٌٌ لا ونلأ كنرً دٌسلأا قًذنصنبث ىًسٌ يلأا محنا ىهع لٌصحهن هزى شفٌر .خٍطخنا خًظنلأن جربًننا ني خعًٌدي ىهع خكجشنا تٌسذر ذعث كنرً داشٍغزًنا ىٍقن مث ىًسر بنزقٌشطً .مثيلأا محهن اذخ شٍجك ةسبقرً اذخ خقٍقد حئبزن ىهع لٌصحنا ىنا يدؤٌ ازى ً اذخ خئطاً بطخ خجسن خقٌشطنا ٍجسن خعٌشس لٌهح شفٌر ًزناً خٍجصعنا دبكجشهن دٌسلاا قًذنصنا مّحن خٌذدي بىسبجزعا نكًي كنزنً خٌذٍهقزنا خًظنلأبث خنسبقي ب خٍّطخنا خًظنلأا خًظنأ عي بيقٍجطزن ىنًلأا حٌطخنا شجزعرً .)اذخ حشٍجك داشٍغزًناً دلادبعًنا دذع( حشٍجكنا دبعثلأا دار ,حشٍجكنا .بٍجسن مٌٌط ذقًً حذقعي دبثبسح ىنإ جبزحر ًزنا ً خٍطخلانا خٌشجدنا دلادبعًنا 1.Introduction: The problem of solving linear equations is common to many scientific and engineering applications. The fundamental problem of the designer of algorithms, for the solution of linear equations, has been to devise means for reducing the solution time. The basic problem in the solution of linear equations is to determine the unknown solution vector, x, in the equation Ax = b (where A is a real non-singular matrix of order N and b is a known right hand side vector). There are mainly two classes of methods for solving Ax = b, namely iterative and direct (it is also possible to have methods that have the features of both) . Iterative methods employ a starting solution and converge to a value that is close to the exact solution by iteratively refining the starting solution. However, iterative methods do not guarantee a solution for all systems of linear equations and they are more frequently used for solving large sparse systems of linear equations. On the other hand, direct methods are those in which for a given linear system of equations, exact solutions (within a round-off bound determined by machine precision) can be obtained with a preestimated number of arithmetic operations. The central idea of direct methods of solving linear equations is : Journal of Kerbala University , Vol. 10 No.4 Scientific . 2012 526 1. either to find the inverse of A and then compute x = A −1 b. The method based on semi rings uses this concept without explicitly finding A −1 (but computes A −1 b implicitly). 2. or to transform the coefficient matrix, A, into an equivalent triangular or diagonal form in which coupling between the unknowns is reduced. The methods based on Gaussian Elimination, LU decomposition, Givens rotations, and Householder reductions use the concept of converting the coefficient matrix A into an equivalent triangular form and then solve the resulting triangular system of linear equations. On the other hand, the methods based on Gauss-Jordan, Cramer’s Rule, And Bidirectional Gaussian Elimination convert the given coefficient matrix A into Diagonal form and then obtain the solution vector x through divisions. The iterative methods are used to solve the linear and nonlinear algebraic equations . Usually it is hard to solve a large system of highly-linear and nonlinear algebraic equations. Although a lot of priori research has been conducted in this area, we still lack an efficient and reliable algorithm to solve this difficult problem. many numerical methods used in computational mechanics, as demonstrated by Zhu, Zhang and Atluri [1], Atluri and Zhu [2], Atluri [3],Atluri and Shen [4], and Atluri, Liu and Han [5] lead to the solution of a system of linear algebraic equations for a linear problem . Neural networks are used in many different application domains in order to solve various information processing problems. They have proven to be successful in pattern recognition, pattern classification, function approximation, prediction, optimization, and controlling [6]. The basic processing elements of neural networks are artificial neurons. In order to perform complex tasks, these neurons have to be interconnected in an adequate way by arranging them in an appropriate architecture and setting the weights of their interconnections. The setting of weights is performed during a training phase. In a supervised learning process, input data and target data pairs are presented to the network and the weights of the connections are altered by a learning algorithm as long as for (ideally) all input data, the output values of the network match to the target data [7]. To find an efficient solution to a certain problem, preprocessing of input data is essential to reduce the complexity of the network and the time needed for training [8], [9], [10]. The input-output description of an artificial neural network system gives a mathematical relationship between the input and output of the system. In developing this description, the knowledge of the internal structure of a system may be assumed to be unavailable; the only access to the system is by means of the input and output terminals [12]. Under this assumption, a system may be considered a Black-Box. Clearly what one can do to a black box is to apply inputs and measure their corresponding outputs, and then try to abstract key properties of the system from these input output pairs. An input-output model assumes that the new system output can be predicted by the past inputs and outputs of the system [12, 13]. In this research we present the problem of solving linear algebraic equations as inputoutput description and used that blackbox to find the optimal values of variables in linear system. 2. Systems of Linear Algebraic Equations: Linear systems of equations are the basis for many, if not most, of the models of phenomena in science and engineering, and their efficient. numerical solution is critical to progress in these areas. 2.1 Linear Algebraic Equations Many physical systems are described by systems of linear equations that describe the relationships between the variables of these systems. The general shape of the system of linear equations . Given numbers a11, . . . , amn, b1, . . . , bm, a non-homogeneous system of m linear equations in n unknowns x1, x2, . . . , xn is the system : a11x1 + a12x2 + ・ ・ ・ + a1nxn = b1 a21x1 + a22x2 + ・ ・ ・ + a2nxn = b2 ... (1) am1x1 + am2x2 + ・ ・ ・ + amnxn = bm Journal of Kerbala University , Vol. 10 No.4 Scientific . 2012 527 Constants a11, . . . , amn are called the coefficients of system (1). Constants b1, . . . , bm are collectively referenced as the right hand side,right side or RHS. The homogeneous system corresponding to system (1) is obtained by replacing the right side by zero: a11x1 + a12x2 + ・ ・ ・ + a1nxn = 0 a21x1 + a22x2 + ・ ・ ・ + a2nxn = 0 ... (2) am1x1 + am2x2 + ・ ・ ・ + amnxn = 0 An assignment of possible values x1, . . . , xn which simultaneously satisfy all equations in (1) is called a solution of system (1). Solving system (1) refers to the process of finding all possible solutions of (1). The system (1) is called consistent if it has a solution and otherwise it is called inconsistent . Can be classify linear systems to: 1 square systems: Are those systems where the number of equations equals the number of variables, i.e m = n. 2 over-determined systems : The systems are those where the number of equations is greater than any number of variables m > n . 3 under-determined systems: The systems are those where the number of quations is smaller than any number of variables m < n . Since most systems of linear equations that appear in practical applications and engineering systems is a kind of square, we will look to study methods of finding solutions to these systems only. 2.2 Methods of Solving Linear Equations Systems Find methods to divide the solutions to these systems of linear equations for the square to the two main groups: A) Direct methods: the methods in which those Pettm where access solutions to systems of linear equations specified number of calculations. These methods suitable for finding solutions to relatively simple systems, i.e systems that contain a small number of equations. these methods is : 1Gaussian Elimination with Backward Substitution . 2Gauss-Jordon 3Using Matrix Factorization . 4Using Gramer rule . B) Iterative methods: the methods in which these are the repetition of step (or set of steps) until reaching values of variables to find the required accuracy. These methods suitable for finding solutions to systems consisting of a large number of equations. these methods is : 1 Jacobi method. 2 Gauss – Seidel method. 3. Artificial Neural Networks Artificial Neural Networks (ANNs) are simplified models of the central nervous system. They are networks of highly interconnected neural computing elements that have the ability to respond to input stimuli. Among the capabilities of ANN, are their ability to learn adaptively from dynamic environments to establish a generalized solution through approximation of the underlying mapping between input and output. This simple type of network is interesting because the hidden un

Noor D. Al-Shakarchy

Papers

Abnormal head movement classification using deep neural network DNN

Person identification based on voice biometric using deep neural network

Detecting abnormal movement of driver's head based on spatial-temporal features of video using deep neural network DNN

Cryptographic system based on Unicode

Application of Neural Network For Solving Linear Algebraic Equations