Man has learned much from studies of natural systems, to develop new algorithmic models able to solve increasingly complex problems. Enormous successes have been achieved through modeling of biological and natural intelligence, resulting so-called "intelligent systems". These nature-inspired intelligent technological paradigms are grouped under the umbrella called computational intelligence (CI). On the other side, modern environmental remote sensing satellite imagery, owing to their large volume of high-resolution data, offer greater challenges for automated image analysis. The algorithms are based on the fact that each class of materials, in accordance to its molecularcomposition, has its own spectral signature. Applications are needed both for remote sensing of urban/suburban infrastructure and socio-economic attributes as well as to detect and monitor land-cover and land-use changes. Conventionally, pattern recognition in remote sensing imagery has been mainly based on classical statistical methods and decision theory. Last years, several computational intelligence approaches have been used with promising degrees of success in remote sensing image analysis. This lecture is an approach dedicated to the improvement and experimentation of several nature-inspired intelligent models for pattern recognition in remote sensing imagery. One considers threemain models and corresponding applications. First model is an Unsupervised Artificial Immune System (UAIS), inspired from the vertebrate immune system, having strong capabilities of pattern recognition. We have implemented this model for a LANDSAT 7ETM+ multispectral image from the region of Bucharest (Romania) with four pixel categories (agricultural fields, artificial surfaces, forest, and water); using UAIS, one leads to the correct clustering multispectral pixel score better than performances obtained by applying K-Means and Fuzzy K-Means algorithms. Second modeluses pixel classification byAnt Colony Optimization (ACO) algorithm which takes inspiration from the coordinated behavior of ant swarms. Using the ACO algorithm to remote sensing image classification does not assume an underlying statistical distribution for the pixel data, the contextual information can be taken into account, and it has strong robustness. The results of ACO classification for a Landsat 7ETM+ image dataset (the same as that used in the first model) leads to very good results. Third model corresponds to change detection using neural network techniques: (a) Multilayer Perceptron (MLP); (b) Radial Basis Function Neural Network (RBF); (c) Supervised Self-Organizing Map (SOM). For comparison, one has tested change detection with statistical techniques (Bayes, Nearest Neighbor). The data used for the experiments of neural change detection are selections from a sequence of two LANDSAT 7 ETM+ multispectral images corresponding to the region Markaryd (Sweden), acquisitions from 2002 and 2006. Change detection by neural classifiers have led to better results than those obtained using statistical techniques.
This paper presents an improved Artificial Immune System (AIS) approach for unsupervised classification in multispectral remote-sensing imagery. For benchmarking, one has considered several unsupervised nature-inspired intelligent classifiers (AIS, neural, fuzzy) versus statistical ones. We have comparatively evaluated the following pattern recognition techniques: the proposed AIS model; Self-Organizing Map (SOM); Vector Quantization SOM (VQSOM); Fuzzy C-means, and K-means. The considered techniques have been evaluated using both synthetic and real datasets. The real datasets correspond to the LANDSAT 7 ETM+ multispectral image (341 × 343 pixels) taken in June 2000, representing a region of Bucharest, Romania. There have been considered four pattern classes: artificial surfaces, agricultural area, forest, water. One has also evaluated the case of choosing a balanced dataset from the LANDSAT image, with equal number of 800 selected multispectral pixels per class. For the balanced LANDSAT dataset with 3 bands (1, 4, 5), the best experimental correct recognition score is of 93.78% for AIS model followed by the scores of 89.09% for the 5 × 5 neuron SOM model, 83.28% for VQSOM, 84.18% for Fuzzy C-means, and 83.15% for K-means.
The paper describes the development of software called interactive scenarios, used to develop the student creativity, imagination, dexterity and computer skills by managing a personal computer in a graphical environment. The proposal is aimed at students in preschool, primary and secondary for use inside and outside the classroom. The technique used as a method of learning is PBL "Project Based Learning".
This paper reveal specific quality characteristics of mobile learning applications developed for assisting the collaborative learning process inside virtual organizations. The features of virtual organizations are analyzed and their collaborative character is presented, in order to identify their advantages compared with the classical organizations. M-learning applications are described and mobile learning processes are analyzed inside a virtual campus. An m-learning application is used in the virtual campus of Bucharest Academy of Economic Studies, in order to calculate specific indicators necessary to evaluate the performance of mobile learning processes in collaborative virtual organizations.
In this paper a novel quantification method for large state space attractors is proposed. The suggested approach is briefly described and tested on several dynamical systems with three degrees of freedom. Generalization of the method is for higher dimensional deterministic dynamical systems is also presented. The preliminary results shows that the method can be used for rough recognition of attractor nature and geometry. The significant contribution of proposed approach lies in speed-up the calculation process due to the reduction of one manifold.
In this paper we present numerical solution of Burgers equation using cubic B-spline functions. In the past, difficulties have been experienced while getting its numerical solution. So it is worth to find the schemes that can solve it efficiently giving accurate and stable solutions. The current work aims to develop an algorithm that is easy to understand and implement. Numerical experiments shows that the scheme is capable in achieving results of high accuracy.
This paper analyses Fourier transform used for spectral analysis of periodical signals and emphasizes some of its properties. It is demonstrated that the spectrum is strongly depended of signal duration that is very important for very short signals which have a very rich spectrum, even for totally harmonic signals. Surprisingly is taken the conclusion that spectral function of harmonic signals with infinite duration is identically with Dirac function and more of this no matter of duration, it respects Heisenberg fourth uncertainty equation. In comparison with Fourier series, the spectrum which is emphasized by Fourier transform doesn't have maximum amplitudes for signals frequencies but only if the signal lasting a lot of time, in the other situations these maximum values are strongly de-phased while the signal time decreasing. That is why one can consider that Fourier series is useful especially for interpolation of nonharmonic periodical functions using harmonic functions and less for spectral analysis.
In this paper, we present a numerical technique for approximating a solution of Love's integral equation. The Love's integral equation is a class of second kind Fredholm integral equations, and it can be used to describe the capacitance of the parallel plate capacitor (PPC) in the electrostatic field. This numerical technique developed by Huabsomboon et al. bases on using Taylor-series expansion [4], [5], [6] and [7]. We code a computer program for determining numerical solutions of the Love's integral equation. We compare the numerical solution using Taylor-series expansion technique with the exact solution or with numerical solution obtained by using chebyshev expansion. It is shown that the numerical results are excellent.
A majority of systems that take advantage of human motion in order to recognize gestures are developed through temporal image processing algorithms. However, thanks to the increasing development of acceleration sensors in recent years, it has become possible to use actual arm movements as an acquisition system. This feature could be used in more intuitive systems to communicate reach-to-grasp movements. This research proposes placing an accelerometer on a user's arm to recognize grasping movements in an unique way. The most complex part of this problem revolves around the fact that an accelerometer is unable to evaluate whether a user is performing an reach-to-grasp movement. Given that the movement involves a temporary action, it is possible to use a hidden Markov system to dynamically predict user grasping movements. The results indicate that the model can correctly predict all movements with an F-score = 99% on average.
According to the component-based design and developing pattern, we present a plug-in architecture for the dependable component-based software. After that, the dependable encapsulation is put forwards, which mainly focuses on the component security and availability attributes. Finally, the developing and implementing method of security interceptor, high available load balancing and fault tolerance services have been put emphasis on in this paper.
Bone segmentation in radiographic imaging is an intermediate level processing stage for an automated vision system for the skeletal assessment of children. It is one of the challenging problems in medical image analysis due to high noise levels and low contrast with non-uniform and complex intensity distribution of radiographic image. In this paper, we present a local merging algorithm for automatically segmenting bones from the hand radiograph. With an initial over-segmented image, in which the many primitive (homogeneous) regions are generated by watershed transform and image pre-processing, the hand bone X-ray image segmentation is performed by the local merging process on regions of interest (ROIs). Firstly, the hand is separated from the background to get hand boundary. In this phase, aiming the hand separation coincide with the region reduction, an merging algorithm based on the region adjacent graph (RAG) and nearest neighbour graph (NNG) is proposed. Next, the curvature information of the hand boundary is analyzed for determining the desired ROIs on the hand image. Finally, the sub-RAGs which are sub-graph of the RAG associated with the ROI are extracted, and the local merging process on each sub-RAG is individually executed. Experiments are carried out on 30 hand X-ray images of the young children where the carpal bones have distinct, non-overlapping boundaries. The experimental results show that with the proposed method, an accurate and robust segmentation can be achieved.
Large scale network attack, for instance worm, DDoS, will have serious impacts on network service and network infrastructure. Because of the difficult points that the experiment needs complicated network topology and various defense schemes, researches on test and evaluation of large scale network attack are less. This paper presents metrics and indices of effectiveness evaluation of large scale network attack and defense, and introduces in detail the major methods used to implement an emulation environment.
This paper discusses about the implementation of an Artificial Neural Network (ANN) based model for Neutronics power estimation of Prototype Fast Breeder Reactor (PFBR). ANN has been designed to predict the Neutronics power for various positions of control and safety rods. Simulation studies were carried out using the thermo-hydraulics code and the required data has been generated for training the ANN model. Here, three layer neural network architecture has been developed and trained with different learning algorithms to estimate Neutronics power and model the plant dynamics. The best performing algorithm in terms of faster convergence has been identified among the variants of back propagation network. ANN model has more advantages compared to conventional model namely increased speed, reduced complexity in addition to producing accurate results. Neural network being a data driven model, it is possible to derive the operating characteristics of nuclear reactor without exploring the intricacies of the complex subsystems.
This paper describes a distributed software security testing methods and testing process, according to the characteristics of distributed software, the paper describes the security functional testing indicators, and a system of security indicators of distributed software, and the methods and means of detection of each metrics.
This article deals with the impact of virtualization techniques on interactive delay-sensitive applications running in real-time, particularly IP telephony. Many institutions, organizations and home users often adopt the virtualized solutions for their safety, ease of administration and backup. Virtualization, which was chiefly the prerogative of companies and the academic world in its early days, has gradually develop its platform to reach out to the ordinary users who can benefit from running virtual machines. The aim of this thesis is to examine the impact of a virtual machine on real-time traffic, in our case IP telephony based on the SIP and the RTP. This article also analyses the impact of memory size and the number of processor cores on the delay itself and its variance.
This paper presents an implementation technical solution of network forensics system for Chinese text content. The technical solution utilizes Bloom filter algorithm and Chinese word segmentation and meta-aggregation algorithm (CWSMA) to preprocess and effectively store contents of the text aiming at technical challenges caused by characteristics of "unpredictability of the event features" and "unpredictability of forensics operation", information related with the events such as 'where', 'who', 'when' and the like can be provided for investigators through member query, network verification analysis can be carried out under the condition without predefining event characteristics, the forensics analysis time traceability can be prolonged from several days of existing technique to several months, it is particularly suitable for network forensics of network secret disclosure events and illegal content propagation events with sensitive content analysis.
Reverse program compilation (i.e. decompilation) is a process heavily exploited in reverse engineering. The task of decompilation is to transform a platform-specific executable into a high-level language representation, which is usually the C language. Such a process can be used for source code reconstruction, compiler testing, malware analysis, etc. In present, there are several existing decompilers that are able to decompile simple applications. However, we can see a drop-off in terms of the quality of the generated code when the decompiled code is highly optimized (e.g. usage of instruction idioms) or obfuscated (e.g. dead code insertion, register renaming). Optimized or obfuscated applications are usually generated by highly optimizing compilers or metamorphic engines (used by malware authors). In this paper, we present several innovative decompilation methods based on scattered context grammars. These methods are able to effectively decompile optimized or obfuscated code. For demonstration, we used these methods for enhancement of the static analysis phase of an existing decompiler. Experimental results of our solution are presented at the end of the paper.
In this paper we analyze the use of Moodle VLE by students who attended a fully online graduate-level course. Students were split up into two groups regarding their undergraduate education. The aim of this research is to determine if there is a significant difference between the two student groups concerning both time of assignment submission and number of resources and assignments (re)accessed. We conclude that no significant difference arises regarding time of assignment submission, but this appears with respect to the access to resources and assignments. Furthermore, there is a very low correlation between the time of assignment submission and: the final course scores, the number of resources and the number of assignments (re)accessed.
This paper deals with one of key issues of IP telephony regarding SIP Proxy robustness against attacks and compares security risks inherent to various Denial-of-Service (DoS) attacks and addresses effective protection against them. The proposed solution is based on Snort and SnortSam and has been implemented and evaluated in testbed. Denial of Service - is one of most frequent attacks nowadays due to its simplicity and great impact. This paper describes DoS attack types, and the knowledge is used to test the robustness of the SIP proxy server. Attacks are described in detail, and a security precaution is made to prevent each of them. The solution is an IPS system, composed as a combination of Snort, SnortSam and Iptables applications. The presented solutions were tested in experiments.
This paper presents a speed nonlinear predictive control scheme for a Series DC Motor. The proposed scheme is capable of overcoming the singularity presented when the current approaches to zero and the inherently non-linear behaviour. The control scheme presented is composed of a speed predictive control and an internal model control. The effectiveness of this control algorithm has been successfully verified through simulations.
