Computational Intelligence in Data Mining—Volume 1

Biometric system using face recognition is the frontier of the security across various applications in the fields of multimedia, medicine, civilian surveillance, robotics, etc. Differences in illumination levels, pose variations, eye-wear, facial hair, aging and disguise are some of the current challenges in face recognition. The ear, which is turning out to be a promising biometric identifier having some desirable properties such as universality, uniqueness, permanence, can also be used along with face for better performance of the system. A multi-modal biometric system combining 2D face, 3D face (depth image) and ear modalities using Microsoft Kinect and Webcam is proposed to address these challenges to some extent. Also avoiding redundancy in the extracted features for better processing speed is another challenge in designing the system. After careful survey of the existing algorithms applied to 2D face, 3D face and ear data, we focus on the well-known PCA (Principal Component Analysis) based Eigen Faces algorithm for ear and face recognition to obtain a better performance with minimal computational requirements. The resulting proposed system turns out insensitive to lighting conditions, pose variations, aging and can completely replace the current recognition systems economically and provide a better security. A total of 109 subjects participated in diversified data acquisition sessions involving multiple poses, illuminations, eyewear and persons from different age groups. The dataset is also a first attempt on the stated combination of biometrics and is a contribution to the field of Biometrics by itself for future experiments. The results are obtained separately against each biometric and final decision is obtained using all the individual results for higher accuracy. The proposed system performed at 98.165 % verification rate which is greater than either of the dual combinations or each of the stated modality in a statistical and significant manner.

An EOQ (Economic Order Quantity) model for a deteriorating item with quadratic demand pattern and quadratic holding cost and constant deterioration rate is considered in this paper. In addition, shortages and partial backlogging are allowed. It is assumed that the backlogging rate acts as not only a variable, but also depends on the length of the waiting time up to next replenishment during the stock out period. For this model, average total cost is derived. Finally, a numerical example for illustration is provided.

A novel Procedure is presented for modeling of higher order discrete systems based on matching the time responses of the original and reduced order systems. The flexibility of method is shown through familiar example.

Pose is a natural and important covariate in case of newborn and face recognition across pose can troubleshoot the approaches dealing with uncooperative subjects like newborn, in which the full power of face recognition being a passive biometric technique requires to be implemented and utilized. To handle the large pose variation in newborn, we propose a pose-adaptive similarity method that uses pose-specific classifiers to deal with different combinatorial poses. A texture based face recognition method, Speed Up Robust Feature (SURF) transform, is used to compare the descriptor of testing (probe) face with given training (gallery) face descriptor. Probes executed on the face template data of newborn described here, offer comparative benefits towards affinity for pose variations and the proposed algorithm verdicts the rank 1 accuracy of 92.1 %, which demonstrates the strength of self learning even with single training face image of newborn.

SQL based traditional databases like Oracle; SQL server offers the capability to develop programs like Trigger. Trigger is a very important feature provided by many databases, especially useful in monitoring, rule enforcement, data validation and data analytics etc. MongoDB is non SQL document oriented database. MongoDB is the fastest growing and most demanding non SQL database. Since MongoDB primarily is operated using its out of box tools like mongo, mongos, bsondump, mongod, mongoexport and Java script function. MongoDB does not provide in-built feature for triggers which is very efficient in data analytics, monitoring and reporting purpose. Paper presents two utility in which one utility is a listener or poller utility which is developed to give similar feature like trigger and after that second utility is developed which gives historical data analytic capability on Mongo database by using the trigger utility. It pulls the data from analytic collection and generates the graph. Data analytic tools plays vital role in decision making in today’s complex business environment where data size is very huge and unstructured by nature.

Heart diseases are the major cause of mortality in developed as well as developing countries. Early prediction of heart disease is required to reduce the number of deaths occurring due to it and by using handwriting analysis we can achieve this. Handwriting of an individual shows the presence of a heart disease even before physical symptoms appear. The proposed system predicts presence of heart disease based on handwriting analysis using k-Nearest Neighbor classifier. It extracts ten writing features namely right slant, left slant, vertical lines, horizontal lines, total length of vertical baselines, total number of left slant lines, total length of horizontal baselines, total number of right slant lines, pen pressure and size from a writing sample and using this information it predicts heart disease and risk factors for heart disease like low blood pressure, and diabetes. The proposed system provides 75 % accuracy.

Higher order neural networks pay more attention due to greater computational capabilities with good learning and storage capacity than the existing traditional neural networks. In this work, a novel attempt has been made for effective optimization of the performance of a higher order neural network (in particular Pi-Sigma neural network) for classification purpose. A newly developed population based teaching learning based optimization algorithm has been used for efficient training of the neural network. The performance of the model has been benchmarked against some well recognized optimized models and they have tested by five well recognized real world bench mark datasets. The simulating results demonstrated favorable classification accuracies towards the proposed model as compared to others. Also from the statistical test, the results of the proposed model are quite interesting than others, which analyzes for fast training with stable and reliable results.

In this paper, the authors have deployed a Social Web Graph over a Semantic Web Framework. A social web graph is a collection of nodes, connected via directed edges. A graph simplifies the search by reducing the sample space by half after every node. But this is a keyword based search and is not rather intelligent. On the other hand a Semantic web framework organizes information in triples in addition to having the benefits of a graph. This makes the information more understandable and easily accessible.

Real-world complex networks usually exhibit inhomogeneity in functional properties, resulting in densely interconnected nodes, communities. Analyzing such communities in large networks has rapidly become a major area in network science. A major limitation of most of the community finding algorithms is the dependence on the ordering in which vertices are processed. However, less study has been conducted on the effect of vertex ordering in community detection. In this paper, we propose a novel algorithm, DIGMaP to identify the invariant groups of vertices which are not affected by vertex ordering. We validate our algorithm with the actual community structure and show that these detected groups are the core of the community.

Video capsule endoscopy (VCE) is a revolutionary imaging technique widely used in visualizing the gastrointestinal tract. The amount of big data generated by VCE necessitates automatic computed aided-diagnosis (CAD) systems to aid the experts in making clinically relevant decisions. In this work, we consider an automatic tissue detection method that uses an adaptive entropy thresholding for better separation of mucosa which lines the colon wall from lumen, which is the hollowed gastrointestinal tract. Comparison with other thresholding methods such as Niblack, Bernsen, Otsu, and Sauvola as well as active contour is undertaken. Experimental results indicate that our method performs better than other methods in terms of segmentation accuracy in various VCE videos.

The support vector machine (SVM) is a most popular tool to resolve the issues related to classification. It prepares a classifier by resolving an optimization problem to make a decision which instances of the training data set are support vectors. Feature selection is also important for selecting the optimum features. Data mining performance gets reduced by Irrelevant and redundant features. Feature selection used to choose a small quantity of related attributes to achieve good classification routine than applying all the attributes. Two major purposes are improving the classification functionalities and reducing the number of features. Moreover, the existing subset selection algorithms consider the work as a particular purpose issue. Selecting attributes are made out by the combination of attribute evaluator and search method using the WEKA Machine Learning Tool. In the proposed work, the SVM classification algorithm is applied by the classifier subset evaluator to automatically separate the standard information set.

This paper deals with software development for controlling EW (Electronic warfare) subsystem through Embedded system using socket programming. MAS receiver is used to monitor, analyze the communication intelligence signals in the scenario which will be used for Defense purpose. The TCP\IP based socket Application Programming Interface (API’s) will be used to realize this application. Vxworks real-time operating system is essential for meeting the time critical application and performance requirement of any system. The RTOS system will also address computational, efficiency and preemptive priority scheduling that is demanded by the user.

Forest fire is one of the natural calamities. The impact of fire on forest soil depends on its intensity and it also affects soil fertility, nutrients, and properties of soil like texture, color, and moisture content. Fire is beneficial and dangerous for soil nutrients based on its strength and it’s duration. In less severity fire, fermentation of soil matter increases nutrients, which results in rapid growth of plants. The natural/artificial biodegradation of soil can rebuild the forest. In the proposed work the effort is made in prediction of possibility of re-plantation in previously forest fire affected area. The analysis and prediction of soil nutrients is done using Naive Bayesian classification.

To secure multimedia content such as digital color images, chaotic map can be used to shuffle or encrypt the original pixel positions of an image. A chaotic map such as Arnold cat map is used to encrypt images as the behavior is periodic and deterministic given the correct key values. Even a small change also cannot predict the behavior and properties of image. To secure digital color images, Arnold cat map is being used to encrypt and send it across public network. Various dimensions of Arnold cat map can be used to achieve greater level of secrecy and confusion. In this paper, our work has been extended to encrypt digital color images using Arnold 4D cat map. Here, the color image has been decomposed to its respective RGB planes. The plane’s pixel positions and gray scale values are then shuffled using Arnold 3D cat map. Then the resulting shuffled encrypted planes (red, green and blue) are divided into 3 parts vertically respectively. The parts of each plane are shuffled in a specific proposed manner with the parts of other planes to achieve greater level of diffusion and secrecy. Each plane holds information of two other planes. The result and analysis shows better result than Arnold 3D cat map used in image encryption.

We proposed a method using Neutrosophic set and data to take the most suitable decision with the help of three different members truth, indeterminacy and falsity. Neutrosophic logic is capable of handling indeterministic and inconsistent information. We have focused to draw a meaning full outcome using neutrosophic concept about the illness of a patient who is suffering from a disease. Fuzzy logic can only handle incomplete information using truth membership value. Vague logic also can handle incomplete information using truth and false membership values. It is an upgrade part of fuzzy and vague concept.

The emerging field of educational data mining gives us better perspectives for insights in educational data. This is done by extracting hidden patterns in educational databases. In these lines, the objective of this research work is to introduce hierarchical clustering models for student’s collected data. The ultimate goal is to find attributes in terms of set of clusters which severely affect the student’s performance. Here clustering is intentionally used as the most common causes affecting performance within the database which cannot be seen normally. The results enable us to use discovered characteristic or patterns in palpating student’s learning outcomes. These patterns can be useful for teachers to identify effective prognostication strategies for students.

The edge detection has become very popular due to its use in various vision applications. Edge detections produces a black and white image where objects are distinguished by lines (either objects boundary comes in black color or white color) depends upon where sharp changes exist. Many techniques have been proposed so far for improving the accuracy of the edge detection techniques. The Fusion of PCA and HUE based edge detector has shown quite better results over the available techniques. But still fusion technique forms unwanted edges so this paper has proposed a new Color and ACO based edge detection technique. The MATLAB tool is used to design and implement the proposed edge detection. Various kinds of images has been considered to evaluate the effectiveness of the proposed technique. Pratt figure of merit and F-measure parameters has been used to evaluate the effectiveness of the available edge detectors.

Sentiment Analysis is becoming a promising topic with the strengthening of social media such as blogs, networking sites etc. where people exhibit their views on various topics. In this paper, the focus is to perform effective Sentimental analysis and Opinion mining of Web reviews using various rule based machine learning algorithms. we use SentiWordNet that generates score count words into one of the seven categories like strong-positive, positive, weak-positive, neutral, weak-negative, negative and strong-negative words. The proposed approach is experimented on online books and political reviews and demonstrates the efficacy through Kappa measures, which has a higher accuracy of 97.4 % and lower error rate. Weighted average of different accuracy measures like Precision, Recall, and TP-Rate depicts higher efficiency rate and lower FP-Rate. Comparative experiments on various rule based machine learning algorithms have been performed through a Ten-Fold cross validation training model for sentiment classification.

Trigonometric Fourier approximation and Lipchitz class of function had been introduced by Zygmund and McFadden respectively. Dealing with degree of approximation of conjugate series of a Fourier series of a function of Lipchitz class Misra et al. have established certain theorems. Extending their results, in this paper a theorem on trigonometric approximation of conjugate series of Fourier series of a function $$ f \in \,Lip\,\left( {\alpha ,r} \right) $$f∈Lipα,r by product summability (E, s) (N, p _n , q _n ) has been established.

As routing protocols very essential for transmission, power consumption is one of the major prevailing issues to find the optimal path. Improve energy performance of MANET Routing protocol by choosing the optimal path which jointly reduces Total Transmission, Power is consuming the packet latency and improves the network lifetime. Propose a new VFO routing based on energy metrics and to reduce the total transmission power, maximize the lifetime of the connection which is implemented using Venus Flytrap Optimization (VFO). The VFO is the novel algorithm devised based on the closure behavior of the Venus Flytrap plant. The VFO algorithm generates the optimal path and energy aware routing.

Digging up the coregulated gene biclusters using a novel Nature-inspired Meta-Heuristic algorithm named Venus Flytrap Optimization (VFO). This optimized biclustering approach will yield highly correlated biclusters. This algorithm is based on the rapid closure behavior of the Venus Flytrap (Dionaea Muscipula) leaves. Gene temperament is understood from their exposure under specific conditions. So far, Optimal Biclusters are extracted using various optimization algorithms like PSO, Genetic algorithm, SA, etc., were used for this kind of analysis. In this paper, VFO algorithm is used for extracting optimal biclusters and results are compared with those obtained by applying PSO, SA, PSO-SA Biclustering algorithms.

Most recently, attention has turned towards Machine learning techniques to predict software development cost as they are more apt when vague and inaccurate information is to be used. Based on the existing evidences, it is proved that a few of the problems associated with previous models are addressed by soft computing techniques. But, the need for accurate cost prediction in software project management is a challenge till today. In this paper, the analytical structure of a Takagi-Sugeno Fuzzy Logic Controller with two inputs and one output for software development effort estimation with a case study on NASA 93 dataset is discussed. The analytical study is also presented with two sample inputs. The Fuzzy models are developed using triangular and GBell membership functions. The results are compared using various assessment criteria. It has been observed that the fuzzy model with triangular membership function performed better than the other models.

Spell checker is a software tool which can detect incorrectly spelled words in a text document. Developing spell checker for a morphologically rich language like Malayalam is really tedious task. This paper mainly discusses about the construction of spell checker for Malayalam language. Since in Malayalam, many words can be derived from root word, it will be impossible to include all the words in a lexicon. So a hybrid method of different techniques can improve the performance of a spell checker. The method explained is an n-gram based approach and which will be inexpensive to construct without deep linguistic knowledge. Along with n gram, a Minimum edit distance algorithm is also added to detect the errors due to addition, deletion, or interchange of letters in a word. This will improve the efficiency of the spell checker. This approach will be useful when less linguistic resources are available for Malayalam language. And also the performance analysis is done with an existing method of spell checking in Malayalam language.

Microarray datasets contain genetic information of patients analysis of which can reveal new findings about the cause and subsequent treatment of any disease. With an objective to extract biologically relevant information from the datasets, many techniques are used in gene analysis. In the paper, the concepts like functional dependency and closure of an attribute of database technology are applied to find the most important gene subset and based on which the samples of the gene datasets are classified as normal and disease samples. The gene dependency is defined as the number of genes dependent on a particular gene using gene similarity measurement on collected samples. The closure of a gene is computed using gene dependency set which helps to know how many genes are logically implied by it. Finally, the minimum number of genes whose closure logically implies all the genes in the dataset is selected for sample classification.

In this work we present a multi criteria based clustering algorithm and demonstrate its usefulness in clustering documents. The algorithm proposes various metrices to judge the veracity of the clusters formed and then finds a near optimal solution that ensures good fitness scores for the all metrices. In view of the complexity of optimizing multiple clustering goals using classical optimization techniques, the paper proposes the use of an evolutionary strategy in the form of Genetic algorithm to quickly find a near optimal cluster set that satisfies all the cluster goodness criteria. The use of Genetic algorithm also inherently allows us to overcome the problem of converging to locally optimal solutions and find a global optima. The results obtained using the proposed algorithm have been compared with the outputs from standard classical algorithms and the performances have been compared.

Association Rule Generation has reformed into an important area in the research of data mining. Association rule mining is a significant method to discover hidden relationships and correlations among items in a set of transactions. It consists of finding frequent itemsets from which strong association rules of the form A => B are generated. These rules are used in classification, cluster analysis and other data mining tasks. This paper presents an extensive approach to the traditional Apriori algorithm for generating positive and negative rules. However, the general approaches based on the traditional support–confidence framework may cause to generate a large number of contradictory association rules. In order to solve such problems, a correlation coefficient is determined and augmented to the mining algorithm for generating association rules. This algorithm is known as the Positive and Negative Association Rules generating (PNAR) algorithm. An improved PNAR algorithm is proposed in this paper. The experimental result shows that the algorithm proposed in this paper can reduce the degree of redundant and contradictory rules, and generate rules which are interesting on the basis of a correlation measure and dual confidence approach.

Keyword extraction is an important phase in automatic text summarization process because it directly affects the relevance of the system generated summary. There are many procedures for extracting keywords, but all of these aim to find the words that directly represent the topic of the document. Identifying lexical association between terms is one of the existing techniques proposed for determining the topic of the document. In this paper, we have made use of lexical association and graph based ranking techniques for retrieving keywords from a source text and subsequently to assign them a relative weight. The individual weights of the extracted keywords are used to rank the sentences in the source text. Our summarization system is tested with DUC 2002 dataset and is found to be effective when compared to the existing context based summarization systems.

This manuscript prepares a modern hybrid technique by combining the usual particle swarm optimization (PSO) technique with differential evolution (DE) technique, named as fusion based differential evolution particle swarm optimization (DEPSO) in order to enhance the global search abilities, while solving impulse response of linear phase digital band stop filter. The linear phase band stop filter is mostly considered as a nonlinear multimodal problem. The DEPSO is a heuristic based global search method modelled on considering the advantages of both of the methods to enhance the superiority of result and convergence speed. The performance of the intended DEPSO based approach has been contrasted with few renowned optimization techniques such as PSO, comprehensive learning PSO (CLPSO), craziness based PSO (CRPSO) and DE. The proposed DEPSO based result confirms the supremacy of solving design problems of FIR filters.

Today surveillance systems are used widely for security purposes to monitor people in public places. A fully automated system is capable of analyzing the information in the image or video through face detection, face tracking and recognition. The face detection is a technique to identify all the face in the image or video. Automated facial recognition system identifies or verifies a person from an image or a video by comparing features from the image and the face database. When surveillance system is used to monitor human for locating or tracking or analyzing the activities, the challenge of identification of a person is really a hard task. In this paper we survey the techniques involved in face detection and person re-identification.

Inflections in Telugu plays a crucial role in maintaining the meaning of the sentence unchanged though the word order is changed. This is possible since inflections are the internal part of the word. Inflecting nouns and pronouns is customary in Telugu. Main attention is given to inflected pronoun translation in this paper. Various categories of inflections and the ways of translation of pronouns are discussed in detail. For analysis, identification and translation of pronoun inflections, a Morphological Analysis System (MAS) that can perform forward and reverse morphology is developed for Machine Translation (MT) (Telugu (Source Language—SL) to Sanskrit (Target Language—TL)).

Sentiment analysis is the process of finding the opinions present in the textual content. This paper proposes a tweet analyzer to perform sentiment analysis on twitter data. The work mainly involves the sentiment analysis process using various trained machine learning classifiers applied on large collection of tweets. The classifiers have been trained using maximum number of polarity oriented words for effectively classifying the tweets. The trained classifiers at sentence level outperformed the keyword based classification method. The classified tweets are further analyzed for identifying top N tweets. The experimental results show that the sentiment analyzer system predicted polarities of tweet and effectively identified top N tweets.

Wireless sensor networks are collection of spatially distributed autonomous actuator devices called sensor nodes. Tracking target under surveillance is one of the main applications of wireless sensor networks. It enables remote monitoring of objects and its environments. In target tracking, sensor nodes are informed when the target under surveillance is discovered. Some nodes detect the target and send a detection message to the nodes on the targets expected moving path. So nodes can wake up earlier. Face tracking is a new tacking framework, in which divides the region into different polygons called Faces. Instead of predicting the target location separately in a face, here estimate the targets movement towards another face. It enables the wireless sensor network to be aware of a target entering the polygon a bit earlier. Face track method failed when target in dynamic motion, i.e. target in random motion or retracing the path again and again. If the target follows a dynamic motion, the polygons are reconstructed repeatedly and energy is wasted in sending messages to create new polygons. Here proposes a framework to track a mobile object in a sensor network dynamically. In this framework, Polygons are created initially in form of clusters to avoid repeated polygon reconstruction. The sensors are programmed in a way that at least one sensor stays active at any instant of time in the polygon to detect the target. Once the target is detected and it entered into the edge of polygon, the nodes of the neighboring polygon is activated. Then, activated polygon keep tracking the target. Thus energy of the sensor nodes are saved because of polygon is not created and deleted, instead activated.

Gait Recognition plays a major role in Person Recognition in remote monitoring applications. In this paper, a new gait recognition approach has been proposed to recognize the normal/abnormal gait of the person with improved feature extraction techniques. In order to identify the abnormality of the person lean angle and ramp angle has been considered as features. The novelty of the paper lies in the feature extraction phase where the walking abnormality is measured based on foot movement and lean angle which is measured between the head and the hip. In this paper, the feature vector is composed of measured motion cues information such as lean angle between the head and the hip and ramp angle between the heel and the toe for two legs. These features provide the information about postural stability and heel strike stability respectively. In the training phase, classifier has been trained using normal gait sequences by exemplar calculation for each video and distance metric between the sequences separately for samples. Based on the distance metric, the maximum distance has been calculated and considered as threshold value of the classifier. In the testing phase, distance vector between training samples and testing samples has been calculated to classify normal or abnormal gait by using threshold based classifier. Performance of this system has been tested over different data and the results seem to be promising.

The paper presents an approach to rank various Cloud-Service providers (CSPs) on the basis of four parameters, namely, Reliability, Performance, Security and Usability, using a combination of Likert scale, Fuzzy logic and Analytic Hierarchy Process (AHP). The CSPs have been ranked using data collected through a survey of cloud computing users. The purpose of this paper is to help provide a technique through which various CSPs can be ranked on the basis of a number of parameters keeping in mind the users’ perceptions. We have chosen fuzzy logic because it has the capability to model the vague and imprecise information provided by the users and can handle the uncertainty involved. Similarly, AHP has been chosen because of its ability to make complex decisions on the basis of mathematics and psychology. The proposed model derives its strength from the fact that it takes into account the uncertainty and dilemma involved in decision-making.

The paper proposes classification of tabla strokes using multilayer feed forward artificial neural network. It uses 62 features extracted from the audio file as input units to the input layer and 13 tabla strokes as output units in the output layer. The classification has been done using dimension reduction and without using dimension reduction. The dimension reduction has been performed using Principal Component Analysis (PCA) which reduced the number of features from 62 to 28. The experiments have been performed on two sets of tabla strokes, which are played by professional tabla players, each comprises of 650 tabla strokes. The results demonstrate that correct classification of instances is more than 98 % in both the cases.

Newspaper plays a significant role in our day to day life. In a news article, readers are attracted towards headline. Headline creation is very important while preparing news. It is a tedious work for journalists. Headline needs maximum text content in short length also it should preserve grammar. The proposed system is an automatic headline generation for news article. There are several methods to generate a headline for news. Here the headline is generated using n-gram model. After pre-processing, extract all possible bi-grams from the news. Frequency of bi-grams in the document is found out and sentence is identified. Select all sentences in which the most frequent bi-gram appears. After that rank the sentences based on the locality of the sentence in the document. Then highly ranked sentence will be selected. Finally using some phrase structural rules, new headlines are generated.

Community detection in social networks plays a vital role. The understanding and detection of communities in social networks is a challenging research problem. There exist many methods for detecting communities in large scale networks. Most of these methods presume the predefined number of communities and apply detection methods to exactly find out the predefined number of communities. However, there may not be the predefined number of communities naturally occurring in the social networks. Application of brute force inorder to predefine the number of communities goes against the natural occurrence of communities in the networks. In this paper, we propose a method for community detection which explores Self Organizing Maps for natural cluster selection and modularity measure for community strength identification. Experimental results on the real world network datasets show the effectiveness of the proposed approach.

Prediction of future trends in financial time-series data related to the stock market is very important for making decisions to make high profit in the stock market trading. Typically the economic time-series data are non-linear, volatile and many other factors crash the market for local or global issues. Because of these factors investors find difficult to predict consistently and efficiently. The motive of designing a framework for predicting time series data is by using a low complexity, adaptive functional link artificial neural network (FLANN). The FLANN is basically a single layer structure in which non-linearity is introduced by enhancing the input pattern. The architecture of FLANN is trained with Meta-Heuristic Firefly Algorithm to achieve the excellent forecasting to increase the accurateness of prediction and lessen in training time. The projected framework is compared by using FLANN training with conventional back propagation learning method to examine the accuracy of the model.

Cloud computing is the most recent computing paradigm, in the Information Technology where the resources and information are provided on-demand and accessed over the Internet. An essential factor in the cloud computing system is Task Scheduling that relates to the efficiency of the entire cloud computing environment. Mostly in a cloud environment, the issue of scheduling is to apportion the tasks of the requesting users to the available resources. This paper aims to offer a genetic based scheduling algorithm that reduces the waiting time of the overall system. However the tasks enter the cloud environment and the users have to wait until the resources are available that leads to more queue length and increased waiting time. This paper introduces a Task Scheduling algorithm based on genetic algorithm using a queuing model to minimize the waiting time and queue length of the system.

The summary of the paper explains the optimal tuning of integral (I) and proportional integral (PI) controllers are applied to closed loop standalone integrated multi wind energy system by using particle swarm optimization. Tuning of I and PI controller gain values obtained from the optimization techniques to get the best possible operation of the system. For the optimal performance of the integrated wind energy system, the controller gains are tuned by using the PSO and genetic algorithms (GA). The system harmonics of voltage responses are observed with search heuristic algorithm that is nothing but a genetic algorithm. Similarly the system responses are observed and compared with PSO algorithm, and the PSO algorithm is proved better. The results establishes the proposed new stand alone multi wind energy system with I, PI controller gains are tuned by using PSO will gives less harmonic distortion and improves performance. The proposed system is developed in MATLAB/SIMULINK.

Network security is a weak link in wired and wireless network systems which if breeched then the functionality of the underlying network is impaired by malicious attacks and causes tremendous loss to the network. The DoS (Denial of Service) attacks are one of the harmful network security threats. Ad Hoc networks can be easily damaged by such attacks as they are infrastructure less and without any centralized authority. This paper narrates results of an undertaken research work to study the consequence of selected DoS attacks on the performance of Ad Hoc networks. The gained knowledge is used to design and develop a security framework to detect intrusion and perform response actions. The validations of proposed defense system are carried out using statistical approach. All Simulations were carried out in NS-2.

Optical Character Recognition (OCR) System for ancient epigraphs helps in understanding the past glory. The system designed here, takes a scanned image of Kannada epigraph as its input, which is preprocessed and segmented to obtain noise-free characters. Fourier features are extracted for the characters and used as the feature vectors for classification. The SVM, ANN, k-NN, Naive Bayes (NB) classifiers are trained with different instances of ancient Kannada characters of Ashoka and Hoysala period. Finally, OCR system is tested on epigraphical characters of 250 from Ashoka and 200 from Hoysala period. The prediction analysis of SVM, ANN, k-NN and NB classifiers is made using performance metrics such as Accuracy, Precision, Recall, and Specificity.

Spelling correction has been haunting humans in various fields of society like while creating business proposals, contract tenders, students doing their assignments, in email communications, sending request for proposals, while writing content for website and so on. Already existing Dictionary based correction approaches have helped by providing solution to the problem when the written word doesn’t even qualify to be called as a legal word. But is it the only challenge a writer faces while writing the desired documents! The words, which fall in the category of correct spelling words, may sometimes be the word which writer did not intend to write. The above illustrated genre of errors is called Real-Word error. This paper proposes a spelling correction system whose main focus is on automatic identification and correction of real word errors accurately and efficiently. The approach includes hybridization of Trigram and Bayesian approach and using Modified Brown corpus as a training set. A large set of commonly confused words is used in this case for evaluating the performance of the proposed approach.

This paper discusses differential evolution algorithm for tuning of proportional integral controller in modular multilevel converter based STATCOM applications. Unlike conventional VSC based converters, an MMC is known for its distinctive features such as modularity, low harmonic content and flexibility in converter design. The MMC-STATCOM is capable of reactive power compensation, simultaneous load balancing and harmonic cancellation. Differential evolution algorithm is used to tune the proportional integral controller of STATCOM. The proposed model is verified in MATLAB/Simulink and the results are well in proximity with the theoretical analysis.

Crypto circuits can be attacked using the technique of differential power analysis by another/separate party, using power consumption dependence on secret message/information for extracting the critical data (information). To avoid DPA (Differential Power Analysis) and security bases, differential logic styles are basically used, because of constant power dissipation. This paper proposes a new design methodology using OR-NOR gates in 90 nm VLSI technology using SABL (Sense Amplifier Based Logic) for DPDN (Differential Pull down Logic) to secure/protect differential logic gates and to eliminate charge in the pull-down differential gate and to remove the memory effect.

A typical web user imposed small and vague queries onto web based search engines, which requires higher time for query formulation. In this paper, a nature inspired optimization approach on term graph is employed in order to provide query suggestion by assessing the similarity. Term graph is simulated according to the pool of relevant documents of user query. The association among terms graphs is based on similarity and will be act as fitness values for genetic algorithm (GA) approach, which converges by deriving query reformulations and suggestions. Each user interactions with the search engine is a considered as an individual chromosome and larger pool help in convergence for significant reformulations. Proposed algorithmic solution select optimal path and extracts the most relevant keywords for an input search query’s reformulation. The query user will select one the suggested reformulated query or query terms. The optimization performance of the proposed method is illustrated and compared with different optimization techniques, e.g. ACO, PSO, ABC.

Previous study has shown the superiority of TCP Vegas over TCP Reno in wired and wireless network. However, when both Reno and Vegas co-exist on a wired link, protocol Vegas dominated by Reno due to its conservative nature. However, previous study does not include all the issues which might be arising during communication in the network. This paper finds the performance (throughput and fairness) of a network shared by Reno and Vegas in static multi hop wireless ad hoc network focusing on four issues. Here we introduced a new parameter called PST (Percentage Share of Throughput) which measures fairness among different connections shared by a common link. The issues are: (1) Both Reno and Vegas flows in forward direction, (2) Reno flows in forward and Vegas flows in reverse direction, (3) Both Reno and Vegas flows in forward direction along with bursty UDP traffic in background, (4) Both Reno and Vegas flows in forward direction where receiver is enabled with delayed acknowledgement (DelAck) schemes. Our result shows that Reno and Vegas are more compatible (or fairer) when DSDV protocol is used and their PST is much closer to each other (around 50 %). For increased hop length, DSDV is better routing protocol for all issues we have considered. However, use of DelAck techniques improves the throughput than others.

This paper introduces a model that uses a rule based technique to determine the pronunciation of numerals in different context and is being integrated with the waveform concatenation technique to produce speech from the input text in Odia language. To analyze the performance of the proposed technique, a set of numerals are considered in different context and a comparison of the proposed technique with an existing numeral reading algorithm is also presented to show the effectiveness of the proposed technique in producing intelligible speech out of the entered text.