Proceedings of the 5th International Conference on Frontiers in Intelligent Computing: Theory and Applications

Human action recognition has gained popularity because of its wide applicability in automatic retrieval of videos of particular action using visual features. An approach is introduced for human action recognition using trajectory-based spatiotemporal descriptors. Trajectories of minimum Eigen feature points help to capture the important motion information of videos. Optical flow is used to track the feature points smoothly and to obtain robust trajectories. Descriptors are extracted around the trajectories to characterize appearance by Histogram of Oriented Gradient (HOG), motion by Motion Boundary Histogram (MBH). MBH computed from differential optical flow outperforms for videos with more camera motion. The encoding of feature vectors is performed by bag of visual features technique. SVM with nonlinear kernel is used for recognition of actions using classification. The performance of proposed approach is measured on various datasets of human action videos.

In this study, response surface methodology (RSM)-based prediction model was prepared for specific fuel consumption (SFC) as a response. A regression model was designed to predict SFC using RSM with central composite rotatable design (CCRD). In the development of regression models, injection timing, compression ratio, injection pressure, and engine load were considered as controlled variables. Injection pressure and compression ratio were observed as the most influencing variables for the SFC. The predicted SFC values and the succeeding verification experiments under the optimal conditions established the validity of the regression model.

Hereditary traits are anticipated by the mutations in the gene sequences. Identifying a disease based on mutations is an essential and challenging task in the determination of genetic disorders such as Muscular dystrophy. Silent mutation is a single nucleotide variant does not result in changes in the encoded protein but appear in the variation of codon usage pattern that results in disease. A new ensemble learning-based computational model is proposed using the synonymous codon usage for identifying the muscular dystrophy disease. The feature vector is designed by calculating the Relative Synonymous Codon Usage (RSCU) values from the mutated gene sequences and a model is built by adopting codon usage bias pattern. This paper addresses the problem by formulating it as multi-classification trained with feature vectors of fifty-nine RSCU frequency values from the mutated gene sequences. Finally, a model is built based on ensemble learning LibD3C algorithm to recognize muscular dystrophy disease classification. Experiments showed that the accuracy of the classifier shows 90%, which proves that ensemble-based learning, is effective for predicting muscular dystrophy disease.

Data compression plays an important role for storing and transmitting text or multimedia information. This paper refers to a lossless data algorithm is developed in C-platform to compress character string based on Basic Arithmetic Coding. At the preliminary stage, this algorithm was tested for the character array comprising of vowels only and the probability distribution is assumed arbitrarily. The result being obtained is encouraging with compression ratio far beyond unity. Though the algorithm was tested for vowels only but the work can be extended for any character array with probability of distribution as obtained from the survey of few randomly selected articles.

Malwares have been rising in drastic extent as Android operating system enabled smart phones and tablets getting popularity around the world in last couple of years. For efficient detection of Android malwares, different static and dynamic malware detection methods have been proposed. One of the popular methods of static detection technique is permission/feature-based detection of malwares through AndroidManifest.xml file using machine learning classifiers. But ignoring important feature or keeping irrelevant features may specifically cause mystification for classification algorithms. So to reduce classification time and improvement of accuracy different feature reduction tools have been used in different literature. In this work, we have proposed a framework that extracts the permission features of manifest files, generates feature vectors and uses six different feature ranking tools to create separate feature reducts. On those feature reducts different machine learning classifiers of Data Mining Tool, Weka have been used to classify android applications. We have evaluated our method on a set of total 734 applications (504 benign, 231 malwares) and results show that highest TPR rate observed is 98.01% while accuracy is up to 87.99% and highest F1 score is 0.9189.

In this paper, a novel approach is proposed that cope with challenges such as illuminations, expressions, poses, and occlusions. The proposed methodology is a non-domination-based optimization technique with a semi-supervised classifier for recognizing a known and unknown face based on different scenarios. The classification is a robust method attaining aptness at different stages resulting in identification of proper training set with actual face image. Different datasets Yale Face Database, Extended Yale Face Database B, ORL database has been considered for our experiments. The performance of the proposed method has been evaluated on several grounds. Results show that the proposed method attains a better performance than the statistical methods.

The algorithm presented in this paper deals with use of soft computing technique of Fuzzy logic applied with dynamic graph theory to create graphs which can be efficient in resource allocation process in varied environments, i.e., software project management, operating systems, construction models, etc. The algorithm implies one unique factor of dynamicity which makes graph of resource allocation evolving even after primary design due to chaotic nature of the afore mentioned nature of environments. The use of Fuzzy imparts a logical inference mechanism which rules out non-monotonous reasoning perspective of this dynamicity. The algorithm is robust and adaptive to varied environments. The proposed algorithm will be beneficial for more accurate Engineering in terms of reducing the failures and being more specific in answering the allocation of the resources and how the work has to be undertaken using those resources. It will also emphasize on devising a model which can be adhered to with the proper follow ups such that it could be referred to at the time of chaos or failures. “The development of the Algorithm will be much more product centric and will stick to developer’s view of development along with customer’s view of required functionalities.”

Cognitive Radio Networks (CRNs) is an intelligent wireless communication network that senses its environment to adjust the transmitter parameters in order to exploit the unused portions of available spectrum. The objective here is to ensure reliable communication with minimum intereference to Primary Users (PUs) and efficient spectrum utilization. The spectrum assigned to licensed users is underutilized and the growing demand causes starvation to the unlicensed users. Thus, CRN senses the available spectrum to find the most appropriate spectrum for allocation. Further, to maximize the efficient use of available spectrum, agreement (consensus) may be used wherein all users agree on a common decision value. In the paper, we discuss various techniques of spectrum allocation in CRN. Lastly, we propose an interference-aware protocol that achieves load balancing, high throughput and less number of reallocations to maximize spectrum utilization. Also, the paper validates the proposed algorithm using the simulation results.

Cloud computing is an internet based technology that provisions the resources automatically on the pay per use basis. With the development of cloud computing, the amount of customers and requirement of resources increases exponentially. In order to balance the load, the tasks must be equally distributed among multiple computing servers thereby, fulfilling Quality of Service (QoS) with maximum profit to cloud service providers. In addition, cloud servers consume huge amount of electrical energy leading to increased expenditure and environment degradation. Therefore, certain solutions are needed that results in efficient resource utilization while minimizing the environmental influence. In the paper, we present a survey of load balancing algorithms along with their limitations and propose a framework for an energy efficient resource allocation and load balancing for heterogeneous workload in cloud computing along with the validation of the framework using CloudSim toolkit.

Accent and Emotion recognition for speech has become most important research area because of the increased demand of speech processing systems in handheld devices. Most of the research in speech processing is done for the English language only. In this paper, we present accent recognition system for Telugu speeches. Three important accents of Telugu were chosen and text-dependent speeches of Coastal Andhra, Rayalaseema, and Telangana accents were collected. Features like tonal power ratio, spectral flux, pitch chroma, and MFCC were extracted from these speeches. deep belief networks are used for the classification purpose. The recognition accuracy obtained in this work is 93%.

We propose a document classifier based on principal component analysis (PCA) and one-class support vector machine (OCSVM), where PCA helps achieve dimensionality reduction and OCSVM performs classification. Initially, PCA is invoked on the document-term matrix resulting in choosing the top few principal components. Later, OCSVM is trained on the records of the matrix corresponding to the negative class. Then, we tested the trained OCSVM with the records of the matrix corresponding to the positive class. The effectiveness of the proposed model is demonstrated on the popular datasets, viz., 20NG, malware, Syskill, & Webert, and customer feedbacks of a Bank. We observed that the hybrid yielded very high accuracies in all datasets.

This paper presents the experimental analysis of data provided by UCI machine learning repository. Weka open source machine learning tool provided by Waikato University reveals the hidden fact behind the datasets on applying supervised mathematical proven algorithm, i.e., J48 and Naïve Bayes algorithm. J48 is an extension of ID3 algorithm having additional features like continuous attribute value ranges and derivation of rules. The data sets were analyzed using two approaches, i.e., first taken with selected attributes and taken with all attributes. The performance of both the algorithm reveals the accuracy of algorithm and predicting the various reasons behind this increasing problem of cardiovascular diseases.

This paper presents a two-stage approach consisting of a real-coded genetic algorithm and goal programming to obtain improved cell formation. In the first stage, the minimum value of each objective is determined using a single-objective genetic algorithm. In the second stage, goal programming is incorporated and the final objective is constructed as the minimization of sum of deviational variables of corresponding objectives. The proposed technique is implemented as a software toolkit using C Sharp.net programming language. Modified grouping efficiency is used as the performance measure to test the efficiency of the proposed technique. Five problems with different sizes have been considered from the literature to show the potentials of the proposed technique.

With the advent of Web 2.0, user-generated content is led to an explosion of data on the Internet. Several platforms such as social networking, microblogging, and picture sharing exist that allow users to express their views on almost any topic. The user views express their emotions and sentiments on products, services, any action by governments, etc. Sentiment analysis allows quantifying popular mood on any product, service or an idea. Twitter is popular microblogging platform, which permits users to express their views in a very concise manner. In this paper, a new framework is crafted which carried out the entire chain of tasks starting with extraction of tweets to presenting the results in multiple formats using an ETL (Extract, Transform, and Load) big data tool called Talend. The framework includes a technique to quantify sentiment in a Twitter stream by normalizing the text and judge the polarity of textual data as positive, negative, or neutral. The technique addresses peculiarities of Twitter communication to enhance accuracy. The technique gives an accuracy of above 84% on standard datasets.

Due to the rapid growth of digital gadgets with various screen sizes, resolutions and hardware processing capabilities, robust video retargeting is of increasing relevance. An efficient retargeting algorithm should not only retain semantic content, but also maintain spatiotemporal resolution of video data. In this paper, the effective seam carving technique for content-aware video retargeting is discussed. Retargeting video is of immense importance as it is frequently played on several gadgets such as television, mobile, tablet, and notebook. The proposed method considers each video frame as an independent image entity and tries to resize it. Our main contribution is a formulation of seam carving using graph cut method. Convention cut techniques fail to defend a meaningful seam. Single monotonic well connected by pixel to pixel is most desirable property in seam carving process. The traditional seam carving method is designed to work based on the minimum energy concept, while ignoring the energy that has been introduced by the operator. To address this issue, we propose a new design criterion in which least amount of energy is introduced in retargeted video.

In today’s world, digital communication plays a vital role. The art of communicating secret information has also evolved. For years, encryption has played a vital role in secure transmission of secret data. But due to lack of covertness, an eavesdropper can identify encrypted data and subject it to cryptanalysis. Here we present a method of hiding information such that its very existence is masked. This study aims to develop an enhanced technique for hiding data in 3-D images ensuring high invisibility. The data will be embedded in 3-D images with .pcd format and the coordinates of the 3-D image are used for data hiding.

The act of network intrusion detection is an obligatory part of network performance under security. Unlike other network security strategies, the act of intrusion detection systems should aware the behavior of the users and signature of the intruded and normal transactions, which is continuous process since the user behavior is not static as well the attack strategies are redefining in magnified speed. Hence, the objective of effective intrusion detection is always a significant factor for research. The bioinspired evolutionary strategies are getting the attention of most of the recent research studies. In order to this, the divergent contexts such as minimal computational complexity, prediction accuracy, ensemble models have been considered as significant objective. The other most significant objective and compatible to current state of art is IDS scalability and robustness in high-speed networks, hence the evolutionary computation approaches are adoptable. In this study, we propose an intrusion detection approach that is based on evolutionary computation technique called Cuckoo search. Further, the proposed detection system is investigated thoroughly in the context of accuracy, robustness, and also from the evolutionary computation point of view.

The waste generated by health care units has been contributing a dreadful share in terms of life threatening diseases and environmental pollution. Erroneous management of this waste has not only invited a serious threat to the environment but also to the personnel associated with it; mainly health care experts, patients, workers as well as the general community. A number of studies advocate that there exists certain factors that inhibit effectiveness of health care waste management (HCWM). Prior knowledge of these factors and their relative importance will be helpful for decision makers to better handle these barriers and improve HCWM effectiveness. This research, through the employment of gray relational analysis (GRA) prioritizes 14 barriers identified from literature, according to the degree of their negative impact. The study reveals that “Unauthorized Reuse of Health Care Waste” and Implementation of “Poor Segregation Practices” ranked 1 are perceived as the two most significant barriers while “Lack of Accountability of Authorities of Health Care Facilities towards HCWM” and “Inadequate Awareness and Training Programs” ranked 5 are perceived as the least important barriers of HCWM in India.

Sharing of data provides mutual benefits for collaborating organizations. Data mining techniques have allowed regimented discovery of knowledge from huge databases. Conversely, in the case of sharing the data with others, knowledge discovery raises the possibility of revealing the sensitive knowledge. The need of privacy prompted the growth of numerous privacy-preserving data mining techniques. In order to deal with privacy concerns, the database is to be transformed into another database in such a way that the sensitive knowledge is concealed. One subarea of privacy-preserving data mining, which got attention in retail businesses, is privacy-preserving association rule mining. A significant feature of privacy-preserving association rule mining is attaining a balance between privacy and precision, which is characteristically conflicting, and refining the one generally reduces the other one. In this paper, the problem has been planned in the perspective of protecting association rules which are sensitive by prudently amending the transactions of the database. To moderate the loss of non-sensitive association rules and to improve the quality of the transformed database, the proposed approach competently estimates the impact of an alteration to the database. The proposed method selects the transactions for alterations using the binary TLBO optimization technique during the concealing process. Experimental outcomes exhibit the efficiency of the proposed algorithm.

In the recent research, compressive sampling (CS) has received attention in the area of signal processing and wireless communications for the reconstruction of signals. CS aids in reducing the sampling rate of received signals thereby decreasing the processing time of analog-to-digital converters (ADC). The energy minimization is the key feature of CS. In this work, CS has been applied to spectrum sensing in cognitive radio networks (CRN). The primary user (PU) signal is optimally detected using the sparse representation of received signals. The received PU signal is compressed in the time domain to extract the minimum energy coefficients and then applied to sensing. Further, the signal is detected using energy detection technique and recovered using $$l_{1}$$-minimization algorithm. The detection performance for various compression rates is analyzed.

Digital image watermarking is widely used for enforcing copyright protection and authentication. Color image watermarking has become essential as most of the images used are colored. A novel multiple composite digital image watermarking technique for color images is proposed in this paper. We have exploited the high embedding capacity property of Least Significant Bit (LSB) technique. It is more robust technique of watermarking multiple images in a single color image. Three different binary watermarks are taken as three channels and are combined to form single composite color watermark. The composite color watermark is further embedded in the color image using LSB technique. Simulation results exhibit that our proposed method has higher PSNR values indicating good visual quality of watermarked image. Experimental results show that the proposed scheme is robust under signal processing and geometric attacks.

In order to predict foreign exchange (FOREX) rates, this paper proposes a new hybrid forecasting approach viz., Chaos+MARS involving chaos theory and multivariate adaptive regression splines (MARS). Chaos theory aims at constructing state space from the given exchange rate data with the help of embedding parameters, whereas MARS aims at yielding accurate predictions using state space constructed. The proposed model is tested for predicting three major FOREX Rates- JPY/USD, GBP/USD, and EUR/USD. The results obtained unveil that the Chaos+MARS yields the accurate predictions than other chaos-based hybrid forecasting models and recommend it as an alternative approach to FOREX rate prediction.

The vital goal of the image compression is to abate the insignificant facts of the image. Image compression is acclamatory while uploading and downloading images over the web. Image compression is the concept to compress the multifarious hyper spectral images, landsat images, multispectral images while maintaining the quality of an image and preventing the noise. The prime aim is to get the compressed image with improved radiometric resolution. Existing approaches are also efficient but still suffers from ringing artifacts. So an efficient technique for compressing the grayscale images is introduced. The proposed approach used particle swarm optimization (PSO), discrete wavelet transform (DWT), and guided image filter (GF). The idea behind the proposed technique is to apply PSO on the DWT along with GF to diminish the ringing artifacts, Gaussian noise and improve the radiometric resolution of the images. The overall result shows that proposed technique has improved radiometric information and lesser ringing artifacts than existing methods.

The analysis of large graph plays a prominent role in various fields of research and application area. Initially, we formally define the partitioning scheme based on user needs and requirements. In this paper, we will be dealing with various methods of graph partitioning, its advantages and disadvantages, and from the result we can conclude which is the most effective method of graph partitioning. We can apply the best method in road navigation, stock market, database modeling, and bioinformatics.

The Water Distribution Management System is an intuitive approach to eradicate the shortage of water in remote areas by providing requisite amount of water to each and every household by virtue of their uniquely generated water cards. In this paper, the main emphasis is given on the fact that the existing manual water collection systems leading to improper water distribution are replaced to the point where the services provided are efficient and cost effective leading not only to eliminate water scarcity but to encourage water conservation as well. Each household is entitled to a certain amount of water per day and the process of water collection is scheduled automatically. A feedback system has also been incorporated to evaluate the quality of water being supplied.

For maintaining power system stability, several parameters like voltage, frequency, etc. are monitored sequentially at regular intervals by SCADA, and the informations are transmitted to data centre through suitable communication schemes. If the volume of data can be reduced, then it is possible to reduce the energy and space requirement. This paper emphasizes on the development of an algorithm to compress the monitoring parameters using Adaptive Huffman Coding in MATLAB environment. The compression ratio obtained by this approach is better than what is obtained by other data compression techniques. This results in the reduction of memory requirement by about 60%, thereby enabling it suitable for the data handling of a large volume of monitoring data encountered frequently in a power system.

Image encryption is one of the most promising fields of research in the conventional scientific society. Visual cryptography is a secured encryption technique which is used to encrypt a secret image based on share generation and superimposition rather than computing. This overcomes the burden of computation but the mammoth risk of attackers is superfluous to the passing of shares in sequence through the communication channel. However, this superimposing problem can be resolved by cracking some additional encryption algorithms alongside the visual cryptography. In this paper, we propose a highly secured visual cryptography scheme which uses encryption and error diffusion halftoning algorithms as intermediate steps in cryptography work. We have also done the comparative analysis in order to select the most optimum technique, among the available algorithms according to the requirement of the system. The proposed work has been tested on various formats of standard color images of varied resolutions and proven more secured than contemporary techniques.

The photovoltaic cell produces pollution less electricity. It requires almost no maintenance and has long lifespan. Nowadays, the photovoltaic is one of the most promising markets in the world because of these advantages. This paper demonstrates the dynamic model of single-stage three-phase impedance source inverter or Z-source inverter connected to grid. Here ZSI connected to PV is analyzed and designed. As the output of the PV array is very low, in order to commercialize and utilize this, the output voltage must be increased. So to boost up the voltage, Z-source inverter (ZSI) is used instead of VSI or CSI. Different pulse width modulation (PWM) techniques are used to provide pulses for PV connected Z-source converter (ZSI). After this, the final model is simulated using MATLAB/SIMULINK, and THD related to different output waveforms are analyzed for different parameters used.

Induction motor was very often used machine in industries. Recent developments in electronics led to use of induction motor-driven electrical vehicles. Faults in inverter-fed induction motor can lead to unusual operation. The knowledge of faults in inverter circuit is as important. Fault identification is as much important as prior knowledge to mitigate the faults that might occur in inverter-driven induction motor drives. In this paper, five-level multi-level inverter was taken up for testing of mitigation using redundant cell. In this simulation model only two fault cases are considered for voltage source inverter (VSI)-fed squirrel cage induction motor drive. Those two cases are switch open fault and switch short fault. In this work redundant cell comes into operation when switch is open fault, short fault in drive system. The Matlab/Simulink-based model clearly explains the effect of adding redundant cell during fault.

A system has been designed and implemented to remotely configure generic automated systems using only conditional logic statements and a web-based application for intuitive user interaction. Current automation systems use company-specific applications and require manual configuring in order to make any changes as per the user’s needs. The purpose of this system is to give users the ability to self-configure and control all their systems from anywhere, using a simple internet-based application, thus requiring no external intervention. Our system is built using open-source software and hardware, thus rendering it cost-effective to implement in real time. Conditional programming, or If-then-Else logic, has been used as it is an easily understandable logic construct to maintain and configure the automated systems.

A new digital image watermarking scheme based on chaotic map was proposed to hide the sensitive information known as watermark. The authors claimed that the scheme is efficient, secure, and highly robust against various attacks. In this paper, the inherent security loopholes of the watermarking embedding and extraction processes are unveiled. The cryptanalysis of watermarking scheme is presented to demonstrate that the scheme is not robust and secure against the proposed attack. Specifically, with chosen host image and chosen watermarks, the successful recovery of securely embedded watermark from received watermarked image is possible without any knowledge of secret key. The simulation analysis of proposed cryptanalysis is provided to exemplify the proposed attack and lack of security of anticipated watermarking scheme.

This paper addresses the issue of distinguishing influential nodes in the complex network. The k-shell index features embeddedness of a node in the network based upon its number of links with other nodes. This index filters out the most influential nodes with higher values for this index, however, fails to discriminate their scores with good resolution, hence results in assigning same scores to the nodes belonging to same k-shell set. Extending this index with neighborhood coreness of a node and also featuring topological connections between its neighbors, our proposed method can express the nodes influence score precisely and can offer distributed and monotonic rank orders than other node ordering methods.

Classification of text documents from a pool of huge collection of the same is performed usually on the basis of certain key terms present in the said documents that distinguish a particular document set from the universal set. Generally, these key terms are identified using some feature sets, which can be statistical, rule-based, linguistic, or hybrid in nature. This paper develops a simple technique based on Venn diagram to prioritize the different standard features available in the literature, which in turn reduces the dimension of the feature sets used for document classification.

In recent times, holistic word recognition has achieved enormous attention from the researchers due to its segmentation-free approach. In the present work, a holistic word recognition method is presented for the recognition of handwritten city names in Bangla script. At first, each word image is hypothetically segmented into equal number of grids. Then gradient-based features, inspired by Histogram of Oriented Gradients (HOG) feature descriptor, are extracted from each of the grids. For the selection of suitable classifier, five well-known classifiers are compared in terms of their recognition accuracies and finally the classifier Sequential Minimal Optimization (SMO) is chosen. The system has achieved 90.65% accuracy on 10,000 samples comprising of 20 most popular city names of West Bengal, a state of India.

In social media directed links can represent anything from close friendship to common interests. Such directed links determine the flow of information and hence indicate an individual influence on others. The influence of a person X over person Y is defined as the ratio of Y’s investment that Y makes on X. Most contemporary networks return source–target paths in an online social network as a result of search ranked by degrees of separation. This approach fails to reflect tie of social strength (i.e., intimacy of two people in terms of interaction), and does not reflect asymmetric nature of social relations (i.e., if a person X invests time or effort in person Y, then the reverse is not necessarily true). In this paper, it is proved that in social graph result can prove to be more effective by incorporating the concept of directed and weighted influence edges taking into account both asymmetry and tie strength. The study is based on two real-world networks: Twitter capturing its retweet data and DBLP capturing its author–coauthor relationship. The experiments have been conducted based on two algorithms—Dijkstra shortest path algorithm and influence-based strongest path algorithm. Then a comparative study was done capturing different cases in which strongest path algorithm was better than shortest path algorithm in different cases.

Symbol table is the environment where the variables and functions/methods exist according to their scope and the most recent updated values are kept for the successful running of the code. It helps in code functioning. It is created during compilation and maintained, used during running of the code. Adding a utility called common file can help in conversion of one code to another code. As common file can be explained as the file containing the common functionalities of different languages, say, every language has a print function but with different syntax; these different syntax of print are added in common file which help in the conversion. In this paper, we present the compilation process mechanism with the help of common file in the symbol table. It also explains how a code is converted into another code.

Nowadays, there is a huge requirement of multimedia data security with an advent use of internet that is being more important to save the data confidentiality and save from various attacks. This paper represents the improvement of LSB (Least Significant Bit) watermarking mechanism by using two host images instead of one image to embed the watermark using some logical operations on the bits of the images. It provides better security and on the other and it preserves the originality of the host image. To compare the image quality we used some image quality parameters like PSNR, MSE, NAE, AD, MD, NCC, and SC.

Internal Model Control (IMC) technique is one of the well accepted model-based controller designing methodologies which is widely accepted in process industries due to their simplicity and ease of tuning. For controlling nonlinear processes IMC controllers are designed based on the linear approximation of nonlinear models. As a result IMC controllers sometimes fail to provide satisfactory performance under model uncertainty and large load variations with its fixed settings. Here we propose an adaptive IMC-PI controller for a level control process where the IMC tuning parameter, i.e., the close-loop time constant ($$ \lambda $$) is varied based on a set of predefined fuzzy rules depending on the process operating conditions in terms of process error ($$ e $$) and change of error ($$ \Delta e $$). Two sets of rule bases are used consisting of 25 and 9 rules for online fuzzy tuning of the IMC-PI controller. Widely different choice of the rule bases defined on two distinct fuzzy partitions justify the effectiveness as well as general applicability of the proposed scheme.

Face is the most easily identifiable characteristic of a person. Variations in facial expressions can be easily recognized by humans, while it is quite difficult for machines to recognize faces portraying varying facial expressions, pose, and illumination conditions efficiently. Face recognition works as a combination of feature extraction and classification. The selection of a combination of feature extraction technique and classifier to obtain maximum accuracy rate is a challenging task. This paper presents a unique combination of feature extraction technique and classifier that yields a satisfactory and more or less same accuracy rate when tested on more than one standard database. In this combination, features are extracted using principle coponent analysis (PCA). These extracted features are then fed to a minimum distance classification system. The proposed combination is tested on ORL and YALE datasets with an accuracy rate of 95.63% and 93.33%, respectively, considering variations in facial expressions, poses as well as illumination conditions.

A procedure based on the use of radial basis function network (RBFN) is presented for black box modeling of nonlinear dynamical systems. The generalization ability of RBFN is invoked to approximate the mathematical model of the given unknown nonlinear plant. This approximate model will then be used to predict the output of the plant at any given time instant. The parameters associated with RBFN are updated using the recursive equations obtained through the gradient-descent principle. The other benefit of using gradient descent principle is that it exhibits the clustering effect while adjusting the radial centers of RBFN. Real-time data of two benchmark problems: Box-Jenkins gas furnace data and Chemical process (polymer production), were used to show the application of RBFN for modeling purpose. Simulation results show that RBFN is well suited as a modeling tool for capturing the unknown nonlinear dynamics of the plant.

Computational grid houses powerful resources to execute computation-intensive jobs, which are submitted by the clients. Resources voluntarily become available in the grid, as a result of which, this collaborative computing becomes more cost effective than traditional HPC. In the grid, since, the participating resources are of varying capabilities, load balancing becomes an essential requirement. This workload distribution mechanism among available resources aims to minimize makespan, optimize resource usage, and prevent overloading of any resource. Eventually, the resources need to be prioritized based on their capability and demand in the current scenario. Thus, prioritization of resources balances workload in grid. In the proposed workload scheduling algorithm, nearest deadline first-scheduled (NDFS), resource ranking, and subsequent job scheduling maintains balanced load across the grid. The ranking of resources in computational grid is achieved using analytic hierarchy process (AHP) model. The primary objective of this paper is to optimize the workload of grid environment while executing multiple jobs ensuring maximum resource utilization within minimum execution time. Service quality agreement (SQA) is met through proper scheduling of jobs among ranked resources. The grid test bed environment is set up with the help of Globus toolkit 5.2. This paper presents the simultaneous execution results of the benchmark codes of fast Fourier transform (FFT) and matrix multiplication in order to balance the workload in grid test bed.

Features of cloud services that users’ data is remotely in obscure machine is neither owned nor controlled by user. For adoption of services, from users’ point of view, cloud security and arbitration are significant. Security gap among different tiers causes privacy issues, customer concern about losing sensitive data available in a cloud computing infrastructure. Disagreement among different components of cloud can be mitigated by Online Dispute Resolution (ODR) to great extent. In this paper, we are focusing mainly on three factors, first trying to identify the customer threats, concerns and cross-border conflicts while using cloud computing. Second on ODR (Online Dispute Resolution) and its mechanisms. Finally on required regulatory framework between consumer, industry, and geographic boundaries. An accepted regulatory framework across participants/consumers/providers is currently at a premature stage but as imperative adoption to cloud computing increases across industry there will be paradigm shift in effort sooner than later to build the same.

In this article, we have devised modified genetic algorithm (MfGA) based fuzzy C-means algorithm, which segment magnetic resonance (MR) images. In FCM, local minimum point can be easily derived for not selecting the centroids correctly. The proposed MfGA improves the population initialization and crossover parts of GA and generate the optimized class levels of the multilevel MR images. After that, the derived optimized class levels are applied as the initial input in FCM. An extensive performance comparison of the proposed method with the conventional FCM on two MR images establishes the superiority of the proposed approach.

A skill set is the ability of performing a particular job. Skill set is acquired by improving the psychomotor domain of the human being. Deficiencies in skills need to be measured and addressed. This may improve the level of skill and reduce deficiency. Deficiency diagnosis is a process of identification of the skills that are lacking in any learner. In this research work, authors have proposed a model that identifies the various deficiencies of a learner.

The competency of an employee is evaluated or identified through some techniques. The success of the Training Program can be evaluated only after it meets the need for which it was called for. This research will help the organization to scale up their training tools and methodology for any kind of training within the organization. An attempt is being made to identify the awareness on satisfaction level of the employees of an IT firm. The nature and behavior of employees are described by descriptive research design. Both the data collection methods have been used and selection of study area includes IT firms, it was selected for conducting the survey based on the judgment of the employees of all cadres. The sample size is 50. Analysis and interpretation of data being done by using statistical tools as percentage method and Chi-square method.

The security issues of cognitive radio network have taken more attention recently due to the new challenges in wireless communication. Confronts related to such a debate seem more predominant in presence of the malicious secondary users when the transmission range of licensed users is shorter compared to the network size. In this paper, a model is introduced which verifies the distance both geographically and through the measuring distance obtained by received power of cognitive user. This is achieved by minimizing the interference to the primary licensed user and upon the faithful operation of the secondary user by calculating the trustworthiness of all users irrespective of their priority. This necessitates focus upon the free space propagation path loss of the transmitted signal. Thus, analysis of the trustworthiness becomes essential followed by calculation of link budget power that ensures the designing cost without extra overhead during a secured communication with the received power and free space propagation path loss.

DRM provides a secure solution for the illegal distribution of digital content through communication networks. We propose a Secure and lightweight protocol for mobile DRM based on DRM community cloud (DCC) and banking community cloud (BCC). Non-repudiation property is a very important property that needs to be ensured for DRM. Non-repudiation property in this protocol is achieved using wireless public key infrastructure (WPKI), universal integrated circuit card (UICC) at the client side, DRM community cloud (DCC) at the cloud provider (CP) and banking community cloud (BCC) at the Issuing Bank. BCC and DCC are a Cloud of Secure Elements (CSE). Our proposed protocol achieves end-to-end security from the client to DCC and BCC.

Security is one of the major concerns in cloud computing now-a-days. Malicious code deployment is the main cause of threat in today’s cloud paradigm. Antivirus software unable to detect many modern malware threats which causes serious impacts in basic cloud operations. This paper counsels a new model for malware detection on cloud architecture. This model enables identification of malicious and unwanted software by amalgamation of multiple detection engines. This paper follows DNA sequence detection process, symbolic detection process, and behavioural detection process to detect various threats. The proposed approach (PMDM) can be deployed on a VMM which remains fully transparent to guest VM and to cloud users. However, PMDM prevents the malicious code running in one VM (infected VM) to spread into another noninfected VM with help of hosted VMM. After detecting malicious code by PMDM technique, it warns the other guest VMs about it. In this paper, a prototype of PMDM is partially implemented on one popular open-source cloud architecture—Eucalyptus.

Outlier detection also popularly known as anomaly detection is the process of recognizing whether the given data is normal or abnormal. Some of the applications of this outlier detection are: network intrusion detection, fraud detection, database cleaning, etc.; In most situations, there is scarcity of abnormal data where as plenty of normal data is available. This is the biggest challenge of novelty detection. The characteristics of abnormal or outlier data are often unknown beforehand. Density estimation methods can be used for novelty detection tasks. These methods work only when the assumed data distribution is same as the underlying data distribution which may not be known in advance. C-SVDD and ν-SVDD are used for novelty detection tasks in our experiments. Experiments are performed on a toy data set of bivariate and overlapping classes and real-time multivariate data. Different kernels are also used for experimental studies. All experiments shows that RBF (Gaussian) kernel gives better performance than the other types of kernels. Experimental results on both artificial and real-world data are reported to illustrate the promising performance of outlier data detection.

This paper is aimed at diagnosing automotive engine fault in real-time utilizing BigData framework called spark. An automobile in the present day world is equipped with millions of sensors which are under the command of a central unit the ECU (Electronic Control Unit). ECU holds all information about the engine. A network of ECUs connected across the globe is a source tap of BigData. Leveraging the new sources of BigData by automotive giants boost vehicle performance, enhance loco driver experience, accelerated product designs. A piezoelectric transducer coupled to the ECU captures the vibration signals from the engine. The engine fault is detected by carving the problem into a pattern classification problem under machine learning after extracting cyclostationary features from the vibration signal. Spark-streaming framework, the most versatile BigData framework available today with immense computational capabilities is employed for engine fault detection and analysis.

In distribution system to determine static states at each node or bus and operating conditions, the load flow studies are very crucial. The load flow studies are very important, not only in finding static states but also during distribution system planning and its extension. In this paper, the load flow problem has been solved by artificial neural networks and these networks are efficient to describe the relation involved within the raw data. Two types neural networks are proposed to solve load flow problem of a distribution system, first one is Radial Basis Function Neural Network (RBFN) and other one is Multilayer Feedforward Neural Network with Backpropagation Algorithm (MFFN with BPA). The mathematical model of distribution load flow comprises a set of nonlinear algebraic equations that are solved using network topology-based distribution load flow which is usurped as reference off-line load flow. A series of training data is generated using off-line load flow, which is used to train the neural networks. The training data consists of different loading conditions and voltages corresponding to each and every node in the distribution system. The neural networks are trained with series of training data and tested with a loading which is not present in training data. Results obtained from two neural networks closely agrees with the reference off-line load flow result of same loading. The results of neural networks are compared together and computational time of two neural networks is considerably small.

Nowadays bugs are the commonly occurring problems in many types of software. In order to prevent from these issues, a detailed study of bugs is an essential thing. Bugs are classified based on their severity in corresponding bug repositories. Some of the bug repositories are Mozilla, Android, Google Chromium, etc. So finding the most frequently occurring bugs is the right solution for the software malfunctioning. Thus it can help developers to prevent those bugs in the next release of the software. In this paper, our main aim is the mining of bugs from the bug summary data in the bug repositories by applying FP-Growth, one of the best techniques for finding frequently occurring pattern using WEKA.

In Security-as-a-service model the objective is to provide security as one of the cloud services. In this model the security is provided from the cloud in place of traditional on-premise implementation. The objective of this initiative is to provide Anti-Malware functionality as a cloud service. This paper provides implementation framework for Anti-Malware system from the cloud as a service. The framework uses several existing file scanning web-based anti-malware engines. The Anti-Malware SecaaS offers all the benefits provided by Security-as-a-Service (SecaaS) model. The proof-of-concept (POC) prototype of Anti-Malware AM-SecaaS is implemented and evaluated successfully. An innovative approach is used to integrate this POC with other SecaaS options so that various SecaaS options are provided to users intelligently and transparently.

In this paper, an effort has been made to emphasize the usefulness of Hausdorff Distance (HD) and Directed Hausdorff Distance (DHD) based features for the recognition of online handwritten Bangla basic characters. Every character sample is divided into N number of rectangular zones and then HD- and DHD-based features have been computed from every zone to every other zone. These distance measurements are served as feature values for the present work. Experiment has been done on a set of 10,000 character dataset. Multilayer Perceptron (MLP) produces the best result with an accuracy of 95.57% when sample character is divided into 16 rectangular zones and DHD-based procedure has been considered.

Malware stands for malicious software. Any file that causes damage to the computer or network can be termed as malicious. For malware analysis, there are two fundamental approaches: static analysis and dynamic analysis. The static analysis focuses on analyzing the file without executing, whereas dynamic analysis means analyzing or observing its behavior while it is being executed. While performing malware analysis, we have to classify malware samples. The different types of malware include worm, virus, rootkit, trojan horse, back door, botnet, ransomware, spyware, adware, and logic bombs. In this paper, our objective is to have a breakdown of techniques used for malware analysis and a comparative study of various malware detection/classification systems.

In the field of technology, software is an essential driver of business and industry. Software undergoes changes due to maintenance activities initiated by bug fixing, improved documentation, and new requirements of users. In software, code smells are indicators of a system which may give maintenance problem in future. This paper evaluates six types of machine learning algorithms to predict change-proneness using code smells as predictors for various versions of four Java-coded applications. Two approaches are used: method 1-random undersampling is done before Feature selection; method 2-feature selection is done prior to random undersampling. This paper concludes that gene expression programming (GEP) gives maximum AUC value, whereas cascade correlation network (CCR), treeboost, and PNN\GRNN algorithms are among top algorithms to predict F-measure, precision, recall, and accuracy. Also, GOD and L_M code smells are good predictors of software change-proneness. Results show that method 1 outperforms method 2.

Phones are turning into the surely understood method for relationship; strategies helping adaptability connote a genuine asset of issues in light of the fact that their preparatory style did not execute effective assurance. A novel structure work of turn imperceptible framework strikes, known as versatility-based avoidance, where an adversary partitions an unsafe payload in a manner that no part can be recognized by ebb and flow ensuring strategies, for example, the most cutting edge framework assault acknowledgment procedures working in condition full method. Snort is a free Network Intrusion Detection System blending several benefits provided by trademark, strategy, and variation from the norm focused examination and is respected to be the most regularly executed IDS/IPS mechanical advancement globally. This report recommends various changes for improving Snort Security Platform and different gathering is suggested to strengthen the measure of rays which can be inspected, and Snort’s multi-threading open doors are scrutinized.

The internet of thing (IOT) is an upheaval of traditional internet. It is new revolution that not only allows transmission and exchange of data but also communication between the physical objects in the real world. These heterogeneous devices are pervasive, ubiquitous in nature that changes dynamically and frequently. Mostly, these devices are low-powered devices and have less computation power and capacity. Traditional routing techniques in ad hoc network do not take security and energy into consideration. To extend the lifetime of the network, the energy supply and consumption of the node is an important aspect. The routing of the packet should be as such that even the low-powered devices have the ability to receive and transmit the packet. The presence of malicious node will make the network more susceptible to the different attacks and threat. To overcome this problem, an energy-efficient routing protocol with a centralized controller is integrated with IOT devices.

Speech Recognition is a challenging task especially for a multilingual country like India as the speakers are habituated in using mixed language and accent. Bangla is a very popular language in East Asia and a fully functional Automated Speech Recognition System (ASR) for it is yet to be developed. Every language embodies a set of sounds called phoneme set, which is the building block for the words of that language. READ (Record Extract Approximate Distinguish) is a Bangla phoneme recognition system, proposed toward the development of a Bangla ASR. To start with, Mel Scale Cepstral Coefficient (MFCC) features have been used for testing on a database of 1400 Bangla vowel phonemes and an accuracy of 98.35% has been obtained.

This work presents the application of biometrics in animal identification, which is a highly researched topic in human recognition. Here, our analysis presents the identification of zebra in their natural habitat. All the techniques are tested on 824 Plains zebra images captured at Ol’Pejeta conservancy in Laikipia, Kenya. We have used coat strips as a biometric identifier which is unique in nature. To improve the performance of identification, information fusion of coat strips can be taken place from many points in zebra skin such as near legs, stomach and neck. Here two region near stomach (flake) and first limb (leg) is cropped from the textural pattern of strips of zebra is used in feature extraction. GMF, AAD, mean, and eigenface feature extraction methods are applied on flake and limb ROI of zebra. Then a novel image enhancement method: difference subplane adaptive histogram equalization is applied to improve the identification rate. Our technique is based on information fusion in fusing the score from stomach (flake) and first limb (leg) region. For this, sum, product, frank T-norm, and Hamacher T-norm rules are applied to validate the identification results. Information fusion improves the identification results from the previous reported results from eigenface, CO-1 algorithm, and stripecodes. The improvement in results verifies the success of our approach of information fusion using score level fusion.

The motivation of this paper is formation of sports network and characterization of the small world network phenomenon by analyzing the data of individual players of a team. Analysis of the network suggests that sports network can be considered as small world and inherits all characteristics of small world network. Making a quantitative measure for an individual performance in the team sports is important in respect to the fact that for team selection of International football matches, from a pool of best players, only 11 players can be selected for the team. The statistical record of each player is considered as a traditional way of quantifying the performance of a player. But other criteria like performing against a strong opponent or executing a brilliant performance against a strong team deserves more credit. In this paper, a method based on social networking is presented to quantify the quality of player’s efficiency and is defined as the total matches played between each team members of individual teams and the members of different teams. The application of Social Network Analysis (SNA) is explored to measure performances and rank of the players. A bidirectional weighted network of players is generated using the information collected from English Premier League (2014–2015) and used for network formation.

An ultra wideband bandpass filter with a notch band for satellite communication is stated in the paper using pseudo-interdigital structure. The structure is planar and there is no use of via or defected ground structure that makes the structure less complex and easy to fabricate. The insertion loss of the proposed filter is less than 0.8 dB and return loss more than 16.7 dB. The notch band centered at 8 GHz has insertion loss of 12.8 dB. The small size of the filter is 0.265 λg × 0.071 λg. The filter is designed and simulated in ADS software.

The paper discusses an essential data mining task, clustering. Clustering groups similar instances and results in classes of similar instances. In this paper, clustering methods k-means, SOM clustering, and hierarchical method of clustering are discussed and implemented in R. Before the application of clustering algorithms cluster tendency is evaluated to determine whether the data set is appropriate for clustering or not. Cluster tendency is also discussed in the paper.

Contour detection forms a significant module of computer vision frameworks, and is still an active area of research. This paper presents a feature-based edge detection strategy on color images, where the likeliness of a pixel to lie on a border separating two distinct regions is estimated by utilizing joint information obtained from two different visual cues. The first cue draws special attention to regions with presence of discontinuities and is constructed by exploiting standard deviation, busyness and entropy measures on the input image and its intrinsic map. The second cue diminishes the chances of broken edge generation by utilizing a population-based global optimisation heuristic (Particle Swarm Optimization) to detect the final edges from highlighted regions of the former cue. The result achieves noteworthy performance that is orders of magnitude better than most of the competing standard approaches, while attaining promising detection results on BSDS300 dataset.

Subtle structural differences can be observed in the islets of Langerhans region of microscopic image of pancreas cell of the rats having normal glucose tolerance and the rats having pre-diabetic (glucose intolerant) situations. This paper proposes a way to automatically segment the islets of Langerhans region from the histological image of rat’s pancreas cell and on the basis of some morphological feature extracted from the segmented region the images are classified as normal and pre-diabetic. The experiment is done on a set of 134 images of which 56 are of normal type and the rests 78 are of pre-diabetic type. The work has two stages: primarily, segmentation of the region of interest (roi), i.e., islets of Langerhans from the pancreatic cell and secondly, the extraction of the morphological features from the region of interest for classification. Wavelet analysis and connected component analysis method have been used for automatic segmentation of the images. A few classifiers like OneRule, Naïve Bayes, MLP, J48 Tree, SVM, etc, are used for evaluation among which MLP performed the best.

The purpose of materialized view selection is to minimize the cost of answering queries and fast query response time for timely access to information and decision support. Besides various research issues related to data warehouse evolution, materialized view selection is one of the most challenging ones. Various authors have given different methodologies, strategies and followed algorithms to solve this problem in an efficient manner. The main motivation behind this systematic review is to provide a path for future research scope in materialized view selection. Various techniques presented in the papers are identified, evaluated, and compared in terms of memory storage space, cost, and query processing time to find if any particular approach is superior to others. By means of a review of the available literature, the authors have drawn several conclusions about the status quo of materialized view selection and a future outlook is predicted on bridging the large gaps that were found in the existing methods.

The SQL injection is a predominant type of attack and threat to web applications. This attack attempts to subvert the relationship between a webpage and its supporting database. Due to widespread availability of valuable data and automated tools on web, attackers are motivated to launch high profile attacks on targeted websites. This paper is an effort to know the current state of SQL injection attacks. Different Researchers have proposed various solutions to address SQL injection problems. In this research work, those countermeasures are identified and applied to a vulnerable application and database system, then result are illustrated.

Metrics are commonly used to guide the designers to build quality data warehouse models. Recently, researchers have defined various object-oriented metrics for data warehouse conceptual model to access their quality. These metrics require theoretical and empirical validation to confirm their applicability in real time. Empirical validation of object-oriented metrics has already been carried out but theoretical validation has not been taken into account. In this paper, theoretical validation for object-oriented metrics using Zuse’s framework is presented to show that these metrics may be considered as strong measures for evaluating quality of object-oriented conceptual models of data warehouse.

There is an immense potential to solve many of the challenging and persistent traffic and accident related problems by implementing IoT in vehicles. Various sensors come in-built with most of the vehicles, simplifying our task. By monitoring engine parameters, we can warn the vehicle owners of potential breakdowns and also notify the nearest service centre. In case of a breakdown, help is immediately dispatched to the vehicle. This is particularly useful for heavy vehicles and public transport which causes major traffic jams. The ubiquity of vehicles on the roads also makes possible the use of sensors for weather detection by using onboard sensors for collecting real-time data.

Human in an environment sees and then perceives objects of interest before he/she tries to find the correlation and association between the various objects in the region of their interest (ROI). By doing so, the agent here, the human develops an understanding of the environment, may it be static and certain or dynamic and uncertain. This paper simulates such an ability of humans, vital to his/her understanding, after being exposed to a visual stimulus. Filtration or selective attention happens then followed by clustering based on identified associations. These clusters form the basis of understanding and stored as Concept maps inside the long-term memory. In order to simulate this feature, various techniques and clustering algorithms exist. This work is a cognitive architectural approach to tackle the clustering problem, as it is a more natural and intuitive approach followed by humans. The ACT-R architecture has been chosen for the task.

Human action recognition refers to the classification of human action from video clips automatically. Images extracted from the video clips at regular time interval are processed to identify the action contained in them. This is done by comparing these images with images taken from appropriate standard action databases. Thus, human action recognition becomes the task of verifying the similarity between two images. This paper proposes mutual difference score as a measure of similarity between two images. The proposed measure has been validated using the Weizmann and KTH datasets.

In the recent years opinion mining plays an important role by business analyst before launching a product. Opinion mining mainly concerns about detecting and extracting the feature from various opinion rich resources like review sites, discussion forum, blogs and news corpora so on. The data obtained from those are highly unstructured in nature and very large in volume, therefore data preprocessing plays an essential role in sentiment analysis. Researchers are trying to develop newer algorithm. This research paper attempts to develop a better opinion mining algorithm and the performance has been worked out.

In the recent past, Delay tolerant network has gradually evolved as a viable solution to various needs arising in intermittently connected wireless networks. In this paper, a border security architecture, has been proposed where DTN has been used to securely transfer messages within the army units of a particular region that are deployed in harsh and in adaptable terrains, thereby, ensuring integrity and security of the message. In the proposed approach, a special property of the nodes is used which makes them self-disposable in nature after their defined TTL gets over. So, the message as well as the node would be always safe from getting into unauthorized hands.

This analysis of MWSNs illustrates that IEEE 802.15.4 has very wide applications in the province of mobile wireless sensor networks as per the related research in this extent is concerned. One of the major investigation in the consideration of MWSNs sustains from the problem of system throughput and end-to-end delay along with the issue of energy consumption. This idle illustrates a X-layer network (cross-layer network) transformation model that can decline the problem of energy consumption and along with end-to-end delay in these networks. In this paper, the moderate model restrains four layers in the network transformation: (a) application layer, which has been utilized to update the node locality and information; (b) network layer, which has been accomplished to recognize the routing of the internetwork through links; (c) MAC (medium access control) layer, which has been centralized on the effectiveness of the networks, and (d) physical layer has also been recommended keeping the intentional view on transmission power from sensor node to the sink node. The place/location and position/status of the mobile node is interlinked in the routing transformation immediately as soon as the route finding process is successfully terminated and then it is been employed by controlling the transmission power of the MAC layer to rectify the range of the transmission with respect to the route. As per the future expectations of practical characteristics is concerned, adjacent NB(neighbor) list discovery broadcast will be engaged only for active nodes. But this paper is accomplished an modern technique, i.e., mobility aware protocol, which recognizes the mobility (velocity/speed) of the nodes so that only those respective nodes will be upgraded with its adjacent NB-list broadcasting, resulting in minimum power utilized by the network interface and also in the degradation of the energy consumption of the node’s. In spite of the above concern, one more additional cause approaching in this model is the issue of bottleneck problem, which has been established due to multiple sources resulting in huge packet loss. This issue has been solved by pipe-lining those packets of the sensor which are nearer to the sink node, which results in reduction of end-to-end delay and energy utilization resulting in high system throughput. Through NS-II simulation, the results of energy consumption, system throughput, end-to-end delay, etc., has been shown.

Every year, we find a lot of people committing suicide in metro. To avoid such incidents, we have designed a system that would take the captured video of preinstalled CCTV cameras in metro stations and analyze them. On breaking down the captured video into frames, the region of interest for each will be calculated and their histogram found. Processing the histogram values, if danger is detected then an alert message will be fired. In future with this triggering of the message a physical barrier, installed beforehand at the edge of the platform, will come up thereby preventing the victim from making the suicide attempt, saving his/her life and hours of harassment for others.

BGP (Border Gateway Protocol) is one of the core internet backbone protocols, which were designed to address the large-scale routing among the ASes (Autonomous System) in order to ensure the reachability among them. However, an attacker can inject update messages into the BGP communication from the peering BGP routers and those routing information will be propagated across the global BGP routers. This could cause disruptions in the normal routing behavior. Specially crafted BGP messages can reroute the traffic path from a source ASN to a specific destination ASN via another path and this attack is termed as AS Path Hijacking. This research work is focused on the detection of suspicious deviation in the AS path between a source and destination ASNs, by analyzing the BGP update messages that are collected by passive peering to the BGP routers. The research mainly focuses on identifying the AS Path Hijacking by quantifying: (1). How far the deviation occurred for a given AS Path and (2). How much credible is the deviated AS path. We propose a novel approach to calculate the deviation occurred by employing weighted edit distance algorithm. A probability score using n-gram frequency is used to determine credibility of the path. Both the scores are correlated together to determine whether a given AS Path is suspicious or not. The experimental results show that our approach is capable of identifying AS path hijacks with low false positives.

Steganography is an art of hiding of secret information in an innocuous medium like an image. Most of the current steganographic algorithms hide data in the spatial or transform domain. In this paper, we perform attacks on three singular value decomposition-based spatial steganographic algorithms, by applying image processing operations. By performing these attacks, we were able to destroy the stego content while maintaining the perceptual quality of the source image. Experimental results showed that stego content can be suppressed at least by 40%. PSNR value was found to be above 30 dB and SSIM obtained was 0.61. Markov feature and BER are used to calculate the percentage of stego removed.

Most of the researchers around the world focus on developing monolingual Optical Character Recognition (OCR) systems. But in a multilingual country like India, it is quite common that a single document page includes text words written in more than one script. Therefore, OCRing such documents need a script identification module as a prerequisite. This paper reports a complete script recognition system for handwritten mixed-script documents. The document pages are first segmented into their corresponding text-lines and words. Then, the script recognition is done at word-level using texture-based features. The present technique is applied on 100 mixed-script document pages written in Bangla or Devanagari text mixed with English words. Encouraging outcomes would motivate more researchers to work on multilingual handwriting recognition domain.

Stop words removal is an important step in many natural language processing (NLP) tasks. Till now, there is no standardized, exhaustive, and dynamic stop word list created for documents written in Indian Gujarati language which is spoken by nearly 66 million people worldwide. Most of the existing stop words removal approaches are file or dictionary based, wherein a hard-coded static, nonstandardized, and individually created list of stop words is used. The existing approaches are time consuming and complex owing to file or dictionary preparation by collecting possible stop words from a large vocabulary, complex framework and a morphologically variant Gujarati document. Even the other proposed approaches in the literature are also very restricted due to their dependence on word-length, word-frequency, and/or training data set. For the first time in scientific community worldwide, this paper proposes a dynamic approach independent of all factors namely usage of file or dictionary, word-length, word-frequency, and training dataset. An 11 rule-based approach is presented focusing on automatic and dynamic identification of a complete list of Gujarati stop words. Extensive empirical evidence has been presented through deployment of proposed algorithm on nearly 600 Gujarati documents, categorized into routine and domain-specific categories. The respective results with 98.10 and 94.08% average accuracy show that the proposed approach is effective and promising enough for implementation in NLP tasks involving Gujarati written documents.