Recent Advances in Computer Science and Communications - Volume 14, Issue 5, 2021

Aims: Propose an imputation measure for filling missing data values so as to make the incomplete medical datasets as complete datasets. Apply this imputation measure on imputed datasets to achieve improved classifier accuracies. Objective: The basic intention of the present study is to present an imputation measure to find the proximity between medical records and an approach for imputation of missing values in medical datasets to improve the accuracy of existing classifiers. Methods: The performance of proposed approach is compared to existing approaches with respect to classifier accuracy and also by performing non-parametric test called Wilcoxon test. Results & Conclusion: Experiments are conducted by considering three benchmark datasets CLEVALAND, PIMA, ECOLI and by applying proposed imputation technique with KNN, J48 and SMO classifiers and classifier accuracies are determined. The results obtained are then compared to thirteen existing benchmark imputation techniques available in KEEL repository. Experiment results proved the importance of the proposed imputation technique.

Background: E-mail is an efficient way to communicate. It is one of the most commonly used communication methods, and it can be used for achieving legitimate and illegitimate activities. Many features that can be effective in detecting email fraud attacks, are still under investigation. Methods: This paper proposes an improved classification accuracy for fraudulent emails that is implemented through feature extraction and hybrid Machine Learning (ML) classifier that combines Adaboost and Majority Voting. Eleven ML classifiers are evaluated experimentally within the hybrid classifier, and the performance of the email fraud filtering is evaluated by using WEKA and R tool on a data set of 9298 email messages. Results: The performance evaluation shows that the hybrid model of Voting using Adaboost outperforms all other classifiers, with the lowest Error Rate of 0.6991%, highest f1-measure of 99.30%, and highest Area Under the Curve (AUC) of 99.9%. Conclusion: The utilized proposed email features with the combination of Adaboost and Voting algorithms prove the efficiency of fraud email detection.

Introduction: Intrusion detection systems play a key role in system security by identifying potential attacks and giving appropriate responses. As new attacks are always emerging, intrusion detection systems must adapt to these attacks, and more work is continuously needed to develop and propose new methods and techniques that can improve efficient and effective adaptive intrusion systems. Feature selection is one of the challenging areas that need more work because of its importance and impact on the performance of intrusion detection systems. This paper applies an evolutionary search algorithm in feature subset selection for intrusion detection systems. Methods: The evolutionary search algorithm for the feature subset selection is applied and two classifiers are used, Naïve Bayes and decision tree J48, to evaluate system performance before and after features selection. NSL-KDD dataset and its subsets are used in all evaluation experiments. Results: The results show that feature selection using the evolutionary search algorithm enhances the intrusion detection system with respect to detection accuracy and detection of unknown attacks. Furthermore, time performance is achieved by reducing training time, which is reflected positively in overall system performance. Discussion: The evolutionary search applied to select IDS algorithm features can be developed by modifying and enhancing mutation and crossover operators and applying new enhanced techniques in the selection process, which can give better results and enhance the performance of intrusion detection for rare and complicated attacks. Conclusion: The evolutionary search algorithm is applied to find the best subset of features for the intrusion detection system. In conclusion, it is a promising approach to be used as a feature selection method for intrusion detection. The results showed better performance for the intrusion detection system in terms of accuracy and detection rate.

Background: The App Stores, for example, Google Play and Apple Play Store provide a platform that allows users to provide feedback on the apps in the form of reviews. An app review typically includes star rating followed by a comment. Recent studies have shown that these reviews possess a vital source of information that can be used by the app developers and the vendors for improving the future versions of an app. However, in most of the cases, these reviews are present in unstructured form and extracting useful information from them requires a great effort. Objective: This article provides an optimized classification approach that automatically classifies the reviews into a bug report, feature request, and shortcoming & improvement request relevant to Requirement Engineering. Methods: Our methodology used supervised Machine Learning techniques (Multinomial Naive Bayes, Linear SVC, CART Decision Tree) that have been first evaluated based on t h e conventional Bag of Words (BOW) feature extraction model along with other auxiliary features such as review rating, review length, review tense, and review sentiment score to automatically classify app reviews into their relevant categories, and subsequently, a feature selection method is used to provide an optimized classification approach for improving classifier performance using Particle Swarm Optimization (PSO) nature-inspired algorithm. Results: Result shows that we achieved best results with precision of 62.0 % and recall of 47.3 % with Linear SVC Machine Learning technique, which we further optimized with PSO nature-inspired algorithm, i.e., with PSO + Linear SVC, thus, resulting in a precision of 63.6 % and recall of 55.0 %. Conclusion: This optimized automatic classification improves the Requirement Engineering where developer straightforwardly knows what to improve further in the concerned app.

Background: Decision making requires a rigorous process of evaluation, which is an analytical and organized process to figure out the present positive influences, favourable future prospects, existing shortcomings and ulterior complexities of any plan, program, practice or a polity. Evaluation of policy is an essential and vital process required to measure the performance or progression of the scheme. The main purpose of policy evaluation is to empower various stakeholders and enhance their socio-economic environment. A large number of policies or schemes in different areas are launched/constituted by government in view of the citizen welfare. Although, the governmental policies intend to better shape up the quality of life of people but may also impact their everyday's life. Objective: The contemplation of public opinion plays a very significant role in the process of policy evaluation. Therefore, the aim of this paper is to incorporate the concept of opinion mining in policy evaluation. An attempt has been made to elevate the process of policy evaluation by analyzing public opinion. Methods: A latest governmental scheme Saubhagya launched by the Indian government in 2017 has been selected for evaluation by applying supervised learning based opinion mining techniques. The data set of public opinion associated with this scheme has been captured by Twitter. Results: The result validates that the proposed methodology supports the optimizing process of policy evaluation and provides a more accurate and actual status of policy's effect among Indian citizen. As a result, this would aid in identifying and implementing the preventive and corrective measures required to be taken for a successful policy. Conclusion: The proposed methodology will stabilize and strengthen the process of policy evaluation which target towards favourable and flourishing future prospects concerning the socio-economic status of a nation. The results are quite exciting and further extension of work will be performed in order to develop and design a patent framework in the area of social big data analytics.

Introduction: Cloud computing involves the use of maximum remote services through a network using minimum resources via internet. There are various issues associated with cloud computing, such as privacy, security and reliability. Due to rapidly increasing information on the cloud, it is important to ensure security of user information. Biometric template security over cloud is one such concern. Leakage of unprotected biometric data can serve as a major risk for the privacy of individuals and security of real-world applications. Methods: In this paper, we improvise a secure framework named DeepCrypt that can be applied to protect biometric templates during the authentication of biometric templates. We use deep Convolutional Neural Networks to extract features from these modalities. The resulting features are hashed using a secure combination of Blowcrypt (Bcrypt) and SHA-256 algorithm, which salts the templates by default before storing it on the server. Results: Experiments conducted on the CASIA-Iris-M1-S1, CMU-PIE and FVC-2006 datasets achieve around 99% Genuine accept rates, proving that this technique helps to achieve better performance along with high template security. Conclusion: The proposed method is robust and provides cancellable biometric templates, high security and better matching performance as compared to traditional techniques used to protect the biometric template.

Background: Speech Recognition is the most effective and suitable way of communication. Extracted features play an important role in speech recognition. Previous research works for Hindi speech recognition lack detailed comparative analysis of the feature extraction methods using dynamic and energy parameters. Objective: The research work presents experimental work done to explore the effects of integrating dynamic coefficients and energy parameters with different feature extraction techniques on Connected word Hindi Speech recognition. As extracted features play a significant role in speech recognition, a comparative analysis is presented to show the effects of integration of dynamic and energy parameters to basic extracted features. Methods: Speaker dependent system was proposed with monophones based five states Hidden Markov Model (HMM) using HTK Tool kit. Speech data set of connected words in Hindi was created. The feature extraction techniques such as Linear Predictive Coding Cepstral coefficients (LPCCs), Mel Frequency Cepstral Coefficients (MFCCs), and Perceptual Linear Prediction (PLPs) coefficients were applied integrating delta, delta2, and energy parameters to evaluate the performance of the proposed methodology for speaker dependent recognition. Results: Experimental results show that the system achieved the highest recognition word accuracy of 89.97% using PLP coefficients. The PLP coefficients achieved 4% increment in word accuracy than original MFCCs and a 16% increment in word accuracy than LPCCs. Adding energy parameters to original MFCCs increased word accuracy by 1.5% only while adding dynamic coefficients delta and double delta had no significant effect on speech recognition accuracy. Conclusion: Research findings reveal that PLP coefficients outperformed. Explorations reveal that the integration of energy parameters are better than original MFCCs. Investgations also reveal that adding energy parametres improved recognition score while adding delta and delta2 coefficients to basic features did not improve the recognition scores. Research findings could be used to enhance the performance of a speech recognition system by using a suitable feature extraction technique and combining the different feature extraction techniques. Further, investigations can be used to develop language resources for refining speech recognition. The work can be extended to develop a continuous Hindi speech recognition system.

Background: With the complexity of the network structure and the diversity of access, the access control model of distributed permissions management came into being. Aims: The central idea of the model is to store the access permissions of all nodes in the model to the network nodes covered by the access control model, and let the neighboring nodes act as access control agents. That is, each node in the model is a routing node in the execution of a certain access control. Methods: Considering the execution parameters of distributed access control model, the evaluation index of key node identification technology in distributed authorized access control model is established. Access control to achieve the management of distributed permissions. It is necessary to protect the key nodes in the model, increase the robustness of the access control model and support the smooth implementation of distributed access. Results: This paper presents a key node recognition algorithm based on distributed access control magic. The application of key node recognition algorithm in distributed access control model in traditional network is compared. Conclusion: Compared with the traditional key node recognition algorithm in the distributed access control model, the algorithm proposed in this paper is more accurate.

Background: The paper discussed about the malware propagation dynamics in wireless sensor network. Malware attack is harmful for network stability as well as battery consumption of sensor nodes. Objective: The objective of proposed study is to develop a model that describes the dynamic propagation behavior of malware in wireless sensor network and suggest corrective measure through which breakout from the network. Methods: Use the concept of epidemic modeling and spatial correlation to describe the malware dynamics in wireless sensor network. Write the sets of differential equation to study its behavior. Results: Methodically find results have been verified with the help of simulation using MATLAB. The effect of spatial correlation on malware propagation in wireless sensor network has been analyzed. Conclusion: In this paper, an epidemic based model with spatial correlation is presented for the study of malware propagation characteristics in WSN. The equilibrium points of the system have been obtained. The expression of basic reproduction number and threshold value of correlation coefficient have been obtained. The basics reproduction (R 0 ) helps in the analysis of network stability. On the basis of their value we found that if R 0 is less than one the system will be stable and malware-free, and when R 0 is greater than one the system exists in endemic state and malware persists in the network. The spatial correlation helps to control the malware propagation in the network.

Background: In daily life, one may encounter several problems whose solution is based on the consideration of several criteria altogether. To generate a solution for such problems is, in fact, a complex task. If one generates a solution, then there is no guarantee that the solution is optimal. It is really challenging to take into account all the criteria at the same time and generate a solution that is optimal. Several techniques have been proposed to solve multi-criteria decision- making problems. Methods: In this paper, we propose an intelligent solution to multi-criteria decision-making problems by means of fuzzy aggregation using the matrix method. The proposed method finds application in solving such problems where the criteria are defined in the qualitative form (fuzzy linguistic variables) rather than in quantitative form (crisp). A novel application of the proposed approach to rank the employees in an organization is presented in the paper. Results: The proposed method is empirically validated by applying it to find the best employee in an institute of repute. Data collection is done in real-time to prove the utilization of the proposed approach in an efficient manner. Conclusion: The results obtained after the application of the proposed method to rank the employees are realistic as it takes into consideration the ambiguity involved in the human thought process.

Background: The spam message aimed at mobile devices has become a prevalent activity in recent years. About 76% of mobile users received spam messages of Ads, fraud messages and phishing links. It has led to a significant financial loss of 91.5 billion Yuan per year and this problem will be more severe in the coming future. Although the operators are making much effort against spam message, the number of spams are increasing day by day. Aim: Many schemes designed to detect and filter spam have been proposed in the research area. The Bayesian classification algorithm is one of the most popular schemes. Most of the spam detection schemes based on Bayesian are designed for communication providers. The scalability of filter policy is hard to control because the strict policy might filter the normal messages while some spam messages would not be detected due to the flexibility in the policy. Methods: The current mobile device has a robust computational capability so it can execute complex jobs. A spam detection framework is designed in the present study. Results: It can sniffer the coming messages in mobile device by hooking the Android SMS API, and send them to the filter module, -which is responsible to filter messages into normal or spam based on Bayesian classification. This is a light-weight framework of consuming low power, which is suitable for mobile devices. Extra experiments are conducted to prove their accuracy and efficiency. Conclusion: The results showed that it could filter the spam among receiving the messages in realtime and become more accurate by learning the users' feedback.

Background: Recent advances in the World Wide Web and semantic networks have amplified social networking platforms, where the users share their photos, hobbies, location, interests, and experiences such as movie or restaurants. Social media platforms such as Facebook, twitter and LinkedIn are used to recommend the users the things of their interests such as movie, food, locations and friends. Objective: A novel method for the recommendation of movies to a friend using whale optimization has been introduced. Ratings given by friends of various movies are employed to recommend movies. Methods: Different evolutionary-based optimization methods have been applied for movie recommendation. The proposed method has been tested on movie-lense dataset and results are compared with 5 other methods namely, K-means, PCA K- means, SOM, PCA-SOM, PSO and ABC in terms of mean absolute error, precision and recall. Results: The experimental results demonstrate that proposed method outperformed all considered methods for 88.5% clusters centers in terms of precision, recall and mean absolute error. Conclusion: A novel recommendation system based on users rating has been designed to recommend movies to friends. It leverages the strengths of whale optimization to find the optimal solution.

Background: In this competing era, education has become part of everyday life. The process of imparting the knowledge to the learner through education is the core idea in the Teaching- Learning Process (TLP). An assessment is one way to identify the learner’s weak spot in the area under discussion. An assessment question has higher preferences in judging the learner's skill. In manual preparation, the questions are not assured in excellence and fairness to assess the learner’s cognitive skill. To generate questions is the most important part of the teaching-learning process. It is clearly understood that generating the test question is the toughest part. Objective: The proposed system is to generate the test questions that are mapped with bloom’s taxonomy to determine the learner’s cognitive level. The cloze type questions are generated using the tag part-of-speech and random function. Rule-based approaches and Natural Language Processing (NLP) techniques are implemented to generate the procedural question of the lowest bloom’s cognitive levels. Methods: Proposed an Automatic Question Generation (AQG) system which, automatically generates the assessment questions from the input file Dynamically. Results & Conclusion: The outputs are dynamic to create a different set of questions at each execution. Here, the input paragraph is selected from the computer science domain and their output efficiency is measured using precision and recall.

Background: Genes expression is high dimensional data, so it is very difficult to classify high dimensional data through traditional machine learning approaches. In this work we have proposed a model based on combined approach of Convolutional Neural Network and Recurrent Neural Network, both belong to deep learning model. The prediction has shown improved result than other machine learning algorithms. Expressions are generated through histone modification. Methods: To improve the accuracy deep learning model is proposed i.e. based on Convolutional and Recurrent neural network. This proposed model uses filter, causal convolutional layers and Residual Block for predictions. Results: In this work we have implemented the machine learning algorithms and deep learning algorithms like Logistic Regression, SVM, CNN, Deep Chrome and the proposed Temporal Neural Network. The performance is measured on the basis of parameters like accuracy, precision and AUC on the training and testing set. Conclusion: The proposed Temporal Neural Network model has shown better performance than other machine learning and deep learning algorithms. Due to this proposed deep learning algorithm can be successfully applied on the genes expression dataset.

Background: The block-chain is a growing decentralized scheme applied in many applications, such as an auction, smart contract, Health, and banking sector. The cloud service providers can sell the resource to cloud consumers using an auction. The main challenge in resource allocation using auction is to provide reliability to the users. Methods: In this paper, a blockchain-based combinatorial grouping auction with a reserve price mechanism (BCGAWRP) was proposed. The proposed scheme maximizes the total revenue and resource utilization by assuring reliability. The proposed BCGAWRP performance was assessed by simulating the cloud environment. Results: The experimental result shows that the proposed BCGAWRP algorithm increases revenue more than the traditional combinatorial auction algorithm. Conclusion: Moreover, simulation studies show that reserve price is useful and provides a mechanism to achieve the trade-off between the seller's and the buyer’s virtual machines.