International Conference on Innovative Computing and Communications

Fuzzy modeling is a challenging task and becomes more complex when designing T2FS, which requires identification of more parameters as compared to T1FS. The problem of fuzzy modeling can be expressed as a high-dimensional search and optimization process, and EAs have the ability to search for optimal solutions in high-dimensional search space, so researchers used various EAs for fuzzy modeling. GAs are widely used for finding solutions in large search spaces, and MAs have characteristics of both global and local optimizations. This paper describes how to use MAs and GAs to identify IT2FS, including how to build MFs for both input and output, as well as how to generate a rule base from a data collection. The efficiency of T1FS and IT2FS for noisy data is also compared with GAs and MAs in the paper. For comparison, we consider four different problems: a rapid Ni–Cd battery charger, data from Box and Jenkins’s gas furnace, and the iris and wine classification datasets. In the presence of noise, the results imply that IT2FS is more efficient than T1FS, and MAs are more efficient than GAs.

There are a lot of challenges for mobile ad hoc networks (MANET) in the present scenario concerning certificate revocation. Suppose if there is no dynamic access to the central authority, then the certificate revocation of the malicious node is very much crucial. The spoofing of certificates by the intruders will create more threat to the secure communication system. In this paper, we propose to develop a secure multipath Optimized Link State Routing (OLSR) mechanism integrated with certificate revocation and trusted route re-computation mechanisms for MANETs, which helps to overcome these issues. According to the trust value, each node assesses the behavior of its neighbors. The proposed certificate revocation and the route re-computation mechanism minimize the overhead in multipath OLSR. As per the simulation results, the proposed approach could outperform the existing approaches in detecting the malicious nodes.

In the pandemic COVID-19 situation, the world is facing a pandemic of fake information which often stirs the public attention by attacking their emotional quotient. Scenario reached a situation where people in search of worthy information for public health and precaution, getting fake news. This unprecedented expansion of fake information has become a challenging research issue. Deliberate efforts have been attempted in this manuscript for finding a solution to this COVID-19 fake news detection problem with the help of deep learning models. Two deep learning models—BERT, a transfer learning model, and attention-based bi-directional long short-term memory (LSTM), a memory-based model, have been applied in order to get accurate fake news classification outcomes. A comparative outcome of both models is presented which shows BERT outperforms and gives excellent results in comparison to the attention-based bi-directional LSTM model. The achieved training accuracy by BERT is 86% which is much higher than the accuracy achieved by attention-based Bi-LSTM. BERT precision, recall, and F-score are 0.82, 0.79, and 0.80, respectively, which shows that BERT can detect COVID-19 fake news better than the attention-based Bi-LSTM model.

Device-to-device communication is an innovative paradigm which enables user equipment to communicate directly with other user equipment with or without the involvement of network infrastructure. It is an inevitable part of the Internet of Things. Hence, it makes wireless networks more spectrum and energy-efficient with traffic offloading. However, the massive growth of number of devices and the corresponding heavy data traffic generated at the edge of the network created additional burdens on the cloud computing due to the bandwidth and resources scarcity. Hence, edge computing is emerging as a novel strategy that brings data processing and storage near to the end users, leading to what is called edge computing-assisted device-to-device communication. This paper conducts a comprehensive survey on different techniques developed to enhance the performance of device-to-device networks by enabling edge computing capability for the devices in the communication network.

Introducing software-defined networking and network function virtualization has brought new opportunities in cloud, enabling dynamic and autonomous configuration as well as provisioning of resources in the cloud. In this article, we highlight how OpenDaylight can be integrated with OpenStack to enhance network services. After describing SDN with NFV and how they compromise cloud services, we present some advantages of SDN in cloud networks. In addition, the deployment of an SDN solution on a high-performing cluster is presented. Finally, this cluster is integrated into the cloud OpenStack to enrich its services.

Knowledge is the main discussing and explored topic of today’s era. Everyone is working toward improving information and tries to consider it as a ladder to move forward. Data is the main object to get information, and data is considered as a big data nowadays as it contains numerous information in all directions. As knowledge is bliss, it is also possible that an adversary can use this information to harm an individual. To protect data from an adversary privacy preserving data publishing techniques is used. But when multiple sensitive data present in a data set which is correlated to each other’s several model are unable to protect data in an efficient way. In this paper, a novel model K-MNSOA is proposed for privacy preserving data publishing, which protect sensitive data privacy breach, even if the data set contains multiple sensitive numerical overlapped attributes. A proposed model assumes that all sensitive attributes are not actually sensitive, so when data is protected, information loss will increase. To overcome this issue, new model suggests to divide sensitive data into levels of sensitivity and apply generalization only for the privacy of high sensitive attribute.

5G or fifth generation is the latest in the communication technology which is being researched worldwide as a successor to the current 4G technology. 5G operates on higher bandwidth with higher data rates of the order of Gbit/s. 5G is estimated to play a major role in the development of smart cities and IoT use cases. Lumos 5G is one of the groups researching on the topic. In this paper, the throughput obtained under various conditions is analysed as a regression model in machine learning with the features as continuous variables. It is observed that the newer tree machine learning models are performing better on the dataset than the traditional tree models. This is verified by performing a tenfold cross-validation check on the best performing models.

The increased access to the Internet on smartphone has enhanced the possibility of exposure of children to content inappropriate for them. A smartphone app to automatically record and analyze inappropriate online material was developed and provided to children of three age groups, viz. four to six, seven and eight and nine and ten years. The smartphone app determined the time spent using smartphones, and the number of times adult text, adult graphics, violent text and violent graphics was accessed by children in a week on an average. One-way multivariate analysis of variance was used to find out significant differences in the five parameters among the three age groups. Results showed that children aged nine and ten years spent 193.37 min on smartphones and accessed content comprising of adult graphics and violent graphics on 3.93 and 30.63 times per week on an average. Younger children aged four to six and seven and eight years were found to use smartphones less with 128.37 and 151.17 min on an average, respectively, and get exposed to inappropriate content less frequently. Children in all the age groups got exposed to inappropriate graphical content more frequently as compared to inappropriate textual content. The app facilitates awareness of parents about the online activities of their children. Timely intervention of parents may prepare children to counteract unpleasant online experiences. This cognizance of children will thus empower them to benefit from the abundant wealth of information available online via smartphones.

An enormous amount of data available on the e-commerce sites are of different forms as ratings, reviews, opinions, remarks, feedback, and comments about any item, and it is difficult for the system to search the user interest and predict the user preference. The recommender system (RS) came into existence and supports both customers and providers in their decision-making process. Nowadays, recommender systems are suffering from various problems such as data sparsity, cold start, scalability, synonymy, gray sheep, and data imbalance. One of the major problems to be considered for better recommendation is data sparsity. Cross-domain recommendation (CDR) is one way to address data sparsity problems, cold start issues, etc. In the most traditional system, cross-domain analysis is used to understand the feedback matrices by transferring hidden information and imposing dependencies across the domains. There is no vast comparison of existing research in CDR. This paper defines the problem, related and existing work on CDR for data sparsity and cold start, comparative survey to classify and analyze the revised work.

This study describes the deployment of an image processing approach for finding COVID-19 affected lungs. Medical scans are useful in diagnosing illnesses and determining if organs are working normally. Medical image processing is an ongoing research subject in where numerous ways are used to help diagnosis, as well as different image processing techniques that may be used. Picture processing was used in this work, which includes image pretreatment, histogram leveling, smothering, eroding, and dilation. The usage of 2-bit picture is selected since this characteristic is well-known and there are several resources accessible. The Open CV library, which includes a plethora of image processing functions, is likewise free to use. Our experiment has shown how COVID-19 affected lung disorders can easily be identified with the help of a 2-bit image segmentation technique. The plan comprises (1) using a deep robust acquisition access to portion proper regions of interest from bleak medical examination image sizes of 903 total, (2) using a propagative neural network to improve contrast, sharpness, and illuminance of image contents, and (3) from the beginning to the conclusion, a regression strategy plan was used to accomplish medical picture categorization by material design in deep neural networks.

Nowadays, human health is paramount to any other thing in the world. But health is affected due to many reasons. Doctors and scientists are constantly working to find solutions to health issues. The main issue is blood-related problems because blood is the foundation of our body. Cancers related to blood are very critical and cause human death. Leukemia is a kind of most cancers that arises inside the bone marrow and outcomes in an excessive wide variety of peculiar white blood cells. If acute leukemia cannot be treated in a short time, there is less time for humans to survive. It is important to detect cancer at an early stage and be able to treat it. It takes more time to cure so early detection is vital for treating cancers. In this research, machine learning is used to predict cancer cells in the blood. So, we used convolutional neural networks (CNN) to train the model and find cancer cells in the blood at an early stage. This research shows the prediction of blood cancer cells and displays the differences between the normal and cancer cells image using CNN classification process.

The aim of this study is to undertake an empirical inquiry and comparison of the effectiveness of various classifiers with ensembles classifiers in the prediction of student academic performance. A single classifier algorithm will be compared against the performance and efficiency of ensemble classifiers. Reducing student attrition is a major problem for educational institutions all over the world. The search for solutions to increase student retention and graduation rates continues for educators. This is only possible if at-risk students are identified and intervened with as soon as possible. However, the majority of regularly used prediction models are inefficient and inaccurate as a result of inherent classifier limitations and the inclusion of insignificant inputs in their calculations. The majority of data mining and machine learning researcher focused on developing an algorithm that can extract useful information from massive amounts of data after being processed by a computer. The most difficult problem in predictive modelling is identifying the most effective prediction algorithms that are also accurate enough to be useful. Therefore, a multi-level homogeneous ensemble predictive (MLHoEP) model is designed, which uses the different techniques of data mining like feature selection, ensemble learning techniques like boosting and bagging. Seven distinct machine learning algorithms were used on this model to predict and analyse the academic performance of the students. The performance of the classification algorithms in terms of prediction was evaluated using k-fold cross-validation. The study contributes to the body of knowledge by suggesting the development of homogeneous classifiers that may be used to accurately predict students’ academic success. It also proposes the construction of homogeneous classifiers, which may be deployed for accurate student performance prediction, in order to provide a better explanation for the poor performance prediction. As a result of this research, it has been demonstrated that the technique of applying homogeneous ensemble approaches is incredibly efficient and accurate in terms of predicting student performance and assisting in identifying students, who are in danger of dropping out of school. The study compared the accuracy and efficiency of single classifiers to ensembles of classifiers in terms of performance. It was discovered in the research that a homogeneous model with excellent accuracy and efficiency might be developed for anticipating student performance. These key problems have been successfully addressed by the findings of this research study: Which characteristics of students are the most effective predictors of academic performance? How accurate are approaches such as bagging and boosting ensembles for predicting student academic performance? The approach offered in this study will aid educational administrators and policymakers in designing new policies and curriculum-linked to student retention in higher education. This research can also aid in the identification of students who are at risk of dropping out of school early, providing for timely intervention and support. Prospective research will examine the creation and implementation of an automated prediction system known as the students’ academic performance forecast framework, which will collect data from students via online submission and produce a prediction result for their academic performance.

Blockchain adoption has been at all-time high in last one decade. With widespread acceptance, blockchain is also subject to criticism from deployment to sensitive data exploitation in blockchain applications. Blockchain wallet is an essential part of blockchain framework and contains the most sensitive data of blockchain application. In 2020, out of 122 blockchain attacks, 27 were found to be solely focused on the wallet. Even though the comparison and analysis of various types of wallets are available on the market, there is no collective mapping of the best wallets in use. This study delves into wallet types, how they work and a comparative analysis of the state of the art on wallets that are currently in use. Presented research study is also helpful in designing a wallet and also gives assistance in choosing a wallet according to the requirement. The findings seem useful for developers and end-users at large, as it brings about the features of the current wallets in use and also proposes a novel design for blockchain wallets.

AI provides the concept of machine learning that helps to automate the decision-making process by analyzing data inputs. It trains machines by providing it sample data and thus makes the system intelligent that is helpful for real-world AI applications. Machine learning algorithms are applied to such social feedback data to excerpt useful information that confers a competitive edge to several enterprises. There are enough machine learning technologies in the existing literature on sentiment analysis. However, it still needs optimizations for a better decision making process for several enterprises. In this paper, we proposed a scheme for Indian Railways for determining sentiments from Facebook. This is a more specific scheme that clouts business intelligence over different classifiers, viz. SVM, NB, RF, and decision tree, K-NN. The proposed scheme is provided with various parameters like F-measure, recall, precision, logarithmic loss, and accuracy. The first section of this paper provides the preface of sentiment analysis, and the next section presents the related work and motivation for sentiment analysis then methodology adopted for better decision making through machine learning to bring out in depth knowledge for future marketing game plans; it then discussed the experimental results, and finally, the paper encapsulates the conclusion and future scope in the area of sentiment analysis.

Community detection has become pervasive in understanding complex network structures and detecting similar patterns. The main motivation behind using deep learning methods for community detection comes from the brilliant performance results shown by deep neural networks in various fields. Using unsupervised learning models, the problem of community detection can be solved. The high-dimensional feature space representation of the network data leads to a complex neural network architecture that requires a high number of trainable parameters. Deep learning-based models can transform the high-dimensional graph data of complex networks into simple, low-dimensional space or latent representation. The transformation of network representation to latent representation consists of meaningful features of the network data. This mapping preserves the structural information of the network later on, which clustering algorithms can be applied to the converted latent representation. This survey paper provides an overview of the traditional and deep learning-based methods of community detection, followed by a discussion on the challenges and future directions of community detection.

Heart patient number is escalating day by day, and numerous individuals lost their precious lives each year due to sudden heart attack in all over the world. Because of this, before time diagnosis of cardiovascular disease is necessary to prevent death. Some technology-based software is required to help in medical field to recognize heart patients with more accuracy and lesser time. Huge amount of heart patients are present in different hospitals in all over the world, which can be used efficiently to diagnose the heart disease by applying data mining techniques. In the process of data mining, knowledge or useful information is extracted among the large sets of raw data. In the prediction analysis, machine learning techniques are applied to discover valuable patterns and forecast future events or trends. This research work will predict the likelihood of coronary heart disorder in patients by implementing a modified machine learning algorithm. The input data are passed through various procedures comprising preprocessing, clustering, and selection of effective attributes before classification. To determine the heart illness, four algorithms which include random forest, K-means, genetic algorithm, and logistic regression are assimilated. In this technique, the irrelevant attributes of heart dataset are discarded to improve the performance and to decrease the training period time. This process is completed by random forest technique. K-means clusters are optimized by genetic algorithm in order to group all the outlier data points. At last, logistic regression is applied to classify the patients based on the heart disease. Performance comparison among various existing techniques has analyzed on the basis of some performance measures. The calculated accuracy increased up to 95%.

At the beginning of the fall of 2020, the campus network poses new challenges due to the outbreak of the COVID-19. Entire entities of campus are scrambled to set up remote learning. In turn, the number of users and devices on the network multiplies in a tremendous way. This enlargement forces network administration to control and verify the accessibility. To address this subject, this paper puts forward the software-defined perimeter (SDP) integrated with the software-defined campus network (SDCN) framework. SDP controller is united with SDN controller in SDN control plane to yield authentication and access control for network. SDCN with SDP provides strong prospects in minimizing unauthenticate access ratio which strengthens the trust factor of legitimate users and enhances quality of service. SDP with SDCN helps in enabling additional boundaries within the network which acts as a defense, enhances scalability, and shields the network from external attacks as well as from internal malicious users.

Internet of Medical Things (IoMT) is a smart interwoven technology enabled by the advancements made in multi-disciplined fields of medical devices, networking technologies, healthcare applications and artificial intelligence. The current spread of the coronavirus disease (COVID-19) globally has thrown innumerable challenges against human survival. To overcome this pandemic situation, an innovative healthcare solution is vital for saving human lives and mitigating the viral spread. We propose an E-Health+ system that can provide remote patient assistance anytime, anywhere. E-Health+ makes use of artificial intelligence in edge nodes for data processing coupled with Federated learning for swift prognostic medical advice for connected patients during their critical times in IoMT. The medical advice or assistance provided is based on the requests arising in a real-time basis with minimal response times, thereby reducing latency and also the much-needed privacy preservation towards the sensitive patient data.

Diabetes is a condition in which blood glucose, called as blood sugar, is high in an abnormal way. If the prediction of disease is possible at an early stage, then the risk factors associated with diabetes can be considerably lower in severity. The main problem and highly challenging task are to predict diabetes accurately, and the reason of this challenge is the diabetes dataset’s insufficient number of labels data and the existence of outliers. This research paper proposes a strong framework to predict the disease with the help of different types of machine learning (ML) algorithms: K-nearest neighbor (KNN), support vector machine (SVM), decision trees (DTs), Naive Bayes (NB), and logistic regression (LR). For implementation, a dataset has been taken from a PIMA database consisting patient’s health record, and these five machine learning techniques are applied to that dataset. A comparison between all the algorithms is presented in this paper. The motive of the paper is to provide assistance to doctors with their practitioners for the early prediction of diabetes using ML algorithms.

As the technology is growing at a rapid pace, there is an increased demand for various cloud resources and resultant of which is the establishment of large number of cloud data centers (CDCs). A single cloud data center consumes large amount of energy, and eventually, it will lead to the higher operational cost and emission of carbon. To reduce the consumption of energy with better utilization of resources, different virtual machine (VM) and its consolidated approaches have been considered for the dynamic utilization of resources. In this paper, proposed enhanced artificial bee colony (PEA) has been proposed for better migration and placement of various VMs and physical machine (PM) dynamically. There are two distinct phases in this algorithm. Firstly, selection for the location of PM with access delay to the location where it needs to be migrated and secondly, reduction in number of VM migrations. Further, proposed approaches are compared to in terms of SLA-V, energy consumption, number of hosts shutdown, and resource utilization. Results show the gradual reduction in SLA-V by 20 and 31%, number of migrations by 16 and 25% and increase the resource utilization by 8%. There is a better improvement of 13% in energy consumption has been observed in the proposed method compared to others.

The objective of this work is to identify clinical factors that modulate the risk of progression to lung diseases such as asthma, chronic obstructive pulmonary disease (COPD), emphysema, lung cancer, bronchitis, and allergies among patients using data extracted with assistance from machine learning algorithms. In this work, we have gathered 250 instances along with 14 attributes. These information have been gathered from patients experiencing various lung illnesses alongside different indications. The lung illnesses trait contains two sorts of class which are ‘Positive’ and ‘Negative.’ ‘Positive’ implies that the individual has lung illness. The dataset has been trained using K-fold cross-validation technique. Four machine learning algorithms have been used for analysis which are logistic regression, random forest, KNN, and Bayesian networks.

The utilization of multilevel inverter since the last decade has been increased. Due to the ability of these novel inverters to synthesize, the waveforms with better harmonic profile and output are used in number of high voltage and high power applications This paper depicts an asymmetrical 23-level multilevel inverter, using sinusoidal pulse width modulation technique. With increased steps in output voltage, the total harmonic distortion (THD) is reduced. This topology is anticipated to get 23 levels. Simulation results are shown and compared with theoretical results. This topology is proposed to give more number of levels with minimum possible switches, which is more efficient.

The Internet of Things (IoT) is nowadays having outstanding developments in the IT industry and for research purposes. However, it actually experiences security and protection liabilities. Traditional security and protection approaches will, in general, be irrelevant for IoT, principally because of its decentralized geography and the asset imperatives of most of its gadgets. Blockchain that supports the cryptographic money Bitcoin has been as of late used to give security and protection in distributed networks with comparative geographies to IoT. In the IoT, ensuring the validity of the identity of a device accessing the network is the basis of security. A computer network consists of nodes. These nodes are linked through the communication links. The nodes are distributed geographically, and the purpose of the network is to transmit between different types of nodes. These nodes can be personal computers, workstations, sensors, etc. Users carrying wearable devices move from one place to another, and there is a possibility of linking to more than one network. That might raise the anomalies of the information in the network. We aim to remove the anomalies by validating the node with the help of blockchain technology. This paper proposes a system and method for verifying identity using IoT and blockchain technologies. The proposed model will detect anomalies by comparing node traffic profile and behavioral information of the node in the blockchain to the observed node activity.

Increase in connectivity and cost pressure has pushed Industry 4.0 to rely on the systems built over Internet of Things (IoT). These IoT devices are susceptible to cyber-attacks. Intrusion detection system (IDS) protects such IoT devices from such attacks. However, the IoT devices are wreaked with the high-computational costs and curse of dimensionality. Current study presents an intelligent IDS system which is able to reduce the unwanted features. Proposed IDS system shows an accuracy of 99.26% with a precision over 99% to identify the attacks from the CICIDS2018 dataset.

Soft computing-based several techniques had already been applied previously in various industrial applications. This paper tries to provide application of various algorithms on job scheduling problem. The expected future needs of industry are based on proper application of these algorithms. In single objective optimization, point is to discover a schedule that limits general culmination time known as makespan. The paper represents comparative study of algorithms shows calculation of reduced makespan. Modified computation allocates jobs precisely than GA. Thus after applying, it is found that performance of hybrid genetic-cuckoo search algorithm approach is effective in finding ideal solutions contrasted with that of different methodologies.

In computer vision, 3D object detection has numerous applications such as robotics, augmented reality (AR), medical field, manufacturing industries, and safe autonomous driving. But the real-object detection may involve various problems such as noise, missing data, and occlusion problem. From past few years, the great progress in 3D object detection has been made. Object recognition and identification in occlusion remain a difficult challenge, despite recent breakthroughs in 3D object detection. The occlusion problem is one of the difficulties in object tracking. The paper highlights a number of research hurdles and open concerns that researchers must address.

The Internet of things (IoT) is growing more prevalent and popular in recent years as a result of interconnected entities that allow millions of gadgets to talk with one another. We can call IoT an evolution of the Internet, and it has gotten. In recent years, there has been a lot of interest from researchers. The IoT is significant because it allows many low-resource and restricted devices to communicate, calculate, and process many operations and make decisions in the communication network. With each technological improvement, the creation of intelligent systems with high communication and data collection capabilities becomes achievable, opening up new opportunities for a wide range of IoT applications, notably healthcare systems. On the other hand, everything is useful and is not without problems and challenges. There are several difficulties to deploying IoT in the real world, ranging from tiny sensors to servers, such as interoperability, portability, accessibility, privacy, and information security. In this article, we present a complete survey of IoT technologies, processes, statistics, and success stories applied to healthcare, with an emphasis on the security threats and needs of IoT cryptography, technology, and IoT device trends.

The main cause of the death as per today’s scenario is brain sicknesses that effect the human’s body activities along with brain cells. The detection of brain diseases is still a challenge due to many factors like user-friendly diagnoses interfaces, test accuracy, early detection stage, and many more. Keeping all this in mind, this paper proposes a hybrid approach for detection of brain tumor and eliminating all other replica factors. We use TDE hybridized with bisection mechanism that having history in illnesses detection. In our study, this approach shows a betterment in comparison with existing techniques with 8–10% margin. It is our understanding the use of objective function at classification and segmentation phase as well as enhancing contrast will show this improvement. These results are shown through parameters like MSE and PSNR in relationships of precision.

COVID-19 has immensely affected our routine lives and has made us live in a panic driven environment for the past few years. RT-PCR test is the most preferred test for the detection of COVID despite being a time-consuming process. Research has shown that CT scan of lungs can be used to identify COVID infected people in a shorter span of time. The proposed model uses a U-NET-based two staged detection model to segment the abnormal region from the CT scan images. The segmentation module consists of U-Net which produces a binary segmentation mask of regions infected by COVID. The U-Net was trained with dice coefficient loss to improve the structural similarity between ground truth regions and predicted regions. The segmented regions are used to classify COVID contraction. Multiple classification models such as Xception, ResNet, and Inception-ResNet were tested to classify segmented regions. Xception model outperformed other models, produced an accuracy of 98.77%, and thus was chosen as our backbone network. We compare the results obtained on detection using complete CT scans and segmented CT scans to prove the relevance of using segmented regions for COVID prediction.

COVID-19 has been a major global challenge these days. The pandemic has changed human life, attitude, and behavior. This pandemic added a burden to people’s life and health. With the new variants of SARS-CoV-2, a lot of people are even scared of going to the health centers to get the COVID-19 evaluation in fear of contamination and contagious, which caused the surge in the symptoms at later stages. Data collected across various sources can play an important role in predicting and identifying of COVID-19 virus based on the models and the classifications of this data using the most sophisticated machine learning models. The concern here is accessing or transferring an individual’s data from their personal health devices which defers users’ privacy. In the recent past, there are a lot of research that has been done these days on how blockchain can help to securely track and transfer the data across trusted sources. Adding to this, federated learning also is helping on-device data usage without any critical data to be transferred to various external sources. The proposed study directs the stability of frequent health status with the help of wearable devices that capture health metrics like heart rate, blood oxygen levels, breathing rate, muscle activities, stress, emotions, movement patterns, sleep activity, precipitation, and mind/cognitive functions with the introduction of the data streams and models that can seamlessly transfer the data, with the assurance of data integrity, privacy, and control which is the scope of this paper. The usage of both the emerging technologies provides a value addition in terms of health data exchange with effective data distribution with decentralized privacy and computation. We have also introduced a consent-based personal health device registration mechanism on a blockchain consensus network with digital identity to allow and take back controls over who can access their data. We believe that this solution and the implementation would help everyone to predict the possible COVID-19 infections keeping data privacy at the most priority.

Coronavirus (COVID-19) is an air-borne disease that has affected the lifestyle of people all around the world. Tracing patients infected with coronavirus has become a difficult process because of the limitation of tests based on reverse transcription-polymerase chain reaction (RT-PCR). Recently, methodologies based on imaging have been proposed by various researchers especially using deep learning-based models for the detection of COVID infection. This paper analyzes the effectiveness of deep features for COVID detection from CT scan images. Deep features were extracted from the final layers of deep learning models which are then fed into machine learning frameworks for classification. Transfer learned features obtained from ResNet50, Inception V3, and EfficientNetB7 were employed for the study. A combination of Inception V3 and SVM gave the best accuracy of 86.12 and precision and recall with 83.11 and 80.44, respectively. These results are comparable to recent transfer learning approaches and architecture that is about to be discussed is having an advantage of minimized time when compared to traditional deep learning approaches.

Our planet Earth is presently being disturbed by a variety of environmental concerns. One of the top critical environmental issues affecting our planet’s ecosystem is oil spills. Oil spills mostly occur due to ship leakage which highly influences our food supply chain and leads to a high-level drop in the economic division. Therefore, monitoring and tracking those vessels are extremely vital to determine the responsible ships for the occurrence of an incident. This study revolves around an implementation of an automated ship detection software application by building a high-level algorithm that embeds deep learning networks. The algorithm is built in a way that can predict and classify vessels from high-resolution satellite images with 98.5% accuracy.

A convolutional neural network (CNN) has one or more layers and is mainly used for image processing, classification, segmentation. CNN is commonly used for satellite image capturing or classifying hand written letters and digits. In this particular project, a convolutional neural network is trained to predict whether a person is wearing a mask or not. The training is done by using a set of masked and unmasked images which constitutes the training data. The performance of the trained model is evaluated on the test dataset, and the accuracy of the prediction is observed.

The Internet of Things (IoT) has revolutionized the ways; the physical world is connected to the cloud for real-time data dissemination through embedded sensors and microcontrollers. IoT plays an important role in almost every sphere of the world, may it be physical sensor connection to the cloud, the Structural Health Monitoring Systems (SHMS), Smart Homes, Health care, etc. Healthcare Internet of Things (H-IoT) has taken the healthcare sector to the next level by incorporating remote patient monitoring and diagnosis, Robotic surgeries, patient’s vital data monitoring in real time, etc. This paper presents a novel and a simple technique of remotely monitoring patients suffering from highly contagious diseases like the Corona Virus, thereby reducing the direct patient–doctor physical contact and ensuring the social distance. The patients’ medical data is acquired and end-to-end encryption is done on the data to ensure no loss of the data between the transmitting end and the receiving end. The designed system is based on the Node_Mcu microcontroller platform. The sensor data is acquired and processed using the Arduino-Integrated Development Environment (IDE) and further predictions regarding the patient’s health are performed in the MATLAB 2019a computation software.

Deep learning (DL) has made significant progress in identifying and classifying plant diseases. The convolutional neural network (CNN) model was utilized to classify diseased and healthy tomato plant leaves for this study. Seven predominant DL models, namely LeNet 5, AlexNet, VGG19, Inception Net V3, ResNet50, DenseNet 121, and Efficient Net B0 have been used for tomato leaves disease classification. Deep feature extraction and fine-tuning strategies were utilized to adapt these DL models to the specific task of classification. The obtained features using deep feature extraction were then classified by fully connected layers of the CNNs. The experiments were carried out using the image data acquired from the Indian Agricultural Research Institute, India. The dataset consists of diseased and healthy tomato leaf images with a total count of 155 images. Data augmentation was used to increase the dataset size. Furthermore, three segmentation algorithms were also applied to remove the background and highlight the deep features. In this study, a comparison of the above-mentioned CNNs has been carried out to show the accuracy results achieved on the collected dataset. The evaluation results show that deep feature extraction with image segmentation techniques produced better results (up to 100% classification accuracy) than without segmentation. The outcome of this research will have a substantial impact on tomato disease prediction and early prevention.

A novel palmprint recognition system (PRS) using deep supervised learning (DSL) classifier is proposed in this research work. To divulge the novelty, a deep supervised contour fractal dimension analysis model for palmprint recognition (DCFPR) is put forward. That has a novel region-based contour fractal dimension (RCFD) feature extraction approach and a deep supervised Learning (DSL) classifier approach for acquiring the higher recognition and identification accuracy rate. To accomplish the RCFD approach, traced all the edges/contours of 2D palmprint region of interest (2D-PROI) image using Canny edge detection algorithm and then split into several regions. At each region, fractal dimension (FD) and the slope value (S) are computed in an idiosyncratic manner using the box-counting procedure and then accumulate all FDs and Ss of all regions to create a distinctive feature vector. Classify this feature vector using deep supervised learning (DSL) classifier approach to authenticate the genuine person of the taken palmprint at a higher accuracy rate. In this research, the multi-spectral 2D-PROI image database derived from PolyU, Hong Kong Polytechnic University, Hong Kong. The proposed model has been examined and evaluated with various metrics and found with 98% of authentication accuracy.

Automatic text summarization involves extracting relevant details from the contents of input text documents for generating summaries. This area of Natural Language Processing is widely researched, especially with popular languages like English. There is a need to extend this work to less commonly spoken languages of the world. This paper presents a language-independent text summarization approach using Latent Semantic Analysis in Konkani language. Konkani is a low-resource language with limited language processing tools, stop-word list, etc. Latent Semantic Analysis (LSA) is an unsupervised algebraic method that finds latent semantic structures to be used for performing extractive text summarization. We examined well-known Latent Semantic Analysis-based sentence selection approaches on our dataset, constructed using books on Konkani folk tales written in Devanagari script. The results of the experiments indicated that LSA-based approaches can produce promising summaries, with the Cross method performing the best in most metrics.

Automatic summarization of text documents is a widely researched domain in natural language processing. A lot of research is carried out on the most commonly spoken languages in the world. Automatic text summarization needs to be explored to include some of the less popular languages in the world to help sustain such languages and promote their use. A language-independent summarization system that can be effortlessly extended to other such languages, which could have a limited number of resources to carry out such research is required. In this paper, we examine the efficiency of supervised linear regression models for the performing single document extractive automatic text summarization on Konkani language folktales dataset. We use 13 language-independent features and linear regression models to learn feature weights. These weights are then used to calculate a sentence’s score; top ranking sentences are then chosen for summary generation. We employ a k-fold evaluation strategy to evaluate the system-generated summary against a human-generated summary using ROUGE evaluation toolkit. Additionally, we also evaluate the use of L1 and L2 regularization on the summarization task. The work represents early attempts in automatic text summarization pertaining to Konkani language, and the dataset employed in these experiments is unique and devised particularly to facilitate research in this domain. The language-independent features used can be readily extended to other low-resource languages. The systems implemented in this work performed better as compared to an unsupervised system based on k-means approach and also beat the baseline systems.

In recent years, there is development in the field of computing devices; Internet of Things (IoT) becomes the latest trend. IoT comprises ubiquitous things that are associated with day-to-day life of individuals like smartphones, smart TV, laptops, and now vehicles too. Internet of Vehicles (IoV) has become the latest area of research used to develop applications in the field of traffic management and road safety. A collaborative approach of cloud and edge computing is termed cloud-edge computing. To manage the enormous amount of IoT devices and the coordination among IoT, the cloud and edge concept of computation offloading is required. In the process of computation offloading, tasks are computationally offloaded to the cloud data center that enhances the resource utilization of the cloud server and minimizes the energy utilization for the tasks. This paper represents the literature review related to various computational offloading schemes in cloud-edge computing proposed as part of the study. The resources comprise of related book chapters and research papers from different publishers of international and national reputes. The study is carried out with the analysis of various computation offloading schemes in cloud-edge computing for the Internet of Vehicles. In addition, computing technologies like cloud computing, edge computing, and computation offloading for the Internet of Vehicles (IoV) were also discussed.

One of the most important topics for society is human health care; to find the appropriate diagnose or correct diseases, detection is the primary key to get appropriate care; traditional technique is facing many challenges from delay or unnecessary treatment to incorrect diagnoses which lead to a diagnostic error that can effect on the treatment progress, increasing the bill, and give more time to the disease to spread or affect and harm the patient body. Those such errors could be avoided and minimized by using machine learning algorithms. In recent years, many significant efforts have indeed been developed to increase computer-aided diagnosis detection applications, which is a rapidly increasing area of research, and machine learning algorithms are particularly significant in CAD, which is used to detect patterns from medical data sources and making nontrivial predictions could assist the doctor and clinical in making decisions on time. This paper will discuss different ML algorithms that are used in diagnosing different diseases. Therefore, in this paper two major diseases have been chosen like cancer and heart disease, and the use of several ML algorithms applied their performance and accuracy.

The classification of text is one of the basic tasks of natural language processing with wide-ranging applications. This is essentially a process of assigning markers or categories to the text based on its content. The paper aims to use an improved Naive Bayes classifier to identify the fact-worthy sentence. In this paper, authors have implemented an improved Naive Bayes classifier through which we classify the sentences. This proposed method has been tested with the claim buster dataset contains 23,533 sentences where each sentence belongs to either of these three classes, i.e., non-factual statement, unimportant factual statement, and check-worthy factual statement.

The quantity of unstructured text data in digital archive is continually expanding due to the exponential growth of information technology, so the tasks of analysing, organizing, classifying, and summarizing text have become a big challenge. Since the manual classification of text documents requires a lot of human resources, finance, and time, automatic text classification is obligatory. Latent Dirichlet allocation (LDA) is an unsupervised machine learning algorithm often used in topic modelling. The output of the topic modelling algorithm can be used logically to classify documents. The LDA model is plagued with domain-specific terms. A novel latent Dirichlet allocation (LDA) with domain knowledge framework for document classification was introduced. The experiment was carried out using a dataset with five different categories of data. The experiments showed that LDA with domain knowledge gives better results than standard LDA and LDA using the TF-IDF model. Precision, Recall, F1-score, accuracy, and Purity were all improved using the proposed framework.

Unmanned aerial vehicles are real-time applications for flying networks which encourage multi-UAV structures. A multi-UAV system has a corporate behavior and can finish a mission efficiently. Coverage issues between ground stations and aerial vehicles are easily solved using FANETs. Because of the dynamic nature of flying vehicles, there are several security concerns. Routing protocols can upgrade and secure communication channels in flying networks. In order to increase the lifeline of aerial networks, a qualified routing research study is also carried out. According to simulation findings, zone routing protocol has maintained to secure communication channels and promises higher security without incurring computational expense.

Finding similar biological sequences to categorize into respective families is an important task. The present works attempt to use machine learning-based approaches to find the family of a given sequence. The first task in this direction is to convert the sequences to vector representations and then train a model using a suitable machine learning architecture. The second task is to find which family the sequence belongs to. In this work, deep learning-based architectures are proposed to do the task. A comparative study on how effective various deep learning architectures for this problem is also discussed in this work.

It is well known that visually disabled people often find it difficult to interact with their nearby environment. In this paper, we propose a design and show the real-time implementation of the smart embedded blind stick. The main components of the Smart Blind Stick comprise Arduino Uno, Ultrasonic Sensor, IR sensor, GPS sensor, and Buzzer. Here, the ultrasonic sensor and IR sensor are used for obstacle detection in the path of blind person, buzzer is used to make the person alert and GPS sensor is used to track the blind person if she or he lost their path. Arduino microcontroller will be used to control the whole scenario, where the smart stick is a closed-loop system to monitor the nearby environment continuously, and send us the output, by comparing the input result in the form of a buzzer sound.

With the notion of smart cities transforming cities into digital societies and making people's lives easier in every way, Intelligent Transportation Systems have become an integral element among all. The Intelligent Transportation System (ITS) attempts to improve traffic efficiency by reducing congestion and ensuring the safety and comfort of commuters in real time. Traffic sign detection and recognition is one of the multifaceted conjunctive fields of research in ITS. In this paper, we address the issue of the TSR (traffic sign recognition) problem, i.e., classification of traffic signs along the roadside which plays a crucial role in developing advanced driver assistance and autonomous driving systems. CNN's network design has a huge impact on its performance and convergence. As a result, we use the Genetic Algorithm (GA) to automate the task of selecting a high-performance CNN (Convolutional Neural Network) Architecture for the GTSRB (German Traffic Sign Recognition Benchmark) dataset. The model is optimized through GA using multiple network configurations in the search space. Our model takes into account the limitations of the dataset, and we use certain data augmentation approaches to address the issues. We were able to attain an average accuracy of 98.2% which demonstrates the state-of-the-art performance on the publicly available dataset.

As a result of the global spread of COVID-19, e-Learning has recently experienced extraordinary growth. Many educational sectors have made the transition from traditional classroom learning to virtual learning via various online platforms. In this epidemic, virtual learning has enabled all schools and universities to continue to provide education. This rapidly growing alternative modality necessitates the provision of robust and high-quality education. It is also important to figure out whether online learning satisfies the needs of pupils. Even if learning has become easier, many people still confront difficulties, poor connectivity and e-platform. This study aims to identify the students’ satisfaction by conducting a survey and analyzing it by means of data analysis and data visualization.

For human beings, irrespective of their profession, Internet connectivity became a necessity for survival in this digital age, but at the same time, privacy is becoming a growing concern for every Internet user. With rapid growing trends of personalized Internet activity, digital profiling is becoming a new normal, and all Internet servers are compiling users’ activities to provide better suggestions or recommendations. So, to avoid compilation of such activities, users are using facilities like portable browsers, private browsers, and tor browsers to remain anonymous. When similar facilities are used by criminals, then it is becoming a challenge for law enforcement agencies to investigate cases. Therefore, in this paper, we intend to analyze the portable browser in windows 10 pro in virtual machine for exploring the possible artifacts retrieval using digital forensics software. During analysis, it was found that various types of artifacts were available about the browsing activity from virtual machine hard drive and from RAM dump.

High-performance computer tools have been more widely available, and deep learning systems that utilize deep neural networks have become increasingly common in many fields. Deep learning approaches based on convolution neural networks (CNN) have become more widespread as high-performance computer facilities have grown. An overview of the growth of deep learning models and a concise explanation of various learning approaches, such as supervised learning, trains the neural network using labeled data. Solid experiments are required in medical image analysis studies to prove the efficacy of proposed approaches. Many architectures, such as Pre-trained Networks and Convolution Neural Networks CNN, are employed to achieve breast cancer diagnosis. Various classification measures may be utilized, making comparison of the methodologies challenging. Medical screening methods have grown increasingly important in the detection and treatment of diseases. Early identification of breast cancer is regarded to be a crucial element in lowering women's mortality rates. Several different breast screening modalities are being investigated to improve breast cancer diagnosis. Histopathology is used in a current cancer detection and localization method that uses artificial intelligence to screen for breast cancer and identify the existence of tumors in the breast. This study focused on an experimental dataset that employed convolution neural network (CNN) techniques to detect and localize breast tumors (i.e., pre-trained CNN). CNNs are a powerful tool for solving real-world problems, and neural networks with learning algorithms are a promising new technology.

The smooth transfer of user services from an existing network to a new network in vertical handover for providing better quality of experience (QoE) to the users is a challenging task in the field of mobility management. To achieve that level of the QoE, the subscribers of heterogeneous networks may be forced to change the access network or the network operator. For better user experience, network parameters like throughput, packet loss rate (PLR), cost, jitter, and delay, etc., are considered in vertical handover decisions. In this paper, a graph theory and matrix approach (GTMA) based on a multi-attribute decision-making (MADM) mechanism is proposed for the ranking of the candidate networks to handle the issues in vertical handover. The numerical analysis of the proposed approach is performed using conversational traffic in heterogeneous networks. The proposed approach is compared with gray rational analysis (GRA), and the comparative results have revealed that proposed approach is superior to GRA.

Image forensics is one of the most active research domains. As technology is advancing, we can add or take out crucial features from a picture without any trail of tampering. Therefore, its authenticity is called into question especially when images have impressive power. Copy-move is a kind of forgery, where portions of a picture are transformed and inserted into the same picture. Copy-move forgery detection is one such research domain that has put forward various methods to find out copy-move forgery. Many techniques based on image processing and machine learning have been put forward to detect the forgery. Since the duplicated parts are from the same image, many of the features will be similar to the rest of the image making it difficult to detect forgery using the latest methods. In this work, we propose to use SIFT keypoint-based forgery detection with clustering for quickly identifying copy-move forgeries in highly textured regions. As the SIFT keypoints are difficult to detect in smooth regions, we propose to use Hu’s invariant-based block-based forgery detection strategy to detect the missing cases. We show that the joint approach outperforms the method reported by Li et al. (IEEE Trans Inf Forensics Secur 10:507–518, 2015) on the popular copy-move forgery detection dataset MICC-600.

Classifying environmental sounds such as gunshots and dog barking are gaining popularity. Environmental sound classification(ESC) helps in developing context-aware applications such as security systems and criminal investigation systems. Research in speech and music has been done but environmental sounds are different because of their unstructured nature and attracts extensive attention in the field of research. Researchers have explored various preprocessing techniques, feature extraction and feature selection methods, and classification algorithms for ESC. In this paper, the ensemble technique—Adaboost algorithm— is applied to classify environmental sounds. The accuracy of different base estimators is evaluated on the publicly available dataset UrbanSound8K, and the highest accuracy is obtained in the case of the base estimator as random forest. The results of the Adaboost algorithm are also compared with the benchmark results reported using other machine learning classification algorithms such as support vector machines(SVM), IBK5, random forest 500, J48, and ZeroR.

A Blockchain-based smart contract is widely used in every domain for secure data exchange and data storage. The latest technology operates automatically, controls, or documents legally relevant events and actions according to the agreements described in the contract agreement. The Blockchain in health care can be visualized in managing electronic medical record (EMR) data, preservation of healthcare data, personal health record data control, point-of-care genomics supervision, computerized health reports data control, etc., by using IoT devices for data collections. This paper presents a brief overview of the well-known existing researches based on Blockchain-enabled intelligent contracts in the healthcare system. The paper focuses on existing research, methodologies, and future trends and comparative analysis of smart contracts (intelligent agreement) methods . This paper points out challenges and open problems that require discussion in the future considerations. Moreover, help new researchers to understand the upcoming trends in Blockchain-based intelligent agreements in the healthcare scheme.

Around the whole world, cancer is the most life-threatening disease. Basically, cancer can arise in any tissue of the body, and while each variety of cancer has unique characteristics, the fundamental processes that might cause cancer are highly common in all disease types. Breast cancer is one of the most ubiquitous types of cancer in females. In males, prostate cancer is the most dangerous during recent years. This study focuses on breast cancer as well as on prostate cancer in the direction of their early predictions. For early prediction, eight classification models had been used such as logistic regression (LR), Naïve Bayes (NB), decision tree (DT), stochastic gradient descent (SGD), K-nearest neighbors (KNN), decision tree (DT), random forest (RF), support vector machine (SVM), and artificial neural network (ANN). This work includes three different datasets for research analysis of breast and prostate cancer predictions. Two datasets for breast cancer (Coimbra and Wisconsin) and one for prostate cancer are taken from UCI and Kaggle repository, respectively. For improving the results of prediction, the normalization technique and feature selection method had been used in this paper. Performance in terms of accuracy, precision, recall, F1-score, and curves of each classifier are analyzed in this study. Most of the classifiers did well after using the feature selection method (ANOVA). In the case of Breast Cancer Coimbra, KNN give good results with 80% accuracy in both the cases with or without using feature selection. Logistic regression with feature selection doing the best work on Wisconsin Breast Cancer with 99% accuracy. There are four classifiers (SVM, RF, DT, and SGD) which gives highest accuracy (97%) on prostate cancer.

Cryptocurrency is based on blockchain technology which is ideally decentralised, referring to no superior authority overlooking it. The community is maintained by numerous user machines forming a “peer-to-peer” network. With the recent skyrocket of crypto-assets in the financial markets, many view it as the quickest and riskiest way to earn. Such assets are coined as “Volatile” due to its rapidly fluctuating price, thereby making it extremely hard to forecast its course. The paper at hand explores a novel technique that establishes a relation between signal processing and volatile stock forecasting methods via variational mode decomposition (VMD). Variational mode decomposition aided with BiLSTM neural architecture, a purely data-driven model, is fine-tuned to forecast the daily or interday prices of Bitcoin and Ethereum alongside yielded RMSE of 0.0278 for Bitcoin. The results are then further compared with ARIMA, ARMA and MA to analyse the effect of VMD.

In the current world, there has been an upsurge in the use of social networking sites like Facebook, WhatsApp, Twitter, etc. These are considered suitable sites for the exchange of messages and sharing pictures and videos. Besides providing entertainment to the users, sometimes the information circulating on these platforms may be fake or misleading. In this chapter, we reviewed the literature on AI technologies that address the issue of fake news detection, the process of information flow, different data sets to detect fake news, and future perspectives to improve the credibility of information.

The development of emotional intelligence in children begins during the early years of a child. Although it is the responsibility of parents to help a child in developing emotional awareness, studies have shown the utility of software systems in aiding this process. In this paper, the author presents the design of an emotionally intelligent chatbot for children. The outcomes of an online survey conducted among the parents reported that 70% of the respondents felt that an emotionally intelligent interactive chatbot can be useful for children to cope with intense subject matters related to low grades, no friends, bullies, and others. The study highlights various features of a chatbot like a user interface, personalization, responsiveness, security, and human intervention. From the findings, the author has suggested five design principles along with the detailed architecture of a chatbot framework. The paper will be useful for future studies that seek to design and develop a highly efficient emotionally intelligent chatbot for children which is trusted by their parents.

Protecting an organization’s intellectual property, financial secrets, and performance is crucial because it is sensitive data that if compromised could be catastrophic to the organization in question. As a result of the growing economy, organizations of scale have a significant portion of their infrastructure over technology which makes the organization vulnerable. The security teams of such organizations work to patch such vulnerabilities as they come across them but may spend a significant amount of organization resources fixing vulnerabilities that may not be exploited. After conducting our own research on the existing methods to prioritize vulnerabilities that have a higher probability of being exploited, we found that machine learning can be used to make the process of vulnerability prioritization efficient. This paper discusses our research on using machine learning for vulnerability prioritization and the different machine learning algorithms that can be of use for the same. This paper also discusses our approach on creating a system for vulnerability prioritization in an organization.

The main objective of this research is to analyze and compare the performance of machine learning (ML) and deep learning (DL) algorithms in detecting online hate speech. Therefore, Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Logistic Regression (LR), Convolution Neural Network (CNN), Recurrent Neural Network_Long Short-Term Memory (RNN_LSTM), BERT (Bidirectional Encoder Representations from Transformers), and Distil BERT algorithms have been explored and analyzed in this research. This research has applied the dataset on hate speech which was developed by Andry Samoshyn which is publicly available in Kaggle. ML algorithms and DL algorithms have got good scores in accuracy. In ML, SVM, RF, and LR have got top accuracy values. In DL algorithms, RNN_LSTM, Distil BERT, and BERT have performed well in accuracy. Based on F-measurement, DL classifiers have outperformed ML algorithms. Distil BERT has obtained the highest F-measurement scores. When we compare the overall performances, DL is performed well rather than ML in detecting hate speech. Especially transformer-based models of DL are more efficient than other DL and ML algorithms.

A virtual learning environment (VLE) is the type of environment that can attract more students because it allows them to study anywhere in the world, which means that the student's location is no longer a constraint. In addition, VLE facilitates access to teaching resources, which facilitate the monitoring of teacher activities and interaction between students and teachers. Therefore, the online environment can assess the factors that lead to an increase or decrease in the academic performance of students. Machine Learning approaches are used for the cognitive behavior and academic performance of students in Virtual Learning. There is still no decision on the parameters to be adopted for the evaluation of virtual teaching as each student may submit the same type of assignment and same Practical files, and can have the same attendance. In such a case, evaluation of a student’s academic performance became tough. So we need to adopt some LMS which records various actions of the learners and the teachers like Quiz Submitted On-time/Late, Number of Assignment Submitted On-time/Late, Number of Discussions attended, Number of CA attended, and Practical Submitted On-time/Late, Internet connectivity, etc. So, there is a need for a framework that accounts for all of these parameters’ consideration so that a Predictive model can be designed for Forecasting/estimation performance of students that are recommended system should be framed for enhancing the academic performance of the learner.

The abundance of technology followed by serious cybercrimes makes way for providing better security. Internetworking applications at a paramount level generate the need of securing host system as well as network system from the user having malicious intent. Several enterprises and organizations become victims of these severe attacks in the aspect of using many applications for providing safety like firewall, data encryption, and user authentication. By taking into account, this causes a detection system with the use of machine learning approaches of artificial intelligence which has been developed, known as an intrusion detection system (IDS). An intrusion is a process of entering into the system having the intent of unsolicited duplication, record alteration, and illegal access to confidential resources. Hence, analyzing the network packets of such cases for possible intrusions in near future is intrusion detection. The intrusion detection system has evolved as a vibrant topic for researchers in the last two decades and hope it will be in the future as well because of rapid advancement in technology day by day. In this study, a survey of various research papers has been depicted and it has been the utmost priority that the display of work shall be comprehensible.

Freedom of expression found on social media has various pros and cons. Gender-Based Violence (GBV) is also a major issue in social media. As a part of GBV, hate speech against women is on the rise on all social media. There are some lapses available in the stand-alone classifiers in detecting such speech, and the performance of ensemble classifiers is much better. Also, many research works have focused on common hate speech datasets. Hate speech against women has been used in very few research activities. But such hate speech is very dangerous. As a result, this research employs, Decision Tree (DT), Logistic Regression (LR), Random Forest (RF), and Long Short-Term Memory (LSTM) to compute metrics and performances and then use those algorithms to create a voting classifier to develop a more accurate model for detecting hate speech against at women. Two phases were used in this study. RF, LR, DT, and LSTM were used as foundation stand-alone classifiers in the first phase of the ensemble procedure. In Phase Two, the weights of the second-level classifier were estimated using first-level classifiers. Hate speech against women was detected using an open-source #MeToo dataset that was utilized for training and testing by the researchers. The dataset is publicly available on GitHub which was uploaded by Nazmus Sakib. This dataset consists of 278,765 #MeToo movement posts on social media. It clearly shows that the proposed voting classifier model has the highest values in all metrics including accuracy (89%). When we check the strongly positive classification, the proposed model has performed well in precision (0.90), recall (0.91), and F-measures (0.90) and it can calculate strong positive hate speech more efficiently than other stand-alone classifiers. This voting model takes more time to train since it has multiple models inside. By training it for more epochs, we can further increase accuracy.

Electroencephalography (EEG) is a technique to quantitatively measure brain activity with high temporal resolution. EEG converts brain activity to time series data with amplitude on the y-axis, and this data can then be used to understand brain functions. Mathematical tools can be applied to this data to extract features and to discriminate them in several classes. Once EEG data is recorded, it is needed to make sense of that data. In the past couple of decades, EEG data has revolutionised the healthcare industry and brain–computer interface (BCI) systems. This is made possible by continuous improvements in EEG data classification methods, which includes improvements in feature extraction and classification algorithms. In this study, methods to classify EEG data for various applications such as medical diagnostics, BCI and emotion detection are reviewed.

COVID-19 pandemic has disrupted the normal functioning of the world at both the physical as well as biological level. Various biotechnological approaches form the core for finding the drug for COVID. But at the backend, it is highly supported by intensive computational algorithms for drug designing approaches. Artificial intelligence is one such area that can be used to mimic the living system by training it with multi-dimensional datasets. The rapid advancements in the field of artificial intelligence and machine learning have facilitated to build high-accurate generative models. Deep learning is a subset of artificial intelligence that is being used in the present study for ligand designing models. The results obtained by a deep learning-based model called variational autoencoder—general adversarial networks (VAE-GAN) show promising results in terms of ligand design and can also be utilized for drug repurposing. In this paper, we have attempted to build a VAE-GAN model which was trained on isomeric simplified molecular-input line-entry system (Iso-SMILES) data and molecular structure images of the COVID-19 drug dataset. The Iso-SMILES and molecular structure data analysis are useful, but the system needs more improvement to cater to the data loss in terms of spatial structures and orientation of the chemical compounds taken for the analysis. However, the system can be optimized with the help of higher hardware support and increased training datasets, which can help in more precise analysis for generating ligand molecules of interest.

Knowledge retrieval from a large amount of biological database is one of the challenging tasks, nowadays. Numerous types of data mining techniques are applied to execute the same. For a few years, several researchers have established a lot of information retrieval procedures to extract knowledge from a wide-reaching amount of biological informations like protein and genes. In this paper, the authors try to make a brief review regarding these classification techniques along with their accuracy and computational time, which can classify protein into its family. The authors also try to mention the name of databases and procedures which are used to validate these classification approaches. In the end, a comparative analysis between these classification approaches was established alone with limitations and chance of improvement areas. Finally, a brief idea regarding the protein classification concept along with its need is clearly emphasized here.

Personality of an individual has been a promising variable to understand himself and furthermore the others in the society. It is the logical arrangement of an individual’s attributes like thoughts, feelings, attitudes, behaviour and capability that makes an individual selective. Our personality likewise influences our decisions, medical conditions, assumptions, inclinations and prerequisites. In the scenario of 4G/5G and COVID pandemic, the majority of individuals are dependent on the web gateways as their essential intuitive vehicle for their own and expert necessities; accordingly, it has been a fundamental significance for us to consequently perceive the personality traits of the individual on the opposite side of the screen. Mental analysts have tracked down that an interaction of just 100 ms is adequate to shape judgement about any individual. Thinking about a similar idea towards execution of profound learning for recognition of personality traits, in this work, we propose an intelligent model (iSMART), a combination of depth-wise separable convolution neural network (2D-CNN) and long short-term memory with attention (LSTMwA), that extracts audio and video features through parallel networks and predicts the ultimate personality score of a person. With the top to bottom trial and error, it has been seen that the depth-wise separable CNN reduces the quantity of trainable parameters without compromising the test precision. It is a compelling and lightweight model for recognition of personality traits utilising bi-modular data sources. It likewise accomplishes better accuracy as compared with the outcomes got by the top scoring teams in the ChaLearn Looking at People challenge ECCV 2016. Our proposed model can possibly empower the system with better psychological understandings and improved human–computer interaction.

Correction to: Chapter “Occlusion Problem in 3D Object Detection: A Review” in: D. Gupta et al. (eds.), International Conference on Innovative Computing and Communications, Lecture Notes in Networks and Systems 473, https://doi.org/10.1007/978-981-19-2821-5_26