Intelligent Systems and Machine Learning

Assurance in Quality deals beardly with multi sourcing as well as vendor expansion undertakings. The vendors present in the system are reviewed for extension of their agreement. There are fairly a big number of suppliers available for various items distributed as per directorates, by means of limited number of resources; it turns into hard to prepare review report in time. For requirement of transparent structure, different cases need to deal or handle equivalently as well as visits are being systematized. The dealer assessment device may offer with a transparent device by using that; selection can be done to allocate with ground visits for re-evaluation. The preferred additionally specify that require for improvement of supplier assessment system need to be categorical with retaining cost of vendor rating gadget against expected returns as well as other aspects in assessment. Consequently, the rating gadget primarily centered on four vital elements viz. Quality, Service, Delivery in addition to Machine might be well enough to achieve the necessity. The rating device can be employed in different decision-making equipment additionally.

Modern-day social media is one of the most used platforms by millennials for sharing personal, and professional events, thoughts & other entities. These entities include photos, texts, videos, locations, meta data about other users, etc. Thus, securing this content from fake-users is of utmost importance, due to which a wide variety of techniques are proposed by researchers that includes but is not limited to, deep learning models, high density feature processing models, bioinspired models, etc. But these models are either highly complex, or require large user-specific datasets in order to improve their detection capabilities. Moreover, most of these models are inflexible, and cannot be scaled for large social networks with multiple parameter sets. To overcome these issues, this text proposes the design of a novel fusion model to identify fake profiles from multimodal social media datasets. The proposed model initially collects multimodal information about users that includes the presence of profile pic, username length ratios, number of words in the full name, length of their personal description, use of external URLs, account type, number of posts, number of followers & following users, etc. These information sets are pre-processed via a Genetic Algorithm (GA) based feature selection model, which assists in the identification of highly variant feature sets. The selected feature sets are classified via a fusion of Naïve Bayes (NB) Multilayer Perceptron (MLP), Logistic Regression (LR), Support Vector Machine (SVM), and Deep Forest (DF) classifiers. Due to a combination of these classifiers, the proposed model is capable of showcasing an accuracy of 98.5%, precision of 94.3%, recall of 94.9%, and F-measure score of 94.7% across multiple datasets. Due to such a high performance, the proposed model is capable of deployment for a wide variety of social media platforms to detect fake profiles.

Pandemics bring physical life to a complete standstill; people are bound to remain confined to their homes. Students suffer a lot academically due to closure of educational institutes worldwide due to pandemic fear. In such a scenario, imparting adequate education to them so that their academics is not affected, is a big challenge. During the COVID-19 pandemic time, educational institutions have really played a good role in imparting online education to students. Their career and academic tenure were not affected. It is contrary to the past pandemics throughout the world history where students’ academic years were lost. All this has been possible because of advancement in technologies related to Human Computer Interaction. The educational institutions tried to cope up a lot with the current educational mode but lacked in some or the other international ranking parameters. This brought sudden dips in their international ranks which can be regained only in long periods of time with major extra efforts. This research work provides an insight on the slipped off international ranks of higher educational institutions during global disruptive conditions like pandemics (COVID-19 and combatting with future pandemics). The novel model proposed in this work helps academicians in predicting the impact of pandemics on their overall international rankings so that recovery decisions and plans can be taken timely by academicians to combat with the situation. The work involves developing a model based on Machine Learning advanced algorithms with the inclusion of a humongous ranking dataset. Strong empirical results support the high efficiency as sensitivity = 97.98, Accuracy = 97.54, F1 value = 97.82, Kappa-score = 0.95. Using the proposed model. To the best of our knowledge, till now none of the researchers have proposed any such pioneering tool for academicians using advanced Machine Learning algorithms.

Banks are foregoing their present reserves for future sources of Revenue. This source is associated with a risk called credit default risk which increases defaulting conditions called the Non-performing assets(loans) thus leading to the financial crisis. Machine Learning, a branch of Artificial Intelligence, is the upcoming technology with promising solutions to present limitations of the systems eliminating the human errors or emotions with precision by way of training and testing. The present study is focused on predicting defaulting loans using algorithms of Machine learning. The dataset is preprocessed for dropping the missing values. Further three models - Logistic Regression, KNN and XGBoost are applied for predicting defaulters based on precision, recall and F1-score. The findings of the research concluded that the XGBoost model performed best among the three models for assessment of credit risk which will waive off the crisis situation.

The HR department handles all the data regarding the recruitment process while also analyzing them to select suitable candidates for the organization. The HR department interacts with the data regarding the recruitment using many tools for data analysis and interpretation. Such data should be organized properly to ensure the screening of candidates. Each system gives their own report in their own format. Data is not integrated and aggregated at proper granularity suitable for analysis. This can be mitigated by the use of Dimensional and Analytical DataMarts. The dimensional DataMarts will contain transitional data modified for analysis with dimensions and facts. These DataMarts can be used for any adhoc analysis like drill down/roll up, slice and dice, drill through, comparative analysis etc. The analytical DataMart will contain the aggregated data in one row per employee format. This will be task specific like Quality of Hire modeling, Cost of Hire, Time to Hire, demand prediction etc.This paper has focused on creating a recruitment DataMart for Human Resource Department with the assistance of the data modeling technique. This Paper has also discussed the exploratory data analysis and model building in relation to the DataMarts for the recruitment.

Different topic modeling techniques have been applied over the years to categorize and make sense of large volumes of unstructured textual data. Our observation shows that there is not one single technique that works well for all domains or for a general use case. We hypothesize that the performance of these algorithms depends on the variation and heterogeneity of topics mentioned in free text and aim to investigate this effect in our study. Our proposed methodology comprises of i) the calculation of a homogeneity score to measure the variation in the data, ii) selection of the algorithm with the best performance for the calculated homogeneity score. For each homogeneity score, the performances of popular topic modeling algorithms, namely NMF, LDA, LSA, and BERTopic, were compared using an accuracy and Cohen’s kappa score. Our results indicate that for highly homogeneous data, BERTopic outperformed the other algorithms (Cohen’s kappa of 0.42 vs. 0.06 for LSA). For medium and low homogeneous data, NMF was superior to the other algorithms (medium homogeneity returns a Cohen’s kappa of 0.3 for NMF vs. 0.15 for LDA, 0.1 for BERTopic, 0.04 for LSA).

Attrition is a burning problem for any industry and if the rate of attrition is very high, it creates enormous pressure on the process to function effectively. This is precisely what a leading organization’s transportation Line of Business (LOB) is going through where attrition is averaging around 34% for the last three years. Time and again, it has struggled with managing a healthy attrition rate. As a result, there has been a constant occurrence of missed Service Level Agreements (SLAs) resulting in huge penalties. For managers, managing workload has become extremely tedious in the current context.To tackle this problem, this study aims to forecast attrition using time series analysis at various levels based on only the attrition data available for the last fourteen months. The hypothesis here is, if probable attrition is forecasted well in advance, a plan can be put in place to hire employees and make them available whenever there is demand from contract managers. This in turn can help individual contract managers manage their workload efficiently, and reduce their missed SLAs, thereby reducing penalties.The proposed solution is to compare various Time Series Forecasting techniques like Auto-Regressive Integrated Moving Average (ARIMA), Seasonal Auto-Regressive Integrated Moving Average (SARIMA), Exponential Smoothing (ES), Holt-Winters (HW), Moving Average, Ratio to Moving Average, based on attrition data and compared to arrive at the best possible solution.The novelty of this study is the use of time series forecasting techniques to forecast future attrition trends specifically based on attrition data, which has not been explored much. This forecasted data can be used to better workload management which in turn is expected to reduce missed SLAs and penalties.

The study shown in this paper helps the human resource domain eliminate the time-consuming recruitment process task. Screening resume is the most critical and challenging task for human resource personnel. Natural Language Processing (NLP) techniques are the computer’s ability to understand spoken/written language. Now a day’s, online recruitment platform is more vigorous along with consultancies. A single job opening will get hundreds of applications. To discover the finest candidate for the position, Human Resource (HR) employees devote extra time to the candidate selection process. Most of the time, shortlisting the best fit for the job is time-consuming and finding an apt person is hectic. The proposed study helps to shortlist the candidates with a better match for the job based on the skills provided in the resume. As it is an automated process, the candidate’s personalized favor and soft skills are not affected by the hiring process. The Sentence-BERT (SBERT) network is a Siamese and triplet network-based variant of the Bidirectional Encoder Representations from Transformers (BERT) architecture, which may generate semantically significant sentence embeddings. An end-to-end tool for the HR domain, which takes hundreds of resumes along with required skills for the job as input and provides the better-ranked candidate fit for the job as output. The SBERT is compared with BERT and proved that it is superior to BERT.

The peer review process involved in evaluating academic papers submitted to journals and conferences is very perplexing as at times the scores given by the reviewer may be poor in contrast with the textual comments which are in a positive light. In such a case, it becomes difficult for the judging chair to come to a concrete decision regarding the accept or reject decision of the papers. In our paper, we aim to extract the sentiment from the reviewers’ opinions and use it along with the numerical scores to correlate that in order to predict the orientation of the review, i.e., the degree of acceptance. Our proposed methods include Machine learning models like Naive Bayes, Deep learning models involving LSTM and a Hybrid model with BiLSTM, LSTM, CNN, and finally Graph based model GCN. The dataset is taken from the UCI repository consisting of peer reviews in Spanish along with other parameters used for judging a paper. Bernoulli’s Naive Bayes was the model that fared the highest amongst all the approaches, with an accuracy of 75.61% after varying the parameters to enhance the accuracy.

The conventional practices of farming vegetables are inherently associated with the shortage of vegetables due to its seasonal nature, limited farming land and continuous demand. The introduction of artificial means of vegetable production is labor-intensive because of the processes involved in this paper an Artificial Intelligence (AI) based Nutrient Film Technique (NFT) hydroponics system will be designed to mitigate the shortage of vegetable production and minimizes labor. AI will be coded into the ATMEGA328P microcontroller using C++ to provide the required automatic control necessary for the NFT pumps and valves, while Attention Commands (AT) will be used for the interfaced Sim900 Global System for Mobile Communication (GSM) modem for sending Short Message Service (SMS) to the farm operator in case of any malfunction. The improved NFT is found to maximize farmland sine layers of plants can be arranged on one another at a given distance apart, which varies from plant to plant. The AI minimizes human intervention and provides exact mixing and supply of balanced nutrients to plants for fast growth and healthy plants. Continuous production throughout the year using precise AI techniques was designed. Therefore, an improved user-friendly NFT was achieved by utilizing AI embedded in a microcontroller.

Over the last years, ontology learning processes have gained a vast space for discussion and work, providing essential tools for discovering knowledge, especially from textual information sources. One of the most currently used techniques for extracting ontological elements from textual data is through the application of lexical-syntactic patterns, which aim to explore formalities of the language in which texts are written, for removing hyperonym/hyponym pairs that can be used to identify and characterize ontology concepts and create valuable semantic networks of terms. We applied a lexical-syntactic patterns approach in a set of medicine texts, written in classical Portuguese, during the 16th and 17th centuries, with the goal of extracting hyperonym/hyponym pairs to establish a medicine ontology of the time. In this paper, we discuss the most relevant aspects of an ontology learning system we implemented for extracting the referred ontology, which has the ability for characterizing the knowledge expressed in ancient medicament texts.

The imbalanced class distribution of credit-scoring datasets typically makes the learning algorithms ineffective. In this study, NOSTE is proposed, a novel oversampling technique. It first identifies the informative minority instances by eliminating the noisy samples from the minority subset. Then, weight is assigned to the informative minority instances by considering the density and distance factors. Finally, new minority instances are created by determining the average of two different minority instances to make the dataset balanced. In the experimental study, NOSTE performance is validated by conducting an extensive comparison with four popular oversampling methods using three credit-scoring datasets from the UCI repository. The results confirmed that the proposed method brings significant improvement in the classification in terms of F-measure and AUC (Area under the Curve).

Medical tourism is considered as a potential filed for high growth. There are several problems, including the possible ability of patient tourists to monitor critical elements of healthcare quality and professionals. In addition, patient-doctor confidence, process, openness of risks‚ privacy of the medical record and other threats associated with health in specific treatments come from queries. We study the possible benefits of Blockchain technology in this conceptual paper to answer several outstanding medical tourism concerns. We conclude that technology from Blockchain can serve as the basis for future study. Medical tourism. The four primary objectives that may be reached using Blockchain technology, namely to improve tourism experience, to reward sustainable behavior and to provide benefits for local communities, and to reduce worries about privacy, center on Smart Tourism Destinations. The report also discusses the main hurdles to deploy this new technology successfully. This study seeks to better improve existing understanding regarding blockchain technology’s prospective ramifications within the smart tourism field, particular destinations for smart tourism. In this paper, we have provided the strong need of medical tourism and the significance of Blockchain in secured and reliable medical tourism. Medical record sharing on global platform is security concern for doctor and patients. Blockchain Technology can be silver bullet for enhancing medical tourism for Global medical treatment.

The study examined the relationship between oil revenues and government debt. Most oil-producing countries suffer from a problem in the sustainability of public debt, and the inability of these countries to manage financial resources efficiently, especially oil revenues in order to eliminate the accumulation of government debt.The study used two indicators: the ratio of oil revenues to GDP, and the ratio of government debt to GDP, and through the use of ARDL model for co-integration, they were applied to five oil countries, and that the study period is (2004–2018). The study concluded that the relationship between oil revenues and government debt was negative, which is identical to the logic of economic theory, but some countries have exceeded the internationally safe ratio of government debt to GDP, which means the inability of the financial policy maker to manage oil revenues well.The study recommended the importance of the oil resource in order to strengthen the structure of GDP, and encounter the obstacles and problems in the macroeconomic, and that the oil revenues are able to pay off the government debt and improve the national economy.

The first step in preventing losses in agricultural product output and quantity is to identify plant diseases. A significant loss in crop output and market economic value results due to incorrect identification. The farmers used their own eyesight or prior knowledge of plant illnesses to identify plant ailments. When farmers are doing this for a single plant, it is possible, but when it involves many distinct plants, it is much more challenging to detect and takes a lot of effort. Therefore, it is preferable to utilize image processing to detect plants diseases. Image acquisition, picture pre-processing, image segmentation, feature extraction, and classification are all processes in this approach to diagnose the plant disease. In this study, we would like to present the procedures for identifying plant diseases from their leaf photos. We have used VGG 19 model for efficient processing of trained data and test data. This paper aims to support and help the green house farmers in an efficient way.

COVID-19 has been announced as a new pandemic which has affected almost all the countries of the world. Millions of people have become sick and thousands have died due to the respiratory illness caused by the virus. The virus is known to spread when small droplets from nose or mouth of an infected person gets dissolved in air when he or she coughs or exhales or when a person touches a surface infected with virus. The governments all over the world are working on ways to curb the spread of this virus. Multidisciplinary researchers are working to find the best solutions in their own way. Out of the many solutions wearing surgical facemasks is being one of the best preventive measures to limit the spread of corona virus. These masks support filtration of air and adequate breathability. But the problem is that few people don’t use the masks regularly or occasionally due to various reasons like negligence and discomfort etc. This is one of the main causes of high spread of COVID. So, there is a strong need to detect people without mask at public places and to aware them. There are so many initiatives taken by government in this direction, but all have their limitation in one or the other way. So, there is a strong need of a digital solution to ensure that people comply with the government rules of wearing masks in public place sand to recognize unmasked faces on existing monitoring systems to maintain safety and security. Facial recognition systems were being used to identify faces using technology that includes hardware like video cameras. These systems work by combining AI based pattern recognition system along with biometrics to map facial features from an image and compare it with a database of known faces. This research content is also an initiative in this direction to optimize the results.

The continuous rise in for demand services in large-scale distributed systems led to the development of cloud Computing (CC). Because it provides a combination of various software resources, CC is considered dynamically scalable. However, due to the cloud’s dynamic environment, a variety of unanticipated problems and faults occur that hinder CC performance. Fault tolerance refers to a platform’s capacity to respond smoothly to unanticipated hardware or programming failure. Failure must be analyzed and dealt with efficiently in cloud computing in order to accomplish high accuracy and reliability. Over the years, a significant number of techniques and approaches have been proposed for detecting the faults in CC as well as increasing their tolerance ability. In this review paper, we first provided a brief overview of Cloud computing systems, their architecture, and their working mechanism. Moreover, the services provided by Cloud computing and the issues faced by it are also highlighted in this paper. Also, the taxonomy of various faults that occur in the CC environment along with their mitigation techniques is discussed. Furthermore, it has been analyzed that traditional fault detection methods were not generating effective results which resulted in poor performance in cloud environments. Therefore, an ample number of authors stated to use Machine Learning (ML) based models for fault detection in CC. Nonetheless, ML algorithms were not able to handle a large volume of data therefore the concept of Deep Learning was introduced in fault detection approaches. Moreover, it has been also observed that the performance of DL methods can be enhanced significantly by using optimization algorithms along with them. Some of the recently proposed fault detection and tolerant systems based on ML, DL and optimization have been reviewed in this paper.

With the emergence of low-cost video recording devices, the internet is flooded with videos. However, most videos are uncategorized, necessitating video content analysis. This review effort addresses visual big data feature extraction, video segmentation, classification, and abstract video challenges. Exploring compressive sensing, deep learning (DL), and kernel methods for various tasks in video content analysis include video classification, clustering, dimension reduction, event detection, and activity recognition. DL is used to examine video footage recognition and classification. This study examines the algorithms’ flaws and benefits when applied to datasets. The classification approaches used Naive Bayes, support vector machine (SVM), and Deep Convolution Neural Network (DCNN) with Deer Hunting Optimization (DHO). Other approaches have higher false discovery and alarm rates than the DCNNDHO algorithm.

Cochlear disorder is an audio impairment issue, which causes difficulty in understanding human speech. These disorders can cause difficulty in speech recognition, communication, and language development. Intelligent approaches are proven to be efficient and novel approaches for performing various challenging tasks in the healthcare industry. The primary objective of this study is to use machine learning and computer vision domain, to create a web-based platform enabling early detection of the disorders. Computer vision with a classification model is used for achieving the objective. The model is trained on the static custom audiology dataset formulated from the UCI machine learning repository. Cross-validation over various classification algorithms like Logistic Regression, Decision Tree, Support Vector Classifier, K-Nearest Neighbors, and Multi-Layer Perceptron is performed and is proven that Multi-Layer Perceptron suits the dataset. Application for the purpose is developed using Python flask and is deployed for validation.

Quantum computing is an emerging technology and has yet to be exploited by industries to implement practical applications. Research has already laid the foundation for figuring out the benefits of quantum computing for these applications. In this paper, we provide a short overview of the state-of-the-art in data management issues that can be solved by quantum computers and especially by quantum machine learning approaches. Furthermore, we discuss what data management can do to support quantum computing and quantum machine learning.

This research work aims to analyze the nutritional value of different cereals available in the market through various machine learning models. This analysis is supplemented with the visualization of data also for enhanced understanding. This understanding enables users to devise market strategies as they are competent to evaluate quality of each product and thus its reception in the market. The works starts with statistical analysis through of the data through various plots which provides insight of the data. Further authors perform a comparative analysis of different cereals based on various parameters. This analysis helps to determine the best cereal according to our requirements. The authors have implemented machine learning models on the data to predict the vitamins of any cereal based on their nutritional value. The implementation of various models viz. Linear regression, decision tree, logistic regression, random forest, and KNN advocates the efficacy of various machine learning models to the given problem.

Competitive programming improves our problem-solving abil ity. It helps in writing the source code of computer programs that help in solving given problems, and the majority of the problems are mathemat ical or logical in nature. In view of its staggering and different nature, programming requires a particular level of expertise in the examination of estimations, data structures, science, formal reasoning, and related tasks like testing and investigating. In light of the growing regard for expectations for programming, there exist different genuine programming platforms like HackerRank, CodeChef, CodeForces, Spoj, etc. where students can practice and work on their competitive programming skills. Monitoring the progress on these different platforms becomes hectic as they have to manually check each one. Also, there is no tool that helps in predicting their future scores based on their current practice. Another issue is that if the organisations or institutions wanted to monitor their student’s progress, it would be tougher as it would have to be done for each student manually. This work will help the students, as well as the organisations or institutions, maintain a proper portal with data to monitor their progress, by which students can improve their competitive programming skills, saving a lot of time compared to the time taken to do this monitoring manually.

Employee attrition is the reduction in the employee workforce, which can be defined as the rate of employees leaving the company faster than the rate they are hired. Attrition may be for the whole establishment but sometimes it might be particular for a business field. This happens when there is intervention of technology that contribute in replacing the human workforce. There are several factors contributing to employee attrition, a few being age, number of years in the company, manager, technology change, etc. It is vital to understand the impact of these factors on employee attrition so that necessary action can be taken to avoid this. Thus, Machine learning technique is being used nowadays to inspect and predict the data of several real-life applications. After employing the models, authors performed the analysis on each of them using confusion matrix, F-1 score, recall, precision, etc., and found that the best model is SVM with an accuracy of 85.60%.

As businesses advance toward digitalization by automating an increasing number of procedures, unstructured forms of text in documents present new challenges. Most organizational data is unstructured, and this phenomenon is on the rise. Businesses like healthcare and insurance are embracing business process automation and making considerable progress along the entire value chain. Artificial intelligence (AI) algorithms that help in decision-making, connect information, interpret data, and apply the insights gained to rethink how to make better judgments are necessary for business process automation.A healthcare procedure called Prior Authorization (PA) could be made better with the help of AI. PA is an essential administrative process that is a component of their utilization management systems, and as a condition of coverage, insurers require providers to obtain preapproval for the provision of a service or prescription. The processing of insurance claim documents can be facilitated using Natural Language Processing (NLP). This paper describes the migration of manual procedures to AI-based solutions in order to accelerate the process. The use of text similarity in systems for information retrieval, question-answering, and other purposes has attracted significant research. This paper suggests using a universal sentence encoder, a more focused strategy, to handle health insurance claims. By extracting text features, including semantic analysis with sentence embedding, the context of the document may be determined. The outcome would have a variety of possible advantages for members, providers, and insurers. AI models for the PA process are seen as promising due to their accuracy and speed of execution.

Sign language is a way of communication in which hand gestures and symbols are used to connect with each other. Communication provides interaction among people to exchange feelings and ideas. Similarly, when it comes to the handling of medical equipment using a robot, sign language should not be a barrier to carrying out such applications. The purpose of this work is to provide a real-time system that can convert Sign Language (ISL) to text format. Most of the work is based on the handcrafted feature. This paper concentrates on introducing a deep learning approach that can classify the signs using the convolutional neural network. First, we make a classifier model using the signs, then using Kera’s implementation of convolutional neural network using python we analyze those signs and identify the surgical tools. Then we process another real-time system that uses skin segmentation to find the Region of Interest in the frame. The segmented region is fed to the classifier model to predict the sign. The predicted sign would gradually identify the surgical tool and convert the sign into text.

In the last few years physiotherapists has been treating the individuals by taking their decisions quickly which results in complexity, non effective and decrease in recovery rate in rehabilitation. Walking aids play a vital role in helping the patients in certain conditions, they lack balance and leads to complex usage of device. These type of aids helps for individuals suffering from spondylitis, disc bulging and postural In this proposed system we design a solution for patient using walkers, this device helps them to maintain the proper balance. As a proof of concept, we develop a sensors-based solution which can be easily embed where the walker helps to get real time information of the subject and helps in recovery rate. This device will automatically warn the users when they apply more pressure on a sensor using haptic feedback. To design a smart crutch tool for Elders and Accident patients which balances the force applied in terms of stress and strain to warn the patients by means of haptic feedback.

Deadlock is a major problem for systems that allocate resources (AL). There are many solutions to the deadlock problem in distributed systems, the solutions are divided into the following three groups: deadlock-prevention, deadlock-avoidance, and deadlock-detection. AL and related deadlock prevention originate from the design and implementation of operating systems and distributed computing. In this article, we systematize research related to distributed systems, problems of AL, strategies in AL, and solutions to deal with deadlock situations in AL. We present deadlock avoidance algorithms, and deadlock prevention, in addition, we present a deadlock detection algorithm using a two-way search with running time complexity of the horizontal arc O(m1/2) when the edge (v,w) is added to the graph. Compare the two-way search algorithm with the improved algorithm, and finally the experimental results.

Researchers are working to fill knowledge gaps as new eras are ushering in by the rapid growth of informatics and human-computer interaction. With speech-based communication, touch-free engagement with electronic gadgets is growing in popularity and offers consumers easy-to-use control mechanisms in areas other than the entertainment sector, including the inside of cars, these engagement modes are now being successfully used. In this study, real-time human gesture identification using computer vision is proven, and the possibility of hand gesture interaction in the automobile environment is investigated. With the use of this noncognitive computer user interface, actions can be carried out depending on movements that are detected. By adding Python modules to the system, the design is carried out on a Windows OS. The platforms used for identification are open-cv and keras. The vision-based algorithms recognize the gesture displayed on the screen. A recognition algorithm was trained in keras using the background removal technique and the LeNet architecture. In this paper, four models were created and their accuracy was compared. The convex hull and threshold model outperformed the other models.

For computer and human interaction, human facial recognition is crucial. Our goal is to anticipate the expression of a human face, gender, and age as quickly and accurately as possible in real-time. Understanding human behavior, detecting mental diseases, and creating synthetic human expressions are only a few of the applications of automatic human facial recognition . Salespeople can employ age, gender, and emotional state prediction to help them better understand their consumers. Convolutional Neural Network one of the Deep Learning techniques is utilized to design the model and predict emotion, age, and gender, using the Haar-Cascade frontal face algorithm to detect the face. This model can predict from video in real-time. The goal is to create a web application that uses a camera to capture a live human face and classify it into one of seven expressions, two ages, and eight age groups. The process of detecting face, pre-processing, feature extraction, and the prediction of expression, gender, and age is carried out in steps.

Education is one of the major sources for determining the growth of country with high economic development. But the challenges facing by the education systems are poor decision-making ability, high difficulties in adapting new curriculums, inefficient teaching, and training. These factors could inherently affect the performance of education sectors in terms of increased unemployment, reduced workforce, and dissatisfaction outcomes. In order to solve these problems, this research work aims to deploy the data science technologies for improving the learning strategies in education systems. Here, the data mining techniques are mainly used to extract the relevant or useful information from the data and is widely used for solving the higher education problems. Also, this work investigates some of the challenges associated to the deployment of big data in education systems, which includes consequentialism, scientism, privacy, and security. Moreover, operating characteristics and features of the cyber security model are assessed and validated in Sect. 5. Finally, the overall paper is summarized with its obtainment and future work.

The amount of data produced every day is enormous. According to Forbes, 2.5 quintillion data is created daily (Marr, 2018). The volume of unstructured data is also multiplying daily, forcing organizations to spend significant time, effort, and money to manage and govern the data assets. This volume of unstructured data also leads to data privacy challenges in handling, auditing, and regulatory encounters thrown by governing bodies like Governments, Auditors, Data Protection/Legislative/Federal laws, regulatory acts like The General Data Protection Regulation (GDPR), The Basel Committee on Banking Supervision (BCBS), Health Insurance Portability and Accountability Act (HIPPA), The California Consumer Privacy Act (CCPA) etc.Organizations must set up a robust data protection framework and governance to identify, classify, protect and monitor the sensitive data residing in the unstructured data sources. Data discovery and classification of the data assets is scanning the organization’s data sources both structured and unstructured, that could potentially contain sensitive or regulated data.Most organizations are using various data discovery and classification tools in scanning the structured and unstructured sources. The organizations cannot accomplish the overall privacy and protection needs due to the gaps observed in scanning and discovering sensitive data elements from unstructured sources. Hence, they are adapting to manual methodologies to fill these gaps.The main objective of this study is to build a solution which systematically scans an unstructured data source and detects the sensitive data elements, auto classify as per the data classification categories, and visualizes the results on a dashboard. This solution uses Machine Learning (ML) and Natural Language Processing (NLP) techniques to detect the sensitive data elements contained in the unstructured data sources. It can be used as a first step before performing data encryption, tokenization, anonymization, and masking as part of the overall data protection journey.

In recent months, all kinds of image-generating models got the spot-light, opening many possibilities for further research direction, and from the commercial side, many teams will be able to start experimenting and building products on top of them. A sub-area of image generation that picked the most interest in the eye of the public is text-to-image models, most notably Stable Diffusion and MidJourney. Open sourcing Stable Diffusion and free tier of MidJourney allowed many product teams to start building on top of them with little to no resources. However, applying any pre-trained model without proper testing and experimentation creates unknown risks for companies and teams using them. In this paper, we are demonstrating what might happen if such models are used without additional filtering and testing through bias detection.

This article is about machine Learning (ML) depending spectrum sensing in using cuckoo search optimization method. In Present days as the number of mobile users is increasing, scarcity of spectrum is arising due to allocation of the available spectrum to growing number of the users in cognitive radio. So there is a need to efficiently utilize the limited spectrum that is available for use. Spectrum sensing is one of the prominent method for effective utilization of the spectrum. Among the existing methods of spectrum sensing using Energy detection, Machine learning based sensing is more prominent. For efficiently optimizing the spectrum sensing cuckoo search based optimization has been used in this paper. For analyzing the channels under noise conditions Gaussian function has been considered. Average information per message based classifier is a good technique of detection for spectrum sensing. Classification has been done with the help of support vector machine and K-Nearest Neighbor algorithms. From the obtained results it has been shown that average information based SVM, KNN techniques outperforms the conventional energy detection based techniques and cuckoo search based optimization has yielded better sensing accuracy with minimum loss.