Deaf and hard-of-hearing persons communicate with each other and their communities by using sign language. As this is a natural method for people to interact with computers, many academics are researching it in order to make it simpler and more natural without the use of additional gadgets. So, the main objective of gesture recognition research is to develop systems that can recognise certain human gestures and use them, for instance, to communicate information. Quick, extremely accurate hand detection and real-time gesture identification must be possible with vision-based hand gesture interfaces. Learning sign movements is the first step in creating words and sentences for computer-assisted sign language interpretation. Both static and dynamic sign actions are available. Both types of gesture recognition are crucial to human culture, even if static gesture recognition is easier than dynamic gesture recognition. This paper outlines the processes needed to recognise sign language. The method of gathering data, as well as its preprocessing, transformation, feature extraction, categorization, and outcomes, are all examined. For this area of study, there are also some suggested directions for further research. The visual-manual modality used by sign languages is a collection of established languages. When a hand motion is inputted, the system instantly displays the appropriate recognized character on the monitor screen. In the following, research projects that led to a system that uses convolutional neural networks to identify handwriting on the basis of depth pictures the camera collects.
Some systems correct faulty production to reduce the cost and increase the performance of steel plates. The first element of quality production is to identify and classify defects. However, when the error classification is not done well, the cost of correcting the error increases. This study proposes a classification and regression tree (CART) based fuzzy model to determine which fault class the faulty steel plates fall into. The second model used in the study is the adaptive neuro-fuzzy interface system (ANFIS). In both approaches, error classes were determined using fuzzy logic. In the two models created, it was observed that when the information of the detected faulty steel plate was entered, it was unsuccessful in determining which class this fault belonged to. Although it suggests a quick solution for detecting the error class, it has been seen that these approaches are not appropriate to use because they do not offer the right solution.
HVAC systems are heating, ventilation, and air conditioning units that allow for maintaining indoor air conditions to a desired degree. Other than maintaining occupant comfort, its use in industrial applications and critical infrastructure is vital to maintain operations since many systems and components must operate under controlled environments, such as, to avoid overheating. HVAC systems by design do not incorporate security. Given this, most HVAC systems , especially those connected to the internet are vulnerable to cyber attacks from various angles. Such an example was the Stuxnet virus which sabotaged a nuclear plant in 2010. Due to this and among other events, HVAC security is now under the spotlight, and the need for a solid security model for HVAC security is necessary. In this paper, we present an anomaly-based detection machine learning model to detect malicious sabotage of HVAC systems which achieves an accuracy of 99% and recall of 98.2%.
It is of paramount importance that software applications stay performant as the number of users increases due to certain predictable events such as sales, promotional campaigns and seasons. Modeling, typically using Queuing Networks, is commonly used to predict the application’s performance characteristics to help in finding the extra required computing resources.Building Queueing Network Models to represent the various hardware and software components is a complex and ad-hoc task due to the lack of generalized models. Additionally, obtaining model parameters, known as Service Demands, is a time consuming inaccurate process. Furthermore, solving such complex models requires the use of slow and resource consuming simulation techniques. Simplifying models and solutions to rectify these issues results of inaccurate predictions leading to wasted computing resources.In this work, we propose the use of Deep Neural Networks to build software application models to predict performance characteristics under varying workloads. Unlike Queuing Networks, Deep Neural Networks are built using well-established generalized approaches. Additionally, their input data can be obtained from the applications performance monitoring data or using typical performance measuring techniques. Finally, solving Deep Neural Networks is way more efficient than the simulation techniques of Queuing Networks.
As the world becomes increasingly digitized, the use of artificial intelligence (AI) has the potential to revolutionize sustainable computing practices. In this paper, we provide a comprehensive review of the impact of AI on sustainable computing, including its applications, case studies, challenges, and limitations. Our results demonstrate that AI can significantly reduce the environmental impact of computing systems and enable more sustainable practices, such as energy-efficient data centers, smart grids, and precision agriculture. However, the adoption of AI in sustainable computing also poses ethical and social challenges that must be addressed, such as bias, privacy, and job displacement. To address these challenges, we propose a governance framework for ethical and socially responsible AI in sustainable computing. We conclude by highlighting the contributions of this paper, including a comprehensive review of the literature and a framework for ethical and socially responsible AI. We also provide future research directions for the field of AI and sustainable computing, such as designing AI-powered smart cities and exploring the role of AI in the circular economy.
Digital media triage is a main challenge that faces a digital investigator. Knowing what might be useful during crime investigation could greatly save the investigator’s time and enhance outcomes. Memory investigation can get potential benefit from the triage process because of its scattered and diverse contents. The Operating System’s paging scheme might map the contents of a file into non-consecutive page frames in the physical memory, making the file-type classification and identification process harder. This paper tackles the memory content triage problem at the page level using a machine-learning methodology.
The fourth industrial revolution and the Covid pandemic have brought significant changes in teaching and learning, necessitating the use of digital technology. Our CrossQuestion application uses gamification and flipped classroom pedagogies to enhance student learning. This study aimed to assess the effectiveness of implementing CrossQuestion in a higher education setting. The research involved a structured literature review, establish research methodology, and to perform data analysis. The results showed that the use of CrossQuestion through gamification and flipped classroom pedagogies is an effective approach to enhancing student learning. Therefore, the study concludes that digital technology can be a useful means to enhance student learning in the current educational landscape.
Skin cancer is one of the most common and dangerous diseases due to a lack of awareness of its signs and methods for prevention. Skin cancer disease can be counted as a fourth burden disease around the world, with the rate of deaths dramatically growing globally. Therefore, early detection at an early stage is necessary to stop the spread of cancer. In this paper, we detect and classify multi-label skin cancer and implement the optimal techniques using machine learning and image processing approaches. However, preprocessing methods assist in removing irrelevant and unnecessary features from the label encoder, and standard features are applied to standardize the range of functionality by scaling the input variance unit. Moreover, various machine learning techniques were applied to check the performance of every classifier on the HAM10000_metadata dataset. The experimental analysis was conducted on the HAM10000_metadata dataset, which consists of seven different types of skin cancer. The results analysis shows that machine learning algorithms such as SVM, DT, and GNB obtained the highest accuracy compared to the other classifiers.
This review explores the current state of Explainable Artificial Intelligence (XAI). This study looks at current advances in XAI research, as well as challenges and the future. To accomplish this, the review will explore XAI literature on a wide range of topics, including current techniques, strategies, and applications. The review begins with an overview of XAI, including its definition, history, and motivations. It will then investigate XAI's current situation as well as explainability approaches. This will entail a look at model-based, post-hoc, and interactive explainability, as well as the methods used to get there, such as feature attribution, rule extraction, and counterfactual analysis. The application of XAI in healthcare, finance, and autonomous systems will be examined. The review will look at the ethical and social consequences of XAI, such as bias, accountability, and transparency. Among the challenges that XAI faces include a lack of uniform measurements and evaluation procedures, interpretable data, and domain-specific expertise. The review will include important trends, patterns, and avenues for improvement. The review will also recommend XAI research and development. This review will assist XAI researchers, practitioners, and policymakers in understanding the field's current state, limitations and challenges, and future direction.
Web technologies are the fundamental platforms for educational organizations, and thus, assessing their usability and accessibility is inevitable. The study examines the usability and accessibility of 134 university homepages in the UAE, Saudi Arabia, and Qatar. The websites are evaluated using automated online testing techniques. The research indicates that the evaluated websites had low levels of WCAG 2.0 compliance. According to the findings, websites in Qatar performed better in terms of homepage usability and accessibility than websites in the UAE and Saudi Arabia. Usability and security were not significant concerns, but there was room for advancement. This study provides web developers with recommendations for improving the accessibility and usability of university websites and homepages.
This research examined the role of sustainability in the Airbnb market in London, using machine learning. The study found that approximately 7.59% of words in property descriptions were sustainability-related, implying hosts acknowledge the value of sustainable practices in hospitality. Higher sustainability ratings correlate with higher prices, highlighting sustainability’s potential as a competitive advantage. Proximity to amenities was another key factor for hosts and guests, as indicated by frequently used keywords like "walk", "access", and "central". The findings underscore the importance of sustainability and location convenience in the evolving hospitality industry, influencing guest attraction and pricing.
Rapid developments in computer vision, big data and deep learning have contributed to automated fault diagnosis in many domains. Intelligent pavement damage prediction is of great significance in transportation maintenance to ensure safety. As the depths of convolution neural networks increases, loss function’s gradient move towards zero. Furthermore, severity of the crack’s inhomogeneity, intricacy of the background—which includes the low contrast with the nearby pavement and the possibility of shadows with a similar intensity results in ambiguous crack prediction. Hence, the crack detection image dataset is preprocessed using homomorphic filtering. To avoid the complications of vanishing gradient, this paper presents three different deep learning architectures (custom model (CR Net), Res-Net50 and Fisura_Net) based on residual and skip connections with hyper parameters tuned optimally to classify and identify crack and no-crack images. Experimental analysis was carried out with a comparison of contemporary CNN models like ResNet-50, ResNet 152 and previous existing research works on the basis of precision, recall, accuracy, F1 score, loss accuracy, and receiver operating characteristic (ROC) curves and Area Under Curve (AUC) metric. The comparative results ascertain the proposed Fisura_Net model utilizing less computational complexity with highest classification prediction efficiency (99.85%), elevated F1 score (1.0) and AUC score (0.99). Fisura_Net outperformed the other existing models and exhibited promising damage prediction.
The era of big data and social networking platforms have provided great repositories of the data for mining useful information for the real-world industry. However, along with this benefit comes the noise in the data. Generally, noise is the data-set that are redundant, false, bad, and/or outliers. Data cleaning, outlier identification, feature engineering, data slicing, etc. are few of many techniques used traditionally. End goal remains ensuring good data (signal) is not lost in bad data (noise) and less processing cost are incurred to extract useful knowledge out of given big data. This paper presents a follow up progress on existing work of the author in relevance of machine learning algorithms, academic and career data predictions and personality computing. All of that have been initially inspired by potential of useful relationships and data points in unstructured data and thus Noise becomes very relevant and may appear Signal in other contexts and predictors in goal. This proposed model is collectively titled as ‘Noise Removal and Structured Data Detection’ based on inherited parallel processing and unique n-Dimensional training approach. Personality features can be quantified into talent traits, matrix indicating the max/min for relevance factors in the academics/career of nD. The engine internals examine and train the algorithm that it minimizes the x,y co-ordinates and maximizes the z co-ordinate. It records and compares the engine internal metrics and reports it back to engine to further optimize the machine learning process until the optimum results are obtained or do not improve any further.
As the World moves towards renewable energy, photovoltaic modules are a fundamental option due to their green nature. However, the manufacturing process of solar cells is complex and vulnerable to discrepancies which can impact the overall performance of the system. Although human-led inspection is seen as the de-facto quality inspection protocol, issues pertaining to bias, cost and time can make it an expensive process. To this effect, this paper focuses on the development of a custom convolutional architecture that is lightweight, hence deployable within manufacturing facilities to assist with defective solar cell inspection. In addition, to address the issue of data scarcity, representative data augmentations are producing tailored towards enhancing the model’s generalizability. The high efficacy of the proposed CNN and proposed augmentations can be gauged by the fact that 98% F1 score was achieved overall.
The United Arab Emirates is facing a significant challenge with drug addiction, trafficking, and promotion within its borders. According to Dajani, 6.1 people per million have died from drug use [1]. To address this issue, the Ministry of Interior is dedicated to ensuring the safety and stability of its citizens and residents through improvements in monitoring and tracking systems, communication networks, and collaborative efforts in the maritime, land, and air domains with its partners. Despite the increasing number of drug-related crimes, the UAE is focusing on implementing advanced technologies to swiftly apprehend suspects and reduce the drug crime index. Drug trafficking and usage in the UAE has been a persistent problem since the 1980s, with a high rate of drug-related deaths recorded by Dubai Police in 2001. Over 3.5 tons of drugs were recovered, and 1,307 individuals were arrested for drug-related offenses such as possession, misuse, and sale. Heroin and Hashish were the most commonly used substances. To minimize drug-related crimes and protect society, this paper proposes a drone-based system.
The recommender systems (RS) are significant in academics, business, and industry. They are frequently employed in various fields, including shopping, music, movies, travel, dining, and writing. Recently, RS can be used in education to suggest student learning styles. This paper proposes a recommender system for predicting student personality with emotions. One of the common recommender system methodologies, collaborative filtering, generates the best suggestions by finding related individuals or things based on their prior transactions. One of the main issues with the collaborative filtering process is the poor accuracy of ideas. This paper uses association rule mining to recommend student personality with emotion based on closed-ended questionnaires. This work initially uses the sentiment analysis technique to identify the student's emotions based on the answer provided for a closed-ended questionnaire. Then, polarity-based sentiment analysis is used to classify student emotions. This paper uses the Association rule mining concept to predict student personality with emotion. This is the first study of a recommender system for the student based on closed-ended questionnaires. The real- world closed-ended questionnaire like Emotional intelligence, Eysenck personality, Self-determination scale, Self-efficacy, Rosenberg's self-esteem, Positive and Negative affect schedule, and Oxford Happiness is used to evaluate the performance of the proposed research work.
Heart rate estimation is the most used indicator in clinical medicine for determining how well the cardiovascular system is functioning. Human heart rate is a key indicator of physiological health and a measure of cardiovascular status. Due to its many advantages like now-cost, non-contact and the ability to produce accurate results in a shorter amount of time as compared to most of the traditional techniques, this method is attracting more scholars for further studies. Our proposed system aims to build a non-contact heart rate estimation and detection system using a simple web-camera. In this system, the subject has to place their face in front of a web-camera and our software will detect the face, find the region of interest (ROI) and determine the user’s heart rate. Once the ROI is established, it calculates the estimated heart rate of highest accuracy using the Eulerian Video Magnification Algorithm. The system being low cost, having very less requirement and highly efficient, it can play a very vital role in the health care and telemedicine industries.
This study analyzes the explanatory power of AI text-based emotion ID metrics on the hourly price change of 12 major crypto-assets. Five GARCH models were tested, with the results providing evidence of a positive relationship between the number of mentions of a crypto-asset name in searches and its price change for 10/12 assets. AI text-based emotion detection metrics also provided marginal evidence of forecasting power for most assets (3/12) and little evidence (3/12) of the predictive power of the name "Bitcoin" as a generic reference term. Practical applications for analysts looking to employ AI text-based emotion ID tech in their models are noted.
With the epidemic of Covid-19, a realisation has come into effect that the healthcare industry requires a more efficient method of storing non-tamperable vaccination certificates as well as sharing them with the relevant parties when required. As such, the need for efficient and secure vaccination tracking has led to the development of decentralized solutions. This paper aims to bring forth a Ethereum platform-based vaccination blockchain certificate system which supports authenticated consensual vaccination certificate sharing through the use of solidity smart contracts. We analyse the system architecture, implementation and evaluation, which will support the system's ability to resolve some of the security issues associated with centralised database systems such as low availability, integrity of the vaccination information as well as confidentiality and privacy of patient medical records. Vaxina system is a real-world implementation of a decentralized vaccination tracking system based on the Ethereum platform. It focuses more on the practical implementation and addresses some of the limitations of the previous proposed systems. It has been implemented and tested on the Ethereum blockchain and is available on GitHub for public use. The paper provides valuable insights for future work in the field of decentralized healthcare solutions.
Gliomas, which can appear in different sizes, and locations, and with scattered boundaries, pose a significant challenge as an aggressive type of tumor. Convolutional Neural Networks (CNNs), one of the most effective deep learning approaches for image analysis problems, have been utilized to develop an automatic deep learning-based 2D-CNN model for brain tumor segmentation in this study. The architecture of the model was designed to be deeper by using small convolution filters (3x3). Additionally, the number of convolutional layers was increased to 16 for the HGG model and 13 for the LGG model to facilitate efficient learning of complex features from large datasets and achieve better results. Fine tuning among the dataset and hyperparameters was employed to obtain the results. The pre-processing for this model includes the generation of a brain pipeline, intensity normalization, bias correction, and data augmentation. The proposed method’s performance was evaluated using the BRATS-2015 dataset and the Dice Similarity Coefficient (DSC). Our method achieved DSC scores of 0.79, 0.77, and 0.78 for complete, core, and enhanced tumor regions, respectively. These results are comparable to other 2D CNN architecture-based methods.
