Advancements in Artificial Intelligence and Machine Learning

The incorporation of artificial intelligence (AI) into healthcare systems has demonstrated significant potential to transform patient care, diagnosis, and treatment. Nevertheless, the implementation of artificial intelligence (AI) in the healthcare sector presents difficulties concerning transparency, interpretability, and trust, especially when there are new possibilities for automated decision-making and enhanced efficiency in many different areas, thanks to the combination of artificial intelligence and mechatronics. Automation and robotics are improving as mechatronics integrates AI. Grand View Research expects the global mechatronics and robotics course market to reach $3.21 billion by 2028, expanding 13.7% from 2021 to 2028. This chapter aims to give a general outline of mechatronics-related artificial intelligence (AI), including its applications, advantages, and challenges. The field focuses on developing intelligent machines with the ability to learn, understand data, and react accordingly. Machine learning and deep learning are two forms of artificial intelligence that have enabled robots and autonomous vehicles to detect their environment, traverse complicated scenarios, and make smart decisions using the data they collect. Artificial intelligence (AI) improves mechatronic systems by expanding their capabilities, which boosts their performance, output, and reliability. Nevertheless, ethical considerations and implementation challenges need to be resolved before the full potential of AI in mechatronics can be realized.

Rapid advances in machine learning, deep learning, and AI are changing business and society. The history of these technologies reveals their transformative impact on a variety of industries. Convolutional, Deep, and Recurrent Neural Network studies show their evolution. Deep learning has facilitated the development of novel applications in several industries and stimulated technological progress. Reinforcement learning, a core component of artificial intelligence, has made notable progress, particularly in the realm of autonomous systems. This paper discusses the latest algorithms and their applications, stressing reinforcement learning's impact on robotics and automated decision-making. The study of natural language processing is crucial. Through language modelling, sentiment analysis, and translation, it shows banking, healthcare, and customer service applications. After that, the paper examines real-world AI applications. These technologies help doctors detect, treat, and forecast medical disorders. Fraud detection, risk management, and algorithmic trading benefit financial institutions. Industry 4.0 combines AI-driven autonomous vehicle navigation, control, and ethical decision-making with intelligent manufacturing and predictive maintenance.

The power transformer stands as a critical and costly component within electrical networks. Transformer oil serves the dual purpose of cooling and insulation within these systems. Its insulating efficacy and overall health are reflected in its physical, electrical, and chemical attributes, which inevitably deteriorate over the transformer's operational lifespan. The properties of transformer oil, such as volume resistivity, viscosity, breakdown voltage, and dissipation factor, undergo alterations over time. Consequently, regular monitoring of the oil condition is necessary to judge the aging effect. Six properties, including moisture content, resistivity, tan delta, interfacial tension, and flash point, have been examined. Data pertaining to these properties across varying ages (in days) have been collected from ten operational power transformers ranging from 16 to 20 MVA. In this paper, Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) based models are presented for predicting the age of transformer oil samples in service. The ANN model employs a multi-layer feedforward network with the backpropagation algorithm, while the ANFIS model is based on the Sugeno model. A comparative examination of the two models indicates that the ANFIS model demonstrates superior performance over the ANN model, producing better outcomes.

The rapid development of new digital and online systems that are powered by artificial intelligence (AI), such as social media, targeted marketing, and personalized search engines, has resulted in the introduction of new ways to engage with individuals, collect information about them, and possibly influence the activities they take. Concerns have been expressed, however, regarding the possibility of manipulation brought about by such types of breakthroughs, they provide one-of-a-kind capabilities for targeting and exerting a significant amount of influence over individuals through the persistent application of automated tactics. When it comes to voting, electorates are likely to be influenced to vote for a particular entity. It is important to note that the strategies that involve targeting and profiling on social media platforms serve not just advertising reasons but also propaganda purposes. Through the monitoring of individuals' activity on the internet, social media algorithms can develop user profiles. These profiles are then leveraged to provide recommendations that are specifically customized to a certain audience segment. As a result, propaganda and incorrect information have a greater potential than ever before to influence the perspectives and decisions of voters, particularly in matters related to elections. This chapter provides a detailed survey of the existing literature on manipulation, with a particular focus on the impact of artificial intelligence and other technologies that are connected to manipulation. Furthermore, it considers the ethical imperatives and societal obligations of artificial intelligence developers, social media platforms, and regulatory agencies in the process of minimizing the adverse effects associated with the technology

In the age of technology, the main function of video surveillance is to detect and track individuals in dynamic environments. This chapter extensively explores and reviews comprehensive literature on various human detection methodologies and datasets, focusing on frameworks that detect human presence through object detection methods by processing video sequences.
The object detection techniques discussed include face detection, motion detection, frame differencing, histogram-based, and geometry-based approaches. These techniques classify objects as human or non-human using different deep learning and machine learning models. This survey explores current technological advancements and their frameworks, revealing insights from studies addressing challenges such as occlusion, pose variation, and environmental complexities.
An overview of prominent human detection datasets, such as INRIA, MIT, CAVIAR, CALTECH, and others, offers valuable resources for training and evaluating detection models. This comprehensive exploration aims to provide researchers and practitioners with a cohesive understanding of human detection methodologies, challenges, and diverse datasets for advancing this critical field in computer vision and surveillance technology.

Recently, sentiment analysis and opinion mining have drawn a lot of interest. Because user-generated content on the internet is becoming more and more influential, this research examines the various machine learning (ML) techniques used in opinion mining and sentiment analysis applications. The opinion mining and sentiment analysis utilizing machine learning approaches are thoroughly reviewed in this research article. The objective of this comprehensive review is to find the most accurate and efficient models that can automatically classify and analyze sentiments expressed in textual data. A range of machine learning techniques are utilized and assessed according to performance measures including precision and accuracy. The dataset used consists of real-world text data collected from social media platforms, product reviews, and online forums. The findings indicate that Support Vector Machine (SVM)and Naïve Bayes (NB) achieved exceptionally high values of accuracy. SVM and NB achieved an accuracy of 95%. On the other hand, Logistic Regression (LR) and K-Nearest Neighbor (KNN) demonstrated comparatively lower accuracy scores of 57% respectively. Among all the evaluated techniques, KNN exhibited the lowest precision score of 57%. Overall, ML techniques have proven to be valuable in sentiment analysis and opinion mining.

Visual analysis of human motion has emerged as a frontier vicinity in computer vision in up-to-date years. In visual sequences, it can identify, track, and recognize individuals. It can also comprehend and characterize their movements. With the quick growth and widespread use of information technology, the application of computer vision technology to the study of image processing, pattern recognition, and the efficient extraction and identification of human motion elements from video images has gained significant attention. In order to enhance the capacity to evaluate the attributes of moving human bodies, this work introduces the extraction function method, examines the properties of the extraction function, and proposes a technique for doing so based on picture recognition. The periodic motion in each motion is segmented using the hierarchical clustering approach to obtain precise segmentation. The tests employed self-built data and several benchmark datasets. Tests demonstrate that the method can extract less feature data and retain low computational complexity while producing high classification and recognition results. Additionally, it is capable of naturally integrating both the static and dynamic aspects of human gait. 

The smart grid (SG) system is a novel concept that introduced bidirectional power and communication infrastructure to traditional power grid systems at the beginning of the 2000s. To make information and communication technologies (ICTs) available in utility grids at every point throughout the generation, transmission, distribution, and consumption (GTDC) of power are necessary. This chapter explains the fundamental elements and cutting-edge technologies that make up SGs, including sensor networks, intelligent management systems, monitoring, and metering systems, communication technologies, regulations, standards, and security needs for this idea. Two-way intelligent communication and real-time measurements are all used in the “SG”, an electrical distribution system. A safe, secure, dependable, robust, effective, and sustainable SG is anticipated. Measuring tools like phasor measurement units (PMUs) could drastically alter how grids are monitored. There are a few obstacles, though, like managing the massive volume of data and deploying an adequate number of PMUs. A fundamental need of the SG is communication in both directions. It is necessary to have a communications network that can handle the data traffic and is secure, devoted, and capable. Renewable energy inclusion will change the grid's characteristics. Enhanced distribution-level management and monitoring are necessary, given the current circumstances.

Hardware security is a critical concern for organizations that use information systems to protect their assets from unauthorized access and malicious attacks. Hardware security assessment involves evaluating the security of hardware components and systems to identify vulnerabilities and areas for improvement. This research paper proposes a framework for hardware security assessment using the Analytic Hierarchy Process (AHP) approach. The proposed framework is applied in a case study to evaluate the security of a wireless sensor network (WSN) system. Based on this calculation with respect to each alternative, the author finds that Hardware encryption obtained the highest final weighted score (0.555), which would be the preferred choice according to the AHP method for improving the security of the hardware of WSN. Based on the obtained weight, authors assign the ranks S1<S3<S2 <S4. The results show that the proposed methodology can effectively identify better security algorithms and prioritize actions to improve the security of the WSN system.

In the rapidly evolving landscape of unmanned aerial vehicles (UAVs) and automation technologies, this research delves into the synergistic potential of integrating artificial intelligence (AI) and machine learning (ML) into drone robotics systems. This paper examines the current state of drone technology, focusing on the challenges faced in optimizing their performance for complex tasks. By leveraging AI and ML algorithms, drones can evolve beyond traditional pre-programmed routes and manual control, unlocking the ability to learn from data, adapt to dynamic environments, and make intelligent decisions in real time. As drones continue to play an increasingly pivotal role across diverse industries, from agriculture and surveillance to logistics and emergency response, the fusion of advanced AI and ML promises transformative advancements in efficiency, adaptability, and autonomy. Internet of Drones (IoD) is evidently a promising and versatile application of Unmanned Aerial Vehicles (UAVs) across various domains. The adaptability of drones to unpredictable circumstances and their diverse range of applications make them valuable tools in numerous scenarios. Drones' outstanding mobility and dexterity allow them to identify areas that are dangerous or difficult for people to access.This is especially valuable in scenarios such as inspecting infrastructure, monitoring environmental conditions, or assessing the aftermath of natural disasters.

The incorporation of Artificial Intelligence (AI) and its sub-domains for women's safety and security is a major requirement in the present technology-driven world. Most of the works in this context focus on leveraging Machine Learning (ML) technologies to predict harassment, violence, and other women-related crimes. However, ML approaches are mostly data-driven. In many poor and developing countries like India, where criminals’ data are not well-documented and publicly unavailable, other alternative ways must be planned to prevent crimes against women. One feasible way is to spread awareness about laws and punishments related to women’s crime using AI assistive technology. This will not only help to spread awareness about laws related to crimes against women but also help in preventing by creating fear of consequences for women-related crimes. In this context, a sub-domain of AI, known as a rule-based expert system, is proposed using a top-down inference method to help the user know about the penalty and legal action linked to the crime committed against women in India as per Indian legislative laws. Initially, a comprehensive set of 77 rules is collected from legal domain experts to design the knowledge base (KB) of the proposed expert system. For rapid prototyping, the proposed system is implemented in ES-Builder, an open-source, web-based expert system shell. To check the efficacy of the system, extensive testing of the system has been carried out by querying the expert system, which shows the desired results. 

Brain tumors are a life-threatening disease, and a lot of people are losing their lives. These brain tumors are abnormal cells that develop in and around the brain. This research explores the cutting edge of medical imaging processing, focusing on enhancing the detection and categorization of brain tumors. EfficientNetB0 is the most advanced deep learning architecture that has been thoroughly compared with other deep learning models in order to improve brain tumor classification accuracy using the Kaggle MRI image dataset with 7023 images. The drawbacks of manual tumor identification techniques are discussed, and precise classification using deep neural networks is proposed, with special attention to the transition from binary to multiclassification. This chapter’s primary focus is on improving and optimizing the EfficientNetB0 model through the addition of trainable layers on top of its basic architecture. Several techniques are used like global average pooling for spatial and dimensionality reduction with reduced parameters, dropout to drop layers, and dense net with softmax for multiclass classification. Concurrently, strategic layer freezing is used to refine the deep learning models for foundation design. The results show that the finetuned EfficientNetB0 model with hyper-parameter optimization guarantees exceptional brain tumor accuracy. EfficientNetB0 has achieved a good accuracy of 99.7% and a precision of 99.5% compared to Resnet50, VGG16, InceptionV3 and Xception. This work presents a unique deep-learning method in accordance with a transfer learning strategy for assessing brain cancer categorization accuracy using the enhanced ResNet50 model. As we advance the state-of-the-art, this chapter offers researchers, medical professionals, and patients a solid foundation for accurate and timely brain tumor diagnoses, thus contributing to the research community.