Recent Advancements in Computational Intelligence: Concepts, Methodologies and Applications (Part 1)

Computational Intelligence (CI) is a subfield of Artificial Intelligence (AI) that represents the application of algorithms designed to address specific problems by automating tasks that traditionally require human intelligence. Utilizing algorithms capable of learning and recognizing patterns within data, AI achieves quicker results. With the rapid advancement of CI, its applications have become essential components across diverse domains, transforming industries and reshaping task execution methodologies. This research conducts a thorough examination of the varied techniques and applications of CI, aiming to provide a nuanced understanding of their current status and future potential. The initial segment of the study delves into the foundational tools pivotal for AI development, forming its backbone. Subsequent sections concentrate on the widespread applications of CI in sectors such as healthcare, finance, education, manufacturing, and autonomous systems. The study delivers a comprehensive overview of CI advancements and applications, offering valuable insights for researchers, practitioners, and policymakers. By comprehending the present landscape and anticipating forthcoming developments, stakeholders can navigate the evolving AI ecosystem with informed decision-making, ensuring the responsible and ethical progression of this transformative technology.

The convergence of edge computing with Artificial Intelligence (AI) constitutes a watershed point in technological growth, with transformational implications for a wide range of sectors. This chapter explores the mutually beneficial interaction between edge computing and AI, as well as the inherent difficulties and broad implications for the direction of computing in the future. With its decentralized processing model, edge computing puts data analysis closer to the point of origin, facilitating improved efficiency and real-time insights. Edge devices enable sophisticated cognitive processes and autonomous decision-making through seamless integration with AI algorithms. However, this integration creates architectural challenges that call for creative solutions to handle privacy and security issues and strike a balance between processing power and other resources. The chapter examines how edge computing-enabled AI can revolutionize manufacturing, transportation, healthcare, and smart cities. It shows how AI can change predictive maintenance, individualized healthcare monitoring, and urban infrastructure optimization. New advances offer avenues for collaborative applications and decentralized AI training. Federated learning models, the spread of edge AI chips and algorithms, and the democratization of AI capabilities are some of these themes. In summary, this chapter provides direction on navigating the rapidly evolving area of artificial intelligence (AI) facilitated by edge computing, encouraging collaboration and innovation to fully achieve AI's transformative potential and address digital age concerns.

With the rapid development of the social economy and industrialization, the trend in autonomous vehicles (AV) is growing dramatically. This trend will significantly change the car industry, making sensors indispensable. This chapter addresses the challenges and opportunities in securely collecting, storing, and processing AV data. AV technologies must make fast decisions based on diverse data, including moral dilemmas, and optimize processing for systems like Advanced driverassistance systems (ADAS), Lane Keeping Assist (LKA), and Traffic Jam Assist (TJA). Data security is crucial to avoid threats associated with AV and Edge Cloud technology. The proposed solution involves Edge Cloud Assisted Data Management using secure deep learning algorithms. Data from cameras and sensors in AVs is processed in the cloud and with the help of an advanced deep learning algorithm, Faster R-CNN, for accurate object detection and classification. These processed results notify control systems and actuators for safe and efficient vehicle operation. Additional sensors monitor vehicle behavior and can intervene in difficult situations. Edge Cloud technology enhances data processing efficiency and optimization. Deep learning algorithms accurately identify objects, including in blind spots, providing optimal solutions in complex situations. Object detection and instance segmentation with Faster R-CNN offer fast and accurate object location predictions. Data security is ensured through novel encryption techniques. This chapter explores the data handling units in AV technologies and emphasizes the importance of adopting edge cloud computing for the safe and efficient operation of autonomous vehicles. The metrics considered in the proposed object detection algorithm are intersection over union and mean average precision. Combining efficient neural network architectures with techniques like pruning, quantization, and edge computing significantly enhances performance while maintaining safety and reliability.

In the presence of fog, photographs often suffer from poor visibility due to the atmospheric conditions. This reduction in visibility stems from the obscuring effect of fog on the surrounding environment. Additionally, the presence of particles, dust, and pollutants further contributes to color dilution and diminished contrast in analyzed images, an inevitable consequence of these contaminants. As this phenomenon persists, distinctions between individuals may gradually blur in their collective consciousness. Within the context of this investigation, a novel method is proposed for single-image processing, leveraging a hybrid filter. Specifically, it suggests the adoption of GGIF (Globally Guided Image Filtering) in conjunction with smoothening filters such as Weighted Least Squares (WLS) and Two-Dimensional Bilateral Filters. The experimental results demonstrate enhanced haze removal and denoising across all images, showcasing the method's efficacy with an overall perceptual fog density (PFD) reduction improvement of 55.39% and the flexibility to adjust parameters, including the sigma variable in GGIF, for improved performance

The Industrial Internet of Things (IIoT) has emerged as a prominent area of research, leveraging sensors, actuators, and computing capabilities to revolutionize data collection, communication, and processing. This has led to the development of innovative Industry 4.0 applications across various sectors such as mining, energy, healthcare, agriculture, and transportation. Despite its numerous benefits, security and privacy remain significant challenges in IIoT design, necessitating the integration of intrusion detection systems (IDS) for mitigation. Recent advancements in machine learning (ML) and deep learning (DL) algorithms offer promising avenues for the development of effective IDS techniques tailored to the IIoT environment. However, the scarcity and imbalance of attack datasets pose obstacles to existing IDS models. In response, this paper proposes a deep learning-enabled privacy-preserving technique aimed at improving classification accuracy. The proposed Fused IGAN-IDS model comprises three key components: (i) Generative Adversarial Networks (GAN) for generating high-quality samples to address data imbalance; (ii) An extremely randomized tree-based feature selection algorithm for identifying significant features from the dataset; and (iii) The Sparse Stacked Autoencoder (SSAE) algorithm for selecting relevant features to optimize the attack detection process's timing efficiency. By incorporating adversarial data samples into the original dataset, the fused IGANIDS model reached an accuracy of 99.6%, sensitivity of 99.5%, specificity of 99.7%, precision of 98.4%, and a Jaccard Similarity Score (JSS) of 99.7%. This approach also resulted in a lower false alarm rate, with a 4% False Positive Rate (FPR) and a 3% False Negative Rate (FNR). 

This chapter discusses advanced strategies and evolving technologies in the domain of video processing and analysis using VisualBERT. It explores techniques such as object identification, and understanding, with a specific light on the applications such as video summarization and search. The research emphasizes the integration of visual, audio, and textual information through multimodal fusion and attention mechanisms to enhance video exploration. The role of edge computing in real-time video processing is examined, highlighting its potential applications in onthe-fly summarization and searching capabilities. Furthermore, the paper evaluates the integration of blockchain technology for secure content distribution, particularly in secure video streaming and video-on-demand (VOD) services. Additionally, it delves into hyper-personalization and AI-driven content recommendations, investigating methods to tailor video experiences according to individual user preferences. The study concludes by proposing future research directions for advancing the field of video processing and exploration across these domains.

Chatbots are now widely employed across various applications, especially in systems designed to provide intelligent user support. The chatbots are capable of understanding user requests and delivering appropriate responses promptly and accurately. Traditional systems lack true comprehension and generate results based solely on predefined rules, limiting their use to casual chatting and providing information accessible on the internet. Such systems are inadequate for the specific needs of a college intelligent inquiry system. This work introduces a chatbot system within the educational domain, specifically designed to assist university students with their inquiries. The primary objective is to develop a dedicated architecture, formulate a communication management model, and ensure the delivery of accurate responses to the students. To achieve this, a system is created leveraging Artificial Intelligence (AI) techniques and Natural Language Processing (NLP) to recognize and address any type of query. An experimental campaign is conducted on the implementation of the proposed generative and retrieval-based models to evaluate their feasibility and efficiency. This work explores various chatbot platforms, applications, and potential implementations in college settings, aiming to create a user-friendly and efficient inquiry system. 

In recent years, mental healthcare has emerged as a critical global concern, with escalating rates of stress, anxiety, and depression affecting individuals across diverse demographics. However, conventional approaches to treatment and support often face significant barriers, including stigma, limited accessibility, and resource constraints. In order to tackle these obstacles, we suggest creating a transformative mental health voice generative assistant, harnessing the power of state-of-the-art Transformer models and cutting-edge Artificial Intelligence (AI) technologies. The envisioned solution seeks to revolutionize mental healthcare by offering a readily accessible, personalised, and empathetic support system to individuals in need. This assistant attempts to offer real-time guidance, coping mechanisms, and companionship that are specific to the needs and emotions of each user by combining sophisticated AI algorithms, Natural Language Processing (NLP), and emotional intelligence skills. By leveraging deep learning techniques, the assistant will continuously adapt and evolve its responses based on user interactions, fostering a sense of connection and understanding. 

This chapter delves into the intricate interplay between reinforcement learning (RL) techniques and generative artificial intelligence (AI), offering a thorough exploration of how RL can be leveraged to enhance generative models. Starting with a foundational overview of reinforcement learning, the chapter introduces essential concepts, including agents, environments, actions, states, and rewards, within the framework of Markov Decision Processes (MDPs). The exploration-exploitation dilemma, a pivotal aspect of RL, is discussed alongside prominent algorithms such as Q-learning, SARSA, Deep Q-Networks (DQN), and Policy Gradient methods. The core of the chapter focuses on the integration of RL techniques with generative AI models, highlighting both the advantages and inherent challenges. It examines how RL can be applied to various domains of generative AI, such as sequence generation in natural language processing (NLP) and music composition, as well as image synthesis through the enhancement of Generative Adversarial Networks (GANs) with RL-based training strategies. The discussion extends to the practical difficulties faced in training RLbased generative models, including issues related to stability, sample efficiency, and scalability. To provide a concrete understanding of these theoretical concepts, the chapter includes detailed case studies showcasing real-world applications of RLenhanced generative models. These case studies encompass a range of applications, from RL-driven text generation models used in dialogue systems to RL-guided image synthesis for creative industries. Additionally, the chapter explores the application of RL in personalized recommendation systems and game design, illustrating the diverse potential of RL in generative AI. The chapter also addresses the current limitations and open challenges in this burgeoning field. It critically analyzes issues such as the computational demands of training RL models, the need for large and diverse datasets, and the difficulty in ensuring model robustness and reliability. Furthermore, the chapter contemplates future research directions, identifying emerging trends such as the fusion of RL with other machine learning paradigms and the potential of RL in enhancing model interpretability and fairness. Ethical considerations form a crucial part of the discussion, as the deployment of RL in generative AI raises significant societal and moral questions. The chapter reflects on the ethical implications of RL-driven generative systems, particularly concerning data privacy, algorithmic bias, and the broader impact on employment and creativity. The chapter synthesizes key findings, emphasizing the transformative potential of reinforcement learning in advancing generative artificial intelligence. It highlights the significance of this interdisciplinary approach in pushing the boundaries of what generative models can achieve, ultimately contributing to the fields of AI, machine learning, and beyond. The chapter provides valuable insights and practical knowledge for researchers, practitioners, and students, fostering a deeper understanding of the challenges and opportunities at the intersection of reinforcement learning and generative AI. Through this comprehensive exploration, the chapter aims to inspire future research and innovation, paving the way for new advancements and applications in this dynamic and rapidly evolving area of artificial intelligence.

Reinforcement Learning (RL) has emerged as a transformative paradigm for optimizing power distribution grids, offering adaptive decision-making capabilities crucial for enhancing grid efficiency, reliability, and sustainability. This book chapter explores various facets of RL application in grid optimization, starting with foundational concepts and algorithms. It delves into diverse application areas, state representation, action space design, reward formulation, and performance evaluation metrics essential for effective RL deployment in real-world grid environments. The chapter also addresses implementation challenges and proposes solutions, leveraging advancements in algorithmic techniques, integration of edge computing and IoT, and ethical considerations. Through case studies and practical applications, it demonstrates RL's potential to revolutionize grid management. Finally, the chapter identifies future research directions, including enhanced algorithmic sophistication, socio-technical integration, regulatory frameworks, and collaborative research initiatives, paving the way for smarter, resilient, and sustainable energy systems of tomorrow. 

Consumers are constantly bombarded with options in a high-velocity environment, and big-ticket products, including cars, homes, laptops, etc., require major decision-making. It is crucial to assist consumers in effective decisions in line with their requirements and returns. As things stand today, there is no holistic service that can provide precise and immediate price estimates with the capability to offer similar products in the same price bracket as well. Value Vue is a groundbreaking and transformative solution website, which emerges as the answer to these existing gaps, accompanying a new era of precision and trust so consumers can make informed purchasing decisions. Value Vue boasts an intuitive and user-friendly website design, offering a seamless browsing experience that simplifies price predictions. At its core, Value Vue leverages the powerful synergy between advanced machine learning models and cloud computing, offering an unprecedented level of confidence and convenience to individuals navigating the complex world of significant purchases. Committed to data integrity and transparency, Value Vue is a price prediction model that relies on reliable and reputable sources to ensure accuracy, in addition to the trustworthiness of its information, while its cloud-powered scalability allows it to seamlessly handle extensive data sets, catering to various markets and user needs. Value Vue is an automation of valuation processes, enhanced market transparency, and personalized recommendations in a friendly environment.

In today's digital world, it is increasingly important to generate secure passwords to protect against hacking and identity theft. The GIBP (Grassmann algorithm using Illusion pin with Brightness Password generation) approach creates secure passwords by using the Grassmann algorithm, which combines random characters and patterns to form unique passwords. The GIBP, with a strong password, adds an extra layer of security through multi-factor authentication. One popular multifactor authentication method is facial recognition technology, which involves scanning the user's face to confirm their identity. While combining GIBP with the Grassmann algorithm and facial recognition technology, users can generate highly secure passwords that are unique and difficult to crack. In addition, the use of multifactor authentication provides an extra layer of security, making it even more difficult for hackers to gain access to sensitive information. Overall, the GIBP with secure password generation and multifactor authentication is essential in today's digital world to protect against cyberattacks’ ever-increasing threats. The evaluation parameters include time taken, speed of operation for clustering, speed for face recognition, and time taken for training. The proposed method provides 99% accuracy and takes less time to capture images that are less than 1 ms. The proposed method is compared with three other methods and produces good results.