The new generation of Intrusion Detection Systems (IDS) progressively requires automatic and intellectual network ID strategies for managing security risks made by a rising amount of advanced attackers from the cyber environment. Specifically, there were huge demands for autonomous agent-related IDS solution which need some human intervention as possible whereas can progress and enhance themselves (by making suitable acts for a presented environment) and turns out to be highly powerful to effective threats that were not seen before (for instance, zero-day attacks). Recently, DRL methods were presented that can learns from the environment with uncontrollable massive amount of states for addressing the major drawbacks of prevailing RL methods. This article introduces a Binary Bat Algorithm-based Feature Selection with Deep Reinforcement Learning (BBAFS-DRL) system for IDSs. The major intention of the BBAFS-DRL approach is the recognition and classification of intrusion systems. In the projected BBAFS-DRL method, data pre-processed was primarily executed to change the data as suitable format. Furthermore, the BBAFS system can be utilized for the effectual feature selection. Next, deep Q-network is applied for the effectual identification and classification of intrusion. At last, the root means square propagation (RMSProp) optimizer is utilized for the effectual hyperparameter tuning procedure. The experimental analysis of the BBAFS-DRL algorithm can be tested by utilize of benchmark database and the outcome can be analyzed on many measures. The comparison outcome demonstrated the improvement of the BBAFS-DRL methodology over other existing approaches.

In this research work, we present a novel multimodal chat application that contains sophisticated artificial intelligence systems which can accept input audio, image, or any PDF file aimed at enhancing communication. For example, it is enabled by the use of Whisper AI for speech-to-text capabilities, one of the latest LLaVA model for image query handling, as well as a CDB populated with semantic PDF pages for user interaction within the same application. Thanks to also quantized versions of these models, the app achieves optimal performance even on off-the-shelf consumer hardware, hence making the app functional and available to everyone. This platform, which is free for the end users, is encouraging as it encourages creativity and improvement through the community. The application is equipped with various security features to protect the user's information ensuring safe and effective use of the application. Preliminary experiments show how the system is capable of producing on-point answers with respect to the selected narrative and medium which enhances the capability of the system as a resource for the user. The next steps of the development process will be targeted towards increasing the input modalities, improving the real-time processing speed and the enlargement of the platform to fit more advanced users with complex needs.

Fog computing has the capability to perform tasks in the local distributed environment within the expected time period. The approaches used for managing the faults in a fog computing environment are not competent to resolve the issues of latency and energy consumption in the network. Allocating and executing all the tasks regardless of how they scale up and completing them within the allotted time frame is crucial for achieving a higher throughput and a lower drop rate. In order to achieve higher scalability, timely execution, and dependability, the given paper introduces a reinforcement learning (RL) backup queue strategy (SEEDBACK_RL) that ensures every task is executed irrespective of its size. The proposed strategy is enhanced to manage the scalability and reliability by identifying the best nodes depending on the type of task. With the use of a simulative environment, the algorithm's efficacy is assessed, and the results show a notable improvement in load balancing, reliability, and network efficiency.

Multistage malware poses a significant and evolving threat to Internet of Things (IoT) devices. These threats can range from relatively simple attacks to more sophisticated operations, such as cryptojacking—where attackers exploit system resources for cryptocurrency mining—and Distributed Denial-of-Service (DDoS) attacks, which aim to overwhelm devices and networks, causing service disruptions. One major challenge in addressing these threats lies in the reliance on machine learning and deep learning solutions. These approaches often encounter difficulties due to imbalanced data samples, which can distort detection results, and limited generalization capability, a phenomenon known as model drift. Model drift occurs when a model trained on historical data fails to adapt to new and evolving attack patterns, thereby significantly limiting the effectiveness of intrusion detection systems. To confront these challenges, a novel approach called BoTSIAM-DRL is suggested. This innovative model combines Siamese active learning—a technique that utilizes similar input pairs to enhance understanding—with a reward mechanism that incentivizes accurate detection. This unique combination provides a fresh perspective and a solution not previously explored in existing literature. The design of BoTSIAM-DRL allows it to dynamically learn and adapt to the evolving nature of malware attacks, refining its strategies as new threats emerge. The model's performance has been rigorously evaluated using the MedBIoT and N-BaIoT datasets, which are specifically curated for IoT security challenges. Impressively, BoTSIAM-DRL has achieved a detection accuracy exceeding 99

Cybersecurity is a significant concern for financial institutions due to the fast advancement of technology and the growing use of digital services. Robust safety protocols must be implemented to secure sensitive financial information and shield customers from possible risks arising from the increase in cyber-attacks aimed at digital enterprises. Data privacy is crucial in the banking sector to preserve clients' sensitive information, avert financial crime, assure regulatory compliance, and protect intellectual property. The work has gotten tougher owing to the rise in internet use and digital transactions. In this context, a DDoS assault represents a significant threat that compromises client privacy. Effective and resilient strategies for attack detection and prevention are essential. Deep Learning (DL) is the most efficient method for implementing cyber-attack detection systems. The performance of the models was evaluated based on recall, precision, accuracy, and F1-score measures. Assessed methodologies indicate that deep learning has a validated efficacy in identifying DDoS assaults, with a classification accuracy of 99.77%.

The Internet of Things (IoT) is a rapidly growing technology that has been generating increasing amounts of traffic from multiple devices. However, this growth in traffic has also created vulnerabilities that need to be addressed. To identify attacking traffic while preserving data, it is important to quickly process intrusive data. Federated learning is a popular solution for decentralized training that preserves data, but it can also be susceptible to federated poisoning attacks caused by malicious clients. This work proposes a clustering-based client selection strategy to identify malicious clients based on their run time, followed by a trigger-set-based encryption mechanism that verifies the authenticity of the clients. This approach allows unreliable clients with plain text-based gradients to be ignored by the global model. The methodology was evaluated using the IoT23 dataset, and its efficiency, robustness, false alarms, and ability to handle some of the poisoning attacks that occur due to tuning and pruning were verified. The LeNet and DeepCtrl algorithms were used to determine detection accuracy, and after the implementation of a watermarking strategy, the detection accuracy improved significantly. For the DeepCtrl classifier, the detection accuracy improved from 89.90 to 99.8%, while for the LeNet classifier, it improved from 86.21 to 96.54%. This proposed methodology can be a useful tool for identifying attacking traffic and improving the security of IoT networks.

In recent years, the number of devices being connected to the internet has grown enormously, as has the intrusive behavior in the network.Thus, it is important for intrusion detection systems to report all intrusive behavior.Using deep learning and machine learning algorithms, intrusion detection systems are able to perform well in identifying attacks.However, the concern with these deep learning algorithms is their inability to identify a suitable network based on traffic volume, which requires manual changing of hyperparameters, which consumes a lot of time and effort.So, to address this, this paper offers a solution using the extended compact genetic algorithm for the automatic tuning of the hyperparameters.The novelty in this work comes in the form of modeling the problem of identifying attacks as a multi-objective optimization problem and the usage of linkage learning for solving the optimization problem.The solution is obtained using the feature map-based Convolutional Neural Network that gets encoded into genes, and using the extended compact genetic algorithm the model is optimized for the detection accuracy and latency.The CIC-IDS-2017 and 2018 datasets are used to verify the hypothesis, and the most recent analysis yielded a substantial F1 score of 99.23%.Response time, CPU, and memory consumption evaluations are done to demonstrate the suitability of this model in a fog environment.

Recommendation systems play a crucial role in assisting users by recommending products and services that align with their preferences and needs. However, a common issue faced by the recommender systems is data sparsity, where limited user-item interactions lead to inadequate recommendation accuracy. To address these challenges, we propose a deep learning-based prediction model for a cross-domain recommendation system with multi-auxiliary domains through optimal knowledge transfer (DL-MD-OKT), which ensures consistent and accurate recommendations in cross-domain scenarios. We design a unified framework that integrates the quantum classical with deep residual learning (QC-DRL) model and joint matrix factorization to align user and item latent factor spaces, which can decrease the dispersion inconsistency between every helper and target region. Instead of using selective knowledge transfer, here we introduce optimal knowledge transfer, i.e., the multi objective non-dominated sorting-based butterfly optimization (NDS-BO) algorithm, which is used to optimally transfer more knowledge from the common matrix to a sparse target domain by transferring cluster-level estimation knowledge from multiple auxiliaries. In cross-domain recommendation scenarios, the superiority of DL-MD-OKT is validated with existing recent studies based on evaluation metrics such as mean absolute error (MAE) and root mean squared error (RMSE).

Sequential recommendation involves suggesting subsequent items in a series of user activities. When recommending relevant items to users within the same account, the challenge lies in discerning diverse user behaviours to provide tailored recommendations based on individual preferences and timing. Cross-domain sequential recommendation (CDSR) focuses on accurately extracting cross-domain user preferences from both within-sequence and between-sequence interactions among items. Current approaches typically concentrate on learning preferences within a single domain using intra-sequence item interactions, followed by a transfer module for cross-domain preferences. However, this sequential process and implicit method are constrained by the effectiveness of the transfer module and may overlook inter-sequence item associations. In this study, we propose an optimal cluster with attention-based shared-account cross-domain sequential recommendation (O-SCSR) system using deep reinforcement learning techniques. Our approach commences by formulating a modified hummingbird optimization (MHO) algorithm for clustering, effectively identifying latent users who share the same account to enhance the understanding of user interactions within shared-account scenarios. Additionally, we design a domain filter based on quantum classic deep reinforcement learning (QCDRL), intelligently selecting interactions contributing to O-SCSR. By quantifying rewards from transferred domain knowledge, the QCDRL-based filter retains only valuable interactions for the task of SCDR. Finally, we validate the efficacy of our proposed O-SCDR method using real-world datasets, namely HVIDEO and HAMAZON. Through simulation results comparing the O-SCDR system with existing state-of-the-art systems, we demonstrate its effectiveness and legitimacy.

This chapter explores the integration of AI techniques in water resource management for soil-based and soilless irrigation systems in agriculture. It emphasizes the importance of AI in optimizing water use and data-driven decision-making. AI-powered techniques like machine learning and predictive analytics enable precise water optimization. Soilless systems like hydroponics, aquaponics, and aeroponics also contribute to water efficiency. AI is crucial in weather forecasting, climate adaptation, crop water estimation, and water use optimization in water-stressed environments. The chapter discusses successful AI implementations, cost-benefit analysis, ethical, social, environmental considerations, equity, access, and sustainability. Future prospects, advancements, and challenges in AI techniques for water management are explored, emphasizing the need for AI adoption in water-efficient farming practices.

Innovative environments such as fog computing, which offer computation, data management, and other applications at the network edge, are gaining prominence. Fog networks with Wi-Fi offloading determined to be among the most acceptable methods for combating significant data increases in cellular networks. Fog nodes used to relay traffic and computations are situated nearby, making them excellent relay points. So, why networks have grown as rapidly as they have, and it's left its imprint on their performance. On the other hand, fog computing has recently developed to help Content-Centric Networking (CCN) decrease the amount of traffic. It has shown to be an effective method to reduce network latency and process time in industries and traffic. The use of Vehicle to Device technology for transferring network data in the mobile environment is expected to take off, particularly for autonomous cars. This study considers how to balance energy usage and service delay while allocating mobile traffic among vehicle networks. This article proposes a new paradigm for describing users' desire to provide their mobile vehicle resources to the public. The suggested methods are evaluated via comprehensive simulations, using various mobile movement traces. Additionally, Deep learning algorithm with how an online traffic monitoring for training and how important it is to know that train is possible. In conclusion, this methods proved to be better than previous benchmark techniques for fog networks.

When dealing with any kind of network traffic, an intrusion detection system is the essential element of security. This also applies to the Internet of Things (IoT) network, where the diverse components involved and the wireless connection method make cyber-related attacks more novel. So, it is necessary to use principles that a human would use to recognize attacks in order to comprehend the attacks and develop a workable method of detecting and responding to intrusions. The hardest element of using machine learning models for intrusion detection is that they can't handle the newer sorts of assaults, thus humans have to help by classifying the traffic as either normal or an attack using rules. In order to give the machine learning model, the capacity to employ hard-coded rules whenever necessary and to make predictions in circumstances where it cannot be done, the goal of this work is to incorporate hard-coded rules into the machine learning model. The attacks are recognized by an intrusion detection system that uses a hybrid rule-based machine learning model. The IoTID20 dataset is used for the attack classification, and the hybrid methodology's overall accuracy was 93.17%.

Recommender system (RS) aims to predict user preferences based on automatic data acquisition, and those collected data assist in achieving the final decision. However, RS suffers from data sparsity issues over the newly launched system, and the lack of time to deal with the massive data is also a challenging factor. To acquire proper outcomes, cross-domain RS intends to transfer knowledge from the specific domain with quality enriched data to help recommendations to the target domains. The entities may or may not be overlapped, and it is common for the entities of two domains to be overlapped. These overlapping entities may show variations in their target domain, and avoiding these issues leads to distorted prediction outcomes over the cross-domain RS. To address these issues, this research concentrates on modeling and efficient cross-domain RS using the generative and discriminative adversarial network (CRS-GDAN) model for kernel-based transfer modeling. Domain specific is considered to handle the feature space of overlapped entities, and transfer computation is adopted to handle the overlapping and non-overlapping entity correlation among the domains. Based on the anticipated concept, knowledge transfer is achieved rigorously even in the case of overlapping entities, thus diminishing the data sparsity issues. The experimentation is performed using an available online dataset, and the model attains a 20% better outcome than other approaches. The outcomes specify that the knowledge transfer from source to destination target is advantageous even in overlapping issues.

Recent years have shown us that even when we are not physically there, we may nevertheless connect to the entire world. The enormous growth in the number of linked gadgets has made this possible. Therefore, in times of emergency, our connected era generates enormous amounts of data that require immediate analysis. These days, the majority of businesses rely on cloud computing to store and handle enormous volumes of data. Cloud computing is not the best option for handling crucial activities, nevertheless, due to its compromises of high latency and slow response time. This necessitates the processing of those activities on edge hardware, or "fog computing." Although fog computing has low latency and quick response times, scheduling with the cloud and edge has been found to be inefficient. Randomness, a lack of resources, network structure, heterogeneity, and other elements all play a role in inefficient scheduling. Although there are a number of heuristic- and reinforcement-based solutions available, research on this topic is still active due to the systems' slow generalization, inability to swiftly adapt, and inability to handle workloads that are time-specific. This work focuses on developing a schedule using recurrent neural networks for parameter updating and hierarchical reinforcement learning (HIRO) to learn with an understanding of changing contexts. The proposed model is assessed for better scheduling decisions taking into account the workload's time-specific patterns and its capacity to fine-tune its parameters with various applications. The bitbrain dataset, which simulates real-world data, is used to evaluate the experimental. The improvement found in terms of energy consumption, reaction time, service level agreement, and cost when compared to cutting-edge algorithms is 15.1%, 8.2%, 34.2%, and 5.26%, respectively.

An intrusion detection system (IDS) becomes an important tool for ensuring security in the network. In recent times, machine learning (ML) and deep learning (DL) models can be applied for the identification of intrusions over the network effectively. To resolve the security issues, this paper presents a new Binary Butterfly Optimization algorithm based on Feature Selection with DRL technique, called BBOFS-DRL for intrusion detection. The proposed BBOFSDRL model mainly accomplishes the recognition of intrusions in the network. To attain this, the BBOFS-DRL model initially designs the BBOFS algorithm based on the traditional butterfly optimization algorithm (BOA) to elect feature subsets. Besides, DRL model is employed for the proper identification and classification of intrusions that exist in the network. Furthermore, beetle antenna search (BAS) technique is applied to tune the DRL parameters for enhanced intrusion detection efficiency. For ensuring the superior intrusion detection outcomes of the BBOFS-DRL model, a wide-ranging experimental analysis is performed against benchmark dataset. The simulation results reported the supremacy of the BBOFS-DRL model over its recent state of art approaches.

The drug discovery and design process has been significantly transformed by the integration of data science, artificial intelligence (AI), green chemistry principles, and affordable medicine. AI techniques enable rapid analysis of vast datasets, predicting molecular interactions, optimizing drug candidates, and identifying potential therapeutics. Green chemistry practices promote sustainability and efficiency, resulting in environmentally friendly and cost-effective production processes. The goal is to develop affordable medicines that are not only efficacious but also accessible to a wider population. This chapter explores case studies and emerging trends to highlight the transformation of the pharmaceutical industry and innovation in drug discovery.