Cloud computing offers a scalable and cost-effective platform by providing on-demand access to shared computational resources. Efficient load balancing is essential to maintain optimal performance and maximize resource utilization, ensuring an even distribution of network traffic across servers, preventing overload, enhancing response times, and improving system reliability. This paper proposes a Meta Reinforcement Learning Driven Hybrid Lyrebird Falcon Optimization for Dynamic Load Balancing in Cloud Computing (Meta-RHDC), a novel approach for dynamic load balancing in cloud environments. The Meta-RHDC model leverages convolutional and recurrent neural networks to predict virtual machine loads and dynamically classify them into overloaded and underloaded categories. By integrating reinforcement learning with advanced optimization techniques, Meta-RHDC significantly improves task scheduling and load balancing compared to existing methods such as Load Optimization Algorithm (LOA), Reinforcement Learning (RL), and Falcon Optimization Algorithm (FOA). Extensive experiments conducted on the CloudSim platform demonstrate that Meta-RHDC achieves substantial improvements in key performance metrics. For task counts, makespan is reduced by 19.51%, energy consumption by 22.75%, and balanced CPU utilization by 21.98%. Resource utilization increases by 32.52%, scalability efficiency improves by 49.03%, and the failure rate decreases by 19.72%. For virtual machine (VM) counts, makespan improves by 30.57%, energy consumption by 42.59%, and balanced CPU utilization by 36.85%. Resource utilization rises by 31.75%, scalability efficiency increases by 36.61%, and the failure rate drops by 38.19%. These results confirm the robustness and efficiency of Meta-RHDC in optimizing execution, resource management, and scalability across diverse workloads, making it a superior solution for dynamic load balancing in cloud computing.

Task scheduling poses a wide variety of challenges in the cloud computing paradigm, as heterogeneous tasks from a variety of resources come onto cloud platforms. The most important challenge in this paradigm is to avoid single points of failure, as tasks of various users are running at the cloud provider, and it is very important to improve fault tolerance and maintain negligible downtime in order to render services to a wide range of customers around the world. In this paper, to tackle this challenge, we precisely calculated priorities of tasks for virtual machines (VMs) based on unit electricity cost and these priorities are fed to the scheduler. This scheduler is modeled using a deep reinforcement learning technique which is known as the DQN model to make decisions and generate schedules optimally for VMs based on priorities fed to the scheduler. This research is extensively conducted on Cloudsim. In this research, a real-time dataset known as Google Cloud Jobs is used and is given as input to the algorithm. This research is carried out in two phases by categorizing the dataset as a regular or large dataset with real-time tasks with fixed and varied VMs in both datasets. Our proposed DRFTSA is compared to existing state-of-the-art approaches, i.e., PSO, ACO, and GA algorithms, and results reveal that the proposed DRFTSA minimizes makespan compared to PSO, GA, and ACO by 30.97%, 35.1%, and 37.12%, rates of failure by 39.4%, 44.13%, and 46.19%, and energy consumption by 18.81%, 23.07%, and 28.8%, respectively, for both regular and large datasets for both fixed and varied VMs.

Traditional approaches for video analysis often misdefine anomalies; they usually rely on single-modality input and have inadequate management of complex temporal patterns. This paper resolves these limitations by proposing a comprehensive scheme for multimodal Closed-Circuit Television (CCTV) video analysis. The utilized techniques in this paper comprise the Multimodal Deep Boltzmann Machine (MDBM), Multimodal Variational Autoencoder (MVAE), and Attention-based Fusion Networks, all of which fully utilize the learned representations. MDBM learns shared representations out of heterogeneous data sources, MVAE captures the inherent distribution of multi-modalities, while the mechanism of attention in fusion networks is done to stress important features. Finally, temporal context is modeled using long short-term memory and transformer networks, temporal convolutional networks, and transformer networks with temporal encoding. Long Short-Term Memory (LSTM) can capture long-range dependencies in sequential data, while Temporal Convolutional Network (TCN) efficiently models temporal patterns using convolutional layers and Transformer Networks fathom the relative importance of temporal features against one another through self-attention, thus improving their detection accuracy for anomalies that happen over a long duration. The proposed models also offer good improvements in the performance of anomaly detection. In particular, accuracy improved by 5% using MDBM, the false positive rate reduced by 15% with MVAE, a more than 10% improvement in the F1-score with the attentive fusion network, a 20% reduction in reconstruction error with Deep Convolutional Autoencoder (DCA), detection precision improved by 12% using Adversarially Learned Inference (ALI), and a gain of 8% in Area Under the Curve (AUC) using Deep InfoMax (DIM) operations.

Cloud computing has transformed IT service delivery by offering scalable, on-demand resources essential for managing diverse and complex workloads. However, efficient task scheduling remains a critical challenge due to the NP-hard nature of mapping workflows to virtualized resources, impacting performance, resource utilization, energy efficiency, and reliability. This highlights the urgent need for advanced scheduling strategies.This paper presents ADWEH (A Dynamic Prioritized Workflow Task Scheduling approach based on the Enhanced Harris Hawk Optimization), a novel framework designed to overcome these challenges. ADWEH combines deep reinforcement learning with dynamic prioritization strategies to assign task priorities based on dependencies and runtime requirements. This integrated approach ensures optimal resource allocation, enhances scheduling efficiency, and improves adaptability to varying workloads.The proposed approach has been implemented and thoroughly tested using WorkflowSim, a simulation toolkit for modeling and evaluating workflow scheduling strategies. ADWEH’s performance is benchmarked against state-of-the-art algorithms, including DQN, A2C, and PWSA3C, using scientific workflows including Montage, Cybershake, SIPHT, LIGO. These Four workflows are tested on datasets like HPC2N, NASA, Uniform, and Random, ensuring robust evaluation under diverse scenarios. Simulation results demonstrate that ADWEH significantly outperforms DQN, A2C, and PWSA3C approaches, achieving up to 31.9% shorter makespan, a 24% reduced in energy consumption, a 29% Enhancement in scalability efficiency, and a 52% increase in resource utilization, while reducing failure rates by over 47%. The results position ADWEH as a robust, efficient solution for improving the performance , reliability of cloud-based workflows, addressing the evolving demands of modern cloud environments.

Scheduling of workflows is a critical issue in cloud computing paradigm as dynamic workflows with variable dependencies arises from heterogeneous resources which comes onto cloud application console. Mapping all these interdependent tasks to precise virtual resources is still a challenge for cloud service provider(CSP) as workflows arises to cloud application console varied from time to time with different running time capacities, dependencies, length of tasks. Ineffective scheduling of workflows creates challenges for CSP, cloud users by delaying execution of tasks, increase in rate of failures, underutilization of resources. Many earlier authors proposed various workflow scheduling mechanisms with different metaheuristic optimization approaches but generating schedules in cloud computing is a NP-hard problem and they may suffer when huge workloads coming onto cloud paradigm. Therefore, to handle the above said concerns in cloud paradigm, authors in this research proposed a prioritized workflow scheduler by Asynchronous Advantage Actor Critic (PWSA3C) algorithm by considering priorities of interdependent tasks, number of dependencies to schedule workflows to precise VMs. Extensive simulations for proposed PWSA3C algorithm conducted using workflowsim. Input to proposed approach was given with scientific workflows named Epigenomics, LIGO. For evaluating robustness of proposed PWSA3C, we compared our approach with DQN, A2C algorithms. From simulated results, it is evident that PWSA3C outperformed over DQN algorithm for makespan, rate of failures, utilization of resources, scalability efficiency with Epigenomics, LIGO workflows.

The rapid growth of Internet scale computing has exposed critical limitations in existing edge cloud coordination mechanisms, particularly in terms of resilience, energy efficiency, and adaptability under heterogeneous and dynamic environments. Current optimization and learning based approaches often suffer from slow convergence, limited exploration capability, and poor robustness when managing distributed edge cloud resources. To address these challenges, this research proposes QRGEC: Quantum Reinforcement Learning with Golden Jackal Optimization for Resilient Edge Cloud Coordination in Internet Computing. The proposed hybrid framework integrates quantum focused policy exploration with adaptive metaheuristic tuning to enhance distributed Internet computing optimization. Policy representations are encoded using variational quantum circuits, enabling efficient exploration of high dimensional decision spaces. Furthermore, the Golden Jackal Optimization mechanism adaptively adjusts reinforcement parameters to improve convergence stability and accelerate learning, thereby enabling resilient and energy-efficient coordination across heterogeneous edge and cloud environments. A resilience aware scheduler seamlessly balances energy efficiency, latency, and recovery within dynamic edge cloud workloads. Extensive experimental evaluations in QRGEC demonstrate that the framework is capable of outperforming previously established deep reinforcement and quantum heuristic baselines with a latency reduction of 36.8%, an increase in energy efficiency of 24.7%, an improvement in resilience of 48.2%, and a sustained resource utilization of 94%. QRGEC also displays the ability to automatically recover from network congestion and failures, recover from network congestion, and maintain balances latency energy trade-offs. This emphasizes the efficiency of QRGEC in autonomous recovery from network failures, making latency energy balance adjustments, and conserving energy.