Deep Learning Neural Networks in the Cloud

International Journal of Advanced engineering Management and Science, Journal Year: 2023, Volume and Issue: 9(10), P. 09 - 26

Published: Jan. 1, 2023

Deep Neural Networks (DNNs) are currently used in a wide range of critical real-world applications as machine learning technology. Due to the high number parameters that make up DNNs, and prediction tasks require millions floating-point operations (FLOPs). Implementing DNNs into cloud computing system with centralized servers data storage sub-systems equipped high-speed high-performance capabilities is more effective strategy. This research presents an updated analysis most recent computing. It highlights necessity while presenting debating numerous DNN complexity issues related various architectures. Additionally, it goes their intricacies offers thorough several platforms for deployment. examines already running on highlight advantages using DNNs. The study difficulties associated implementing systems provides suggestions improving both current future deployments.

Language: Английский

---

Enhanced multivariate singular spectrum analysis‐based network traffic forecasting for real time industrial IoT applications

IET Networks, Journal Year: 2024, Volume and Issue: 13(4), P. 301 - 312

Published: March 12, 2024

Abstract Industrial IoT (IIoT) applications are widely used in multiple use cases to automate the industrial environment. Industry 4.0 presents challenges numerous areas, including heterogeneous data, efficient data sensing and collection, real‐time processing, higher request arrival rates, due massive amount of data. Building a time‐sensitive network that supports voluminous dynamic traffic from is complex. Therefore, authors provide insights into networks propose strategy for enhanced management. An multivariate forecasting model adapts Multivariate Singular Spectrum Analysis employed an SDN‐based IIoT network. The proposed method considers flow parameters, such as packet sent received, bytes source rate, round trip time, jitter, rate duration predict future flows. experimental results show can effectively by contemplating every possible variation observed samples average load, delay, inter‐packet sending with improved accuracy. forecast shows reduced error estimation when compared existing methods Mean Absolute Percentage Error 1.64%, Squared 11.99, Root 3.46 2.63.

Language: Английский

---

A workload prediction model for reducing service level agreement violations in cloud data centers

Decision Analytics Journal, Journal Year: 2024, Volume and Issue: 11, P. 100463 - 100463

Published: April 15, 2024

Cloud computing has become an emerging technology that offers services based on the pay-as-usage model. The cloud provides several advantages, but these advantages come with challenges, such as reducing Service Level Agreement (SLA) violations, efficient resource utilization, energy consumption, etc., needing attention to leverage customer satisfaction and benefit service providers. Workload prediction is a strategy many benefits: reduced SLA violation, scaling, optimization by predicting future workload. However, due varying workload of applications, it difficult predict accurately, fails for long-term dependencies. We propose methodology Multiplicative Long Short Term Memory (mLSTM) allows input-dependent transitions considers dependencies address this issue. proposed method implemented compared other variants LSTM used in literature purposes. work outperforms existing terms accuracy.

Language: Английский

---

Comprehensive enhancements for machine-learning based cloud resource orchestration algorithms

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: June 12, 2024

With the increasing utilisation of machine-learning algorithms, autoscaling methods have potential to provide more sophisticated control mechanisms for cloud application operators over virtualized resources in terms provisioning. This paper introduces machine learning-based approach introducing several improvements including metric selection, proactivity, various enhancements neural network among others. Our enhanced learning models enables autoscaler algorithm react quickly sudden load changes, increase proactivity while using fewer and reducing Quality Service (QoS) violations cloud-based services. The comprehensive measurements indicate that QoS may reduce by up 80%, level resource utilization either remains constant or decreases slightly (by 3-4%) certain applications. In cases where there was no reduction violations, saw a significant decline, falling between 20-50%. proposed changes been analyzed tested under conditions, representing 3 distinct common use environments. developed process has implemented on science Hungarian Research Network supporting operation infrastucture hosts 300 scientific projects.

Language: Английский

---

A Comparative Analysis of Generative Adversarial Networks for Generating Cloud Workloads

Published: July 7, 2024

Language: Английский

---