AI Driven Optimization of Resource Allocation and Cost Efficiency in Cloud Computing Environments
Year: 2025
Type: dissertation
Abstract: AI can improve resource allocation and minimise operational costs in cloud computing settings. Machine learning (ML) and reinforcement learning are examined. The study recommends two models. Linear regression, random forests, and neural networks were utilised to optimize resource consumption depending on workload, cost, and performance. Random Forest was the most accurate model, with an R² score of 0.9870, outperforming others in prediction. The second section employed RL to find smart resource-use strategies. By watching states, actions, and rewards interact, the DDPG and Q-learning algorithms learnt how to allocate resources flexibly. Classic rule-based approaches were inferior to RL models, which saved money, used resources better, and improved system reliability. The paradigm combines adaptive decision-making and predictive analytics, delivering a theoretical and practical contribution. The study reveals that AI-driven systems can manage cloud resources in real time with various benefits. The study found certain issues, such as tiny datasets and a lack of computing capacity, but it also opened new possibilities for studying more advanced RL approaches in the real world across multiple cloud platforms.
Source: Theseus (Ammattikorkeakoulujen)
Author: Ullah, Asad
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7064073961