Cloud services have gained widespread adoption in recent years due to features such as pay-as-you-go pricing and dynamic scalability. However, one of the main challenges in such environments is the autonomous and optimal management of resources with the aim of reducing costs, complying with Service Level Agreements (SLAs), and improving quality of service. Over-provisioning of resources can lead to increased operational costs and inefficient resource utilization, whereas under-provisioning results in violations of SLA terms and incurs associated penalties. In this paper, a hybrid approach for autonomous cloud resource allocation is proposed, which employs a Hopfield neural network for dynamic prediction of future workload and a fuzzy multi-criteria VIKOR decision-making method for optimal selection of resource allocation policies. The proposed method has been evaluated in the CloudSim simulation environment under various real and synthetic workloads. Simulation results demonstrate that, compared to other benchmarked methods, the proposed approach achieves a significant average reduction of 6% in total cost, 4% in average response time, and 4% in the number of SLA violations. These improvements stem from the integration of accurate workload prediction with a fuzzy multi-criteria decision-making algorithm, which enables autonomous and intelligent resource management in cloud environments with dynamic workloads