<p>Scheduling is widely recognized as a crucial factor in enhancing the efficiency of cloud computing services. In this context, the scheduling process must be designed to minimize the task completion time while maximizing the productivity of available resources. Numerous task scheduling strategies have been introduced in the literature, and efforts to develop optimal scheduling models remain ongoing. Therefore, this paper presents a scheduling method based on a neural network optimizer in heterogeneous Infrastructure as a Service (IaaS) cloud environments, which seeks to find a more suitable schedule. The proposed method integrates a learning-based optimization algorithm, a neural network mechanism, and mutation operators from evolutionary algorithms to achieve a balance between the exploration and exploitation phases. To evaluate the performance and efficiency of the proposed algorithm, three sets of experiments were conducted. First, the basic NNA and TLBO algorithms were tested as global optimizers alongside the proposed algorithm. Then, experiments using the CloudSim platform were performed to evaluate the proposed algorithm. In the second phase, experiments were carried out using a synthetic dataset generated by the simulator. Finally, the algorithms were evaluated based on two real datasets. The simulation results demonstrate that the proposed algorithm outperforms the IBWC, HGHHC, MSDE, EMPA, and FPGWO algorithms in terms of makespan, resource utilization, degree of imbalance, and throughput.</p>