It is a kind of computing that uses centralized remote servers and the internet to track data and applications. Users of cloud computing can access data, applications, and other digital resources from any device connected with a connection to the internet. Task scheduling is the main concern in cloud-fog environments because of the heavy load on the servers. Traditional meta-heuristic algorithms are not feasible because of their high response time. Therefore, some researchers have worked on machine learning algorithms. The need for low latency in cloud-fog environments has motivated researchers to focus on machine learning approaches. In this work meta-heuristic and machine learning algorithms related to the scheduling of tasks in cloud-fog scenario were analyzed, and improvements were suggested, such as hybrid machine learning algorithms, which provide better latency than previous algorithms. Hybrid models offer intelligent, adaptive, and scalable solutions by combining supervised, unsupervised, reinforcement learning, and traditional optimization methods.