Recently, the technological development in edgecomputing and content caching can provide high-quality servicesfor users in the wireless communication networks. As a promisingtechnology, multi-access edge computing (MEC) can offload tasksto the nearby edge servers, which alleviates the pressure ofusers. However, various services and dynamic wireless channelconditions make effective resource allocation challenging. Inaddition, network slicing can create a logical virtual networkand allocate resources flexibly among multiple tenants. In thispaper, we construct an integrated architecture of communication,computing and caching to solve the joint optimization problem oftask scheduling and resource allocation. In order to coordinatenetwork functions and dynamically allocate limited resources,this paper adopts an improved deep reinforcement learning(DRL) method, which fully jointly considers the diversity of userrequest services and the dynamic wireless channel conditions toobtain the mobile virtual network operator (MVNO) maximalprofit function. Considering the slow convergence speed of theDRL algorithm, this paper combines DRL and ensemble learning. The simulation result shows that the resource allocation schemeinspired by DRL is significantly better than the other comparedstrategies. The output of the result of DRL algorithm combinedwith ensemble learning is faster and more cost-effective.