Hybrid Learning for Cold-Start-Aware Microservice Scheduling in Dynamic Edge Environments
IEEE Transactions on Mobile Computing, vol. 25, no. 5, pp. 7338–7353
This work combines offline behavior cloning with online Soft Actor-Critic to optimize latency and energy in dynamic edge environments. A rule-based expert, GRU-enhanced policy, and feasibility masking improve early-stage learning, convergence speed, and robustness.