Semantic Memory-Regularized UAMetaRL for Adaptive Task Scheduling in 6G Edge Intelligence
Abstract
This repository provides the source code, configuration files, synthetic datasets, and experimental results for the paper entitled "Semantic Memory-Regularized Uncertainty-Aware Meta-Reinforcement Learning for Adaptive Task Scheduling in 6G Edge Intelligence". The repository implements a semantic memory-regularized uncertainty-aware meta-reinforcement learning framework for adaptive task scheduling in 6G edge intelligence environments. It includes modules for semantic context construction, memory-based adaptation, meta-policy selection, uncertainty estimation, governance validation, execution-level reinforcement learning, safety control, and experimental evaluation. The provided implementation enables reproduction of the reported experiments, including baseline comparisons, ablation studies, statistical analysis, and performance evaluations under dynamic edge computing scenarios.
