Fig. 5From: Decentralized computation offloading for multi-user mobile edge computing: a deep reinforcement learning approachAverage reward per episode in the training process for a single user agent with w1=0.8Back to article page