Stabilizing Deep Reinforcement Learning Model Training for Video Conferencing | IEEE Conference Publication | IEEE Xplore