Accelerating Distributed Reinforcement learning with In-Switch Computing | IEEE Conference Publication | IEEE Xplore