Generalizable Reinforcement Learning-Based Coarsening Model for Resource Allocation over Large and Diverse Stream Processing Graphs | IEEE Conference Publication | IEEE Xplore