XLRON: Accelerated Reinforcement Learning Environments for Optical Networks (Th2A.21)
Presenter: Michael Doherty, University College London
We present XLRON: an open source project enabling, for the first time, GPU-accelerated reinforcement learning on optical network problems. We demonstrate 100-1000x speed-up in training time over similar tools, thereby opening new research possibilities.
Authors:Michael Doherty, University College London / Alejandra Beghelli, University College London