ºÚÁÏÍø

Abstract: 

We investigate the use of the evolutionary NEAT algorithm for finding a quantum error decoder for the toric code. The problem is formulated as a single player game, making it amenable to a reinforcement learning approach, with an agent whose policy network we directly optimise. We find that these NEAT-optimized network decoders have similar performance to previously reported machine-learning based decoders, but use roughly three to four orders of magnitude fewer parameters to do so.

[1] https://arxiv.org/abs/2101.08093