|Reinforcement Learning with Neural Networks for Quantum Multiple Hypothesis Testing
|Sarah Anne Brandsen, Kevin Stubbs, Henry Pfister, Duke University, United States
|Q.3: Quantum Error Correcting Codes and Decoding
|Quantum Systems, Codes, and Information
|Click here to download the manuscript
|Click here to watch in the Virtual Symposium
|Reinforcement learning with neural networks (RLNN) has demonstrated great promise for problems in quantum information theory. We apply RLNN to the problem of quantum hypothesis testing, where one tries to distinguish between multiple quantum states while minimizing the error probability. Finding the minimal error measurement is a challenging problem in general, and implementing the optimal measurement on the entire system may be impractical. In this work, we develop an experimentally feasible, locally adaptive measurement strategy for multiple hypothesis testing via RLNN. Numerical results for the network performance are shown. Finally, we introduce the concept of weak and strict t-locality and prove a partial ordering that allows us to simplify the total number of locally adaptive algorithms that need to be considered.