Quantum Seminar: A NEAT Quantum Error Decoder

Zoom: https://tau-ac-il.zoom.us/j/82245900781

Abstract:

I will discuss how quantum error decoding / correcting on a stabiliser code can be formulated as a single player game, thereby making it amenable to solving using reinforcement learning methods. We can then discuss different agents that learn to play this game, and will focus in particular on the evolutionary “NEAT” algorithm. 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. If time permits, I will also quickly divert to showcasing ongoing work revolving around experimental optimisations (tuning quantum dots, reading out spin-qubits, etc) using ML-inspired algorithms.