Neural-network-based qubit-environment characterization

The exact microscopic structure of the environments that produces $1/f$ noise in superconducting qubits remains largely unknown, hindering our ability to have robust simulations and harness noise. In this paper we show how it is possible infer information about such an environment based on a single measurement qubit coherence, circumventing any need for separate spectroscopy experiments. Similarly other spectroscopic techniques, used as probe which interacts with its environment. complexity relationship between observed dynamics impurities makes problem ideal machine learning methods - more specifically neural networks. With algorithm are able reconstruct parameters most prominent environment, well differentiate different models, paving way towards better understanding circuits.