Time-dependent atomic magnetometry with a recurrent neural network

TL;DR

This paper introduces a recurrent neural network approach for estimating fluctuating magnetic fields from optical measurements, offering a flexible alternative to traditional Kalman filters without restrictive assumptions.

Contribution

It demonstrates that an encoder-decoder neural network can accurately infer time-dependent magnetic fields from optical data, surpassing traditional methods in flexibility.

Findings

Neural network achieves comparable accuracy to Kalman filters.

The method is free from restrictive theoretical assumptions.

It effectively processes continuous optical measurement data.

Abstract

We propose to employ a recurrent neural network to estimate a fluctuating magnetic field from continuous optical Faraday rotation measurement on an atomic ensemble. We show that an encoder-decoder architecture neural network can process measurement data and learn an accurate map between recorded signals and the time-dependent magnetic field. The performance of this method is comparable to Kalman filters while it is free of the theory assumptions that restrict their application to particular measurements and physical systems.