Gaussian Processes for the Analysis of Electrochemical Impedance Spectroscopy Data: Prediction, Filtering, and Active Learning

Abstract

Electrochemical impedance spectroscopy (EIS) is a widespread characterization technique used to study electrochemical systems. However, several shortcomings still limit the application of this technique. First, EIS data is intrinsically noisy, hindering spectra regression and prediction at unknown frequencies. Second, many physicochemical properties, such as the polarization resistance, are determined through non-unique equivalent circuits. Third, probed frequencies are usually log-spaced with a fixed number of points per decade, which is not necessarily optimal. This article illustrates how Gaussian processes (GPs) can overcome these three issues by showing that GPs can successfully filter EIS data and be used to determine the polarization resistance as a stochastic variable. Lastly, a GP-based, active-learning framework is developed to select EIS frequencies optimally for quick and accurate measurements.