Inquiry into the Appropriate Data Preprocessing of Electrochemical Impedance Spectroscopy for Machine Learning

Electrochemical impedance spectroscopy (EIS) is an important
analytical
technique for the understanding of electrochemical systems. With the
recent advent and burgeoning deployment of machine learning (ML) in
EIS analysis, a critical yet hitherto unanswered question emerges:
what is the appropriate manner to preprocess the EIS data for ML-based
analysis? While the preprocessing of a model’s input data is
known to be critical for a successful deployment of the ML model,
EIS is known to possess multiple classical venues of data representation,
and moreover, a proper data normalization protocol for comparative
EIS studies remains elusive. Here, we report the methodology and the
outcomes that evaluate the efficacy of multiple data preprocessing
methods in an ML-based EIS analysis. Within our proof-of-concept parameter
space, plotting the input training data’s impedance magnitude
(|Z|) against phase angle (φ) while individually
normalizing each EIS curve yields the highest accuracy and robustness
in the correspondingly established residual neural network (ResNet)
model. Rationalized by additional “importance” analysis
of the input data, such a data representation method extracts information
and hidden features more effectively. While the Nyquist plot is widely
used in manual analysis, a different data representation of EIS data
seems equally plausible for ML-based EIS analysis. Our work offers
a protocol for future researchers to decide on the proper preprocessing
method for different ML applications in electrochemistry on a case-by-case
basis.