Fig. 4: SHapley Additive exPlanations (SHAP) model interpretation and Finite Element (FE) simulations to uncover complex fabrication–structure–property correlations.
From: Machine intelligence accelerated design of conductive MXene aerogels with programmable properties
(a) Normalized SHAP values of MXene, CNF, gelatin, GA loadings, and mixture loading on the \({\sigma }_{30}\) values of conductive aerogels. b Comparison between the FE-simulated and experimentally characterized \({\sigma }_{30}\) values of conductive aerogels at the same MXene/CNF/gelatin/GA ratio yet at different mixture loadings. Data are presented as mean ± s.d., n = 3, with each independent experiment marked by an open black circle. c SEM images of high-, medium-, and low-density conductive aerogels at their uncompressed states. d Localized stress distribution profiles of high-, medium-, and low-density aerogel models (from FE simulations) under 30% compression. e SEM images of high-, medium-, and low-density conductive aerogels at their compressed states. Error bars represent s.d.
