Table 9 Details of the machine learning (ML) approaches used to model and predict different MXene outputsa
Year, authors | Output | Data size | Feature selection | ML problem | Algorithm | Validation /Test |
|---|---|---|---|---|---|---|
Rajan et al.54 | Metal−semiconductor identification | 643 | LASSO | Classification | Bagging | 5-fold CV, 10 % test |
Rajan et al.54 | Band-gap estimation | 70 | LASSO | Regression | Gaussian process | 5-fold CV, 10 % test |
Frey et al.209 | Synthesizability | 792 | PCA | Positive and unlabeled learning | Robust ensemble SVM | 10-fold CV |
Mishra et al.212 | Valence band | 76 | LASSO, NCA | Regression | Gaussian process | 10 % test |
Marchwiany et al.219 | Cytotoxicity | 71 | Random forest feature importance | Classification | Random forest | 10-fold CV |
Venturi et al.214 | Mechanical strength, band gap and formation energy | >3500 | --- | Regression | CGCNNs | 15 % validation; 15 % test |
Wang et al.222 | HER | 420 | RFE; feature importance; Pearson correlation coefficient | Regression | AdaBoost | 10-fold CV |
Zheng et al.221 | HER | 299 | Pearson correlation coefficient | Regression | Random forest | 10-fold CV, 25 % test |
He et al.210 | Stability | 85 | Pearson correlation coefficient; Symbolic regression | Classification | SVM | 20 % test |
Li et al.226 | Energy storage (gravimetric capacity, voltage, and induced charge) | 360 | RFE | Regression and Classification | Multi-output random forest | 5-fold CV; 20 % test |
Song et al.218 | Saturation magnetization | 23825 | --- | Classification | AdaBoost | 10-fold CV; 10 % test |
Tian et al.216 | Tensile stiffness | 157 | --- | Regression | SISSO | --- |
Abraham et al.229 | CO2 activation | 114 | Feature importance | Regression | Random forest | 5-fold CV; 20 % test |
Boonpalit et al.230 | CO sensing | 450 | --- | Regression | CGCNNs | 20 % validation |
Chen et al.224 | ORR and OER | 78 | Feature importance | Regression | Random forest | 4-fold CV; 25 % test |
Cheng et al.227 | Hydrogen adsorption distance | 12647 | --- | Classification and regression | ALIGNNs | 5-fold CV |
Ding et al.232 | Solar spectral absorption | 500 | --- | Regression | Random forest | 10 % test |
Jiao et al.231 | C−N coupling | 54 | LASSO | Regression | SISSO | 10-fold CV |
Ma et al.225 | ORR and OER | 42 | Expert knowledge criteria | Regression | Random forest and gradient boosting | 5-fold CV; 10 % test |
Liang et al.223 | HER (Gibbs free energy of hydrogen adsorption) | 264 | Pearson correlation coefficient; feature importance | Regression | Random forests | 30 % test |
Liang et al.223 | HER (cohesive energy) | 264 | Pearson correlation coefficient; feature importance | Regression | Random forest | 30 % test |
Roy et al.213 | Work function | 275 and 315 | Genetic algorithm | Regression | Artificial neural networks | 5-fold CV; 20 % test |
