Table 3 Wilcoxon signed-rank test results on Cytotoxicity prediction with feature selection (Macro-F1 over 5-fold CV). One-sided test with the alternative hypothesis \(H_1\): TabPFN > baseline.
From: Boosting pre-trained model with silica nanoparticles cellular toxicity prediction
Comparison | Statistic (W) | p-value | Significant |
|---|---|---|---|
TabPFN vs GradientBoosting | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs RandomForest | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs LogisticRegression | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs XGBoost | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs SVC | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs BPNN | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs FTTransformer | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs TabTransformer | 15.0000 | 0.031250 | \(\checkmark\) |
TabPFN vs CatBoost | 15.0000 | 0.031250 | \(\checkmark\) |
