| Converter and converted modelR6 class for converting a passed model into a torch-based model with all methods pre-implemented in each layer | |
|---|---|
| Converter of an artificial neural network | |
| Converted torch-based model | |
| Feature attribution methodsAll implemented feature attribution methods as R6 classes applying the corresponding method to the converted model during initialization | |
| Vanilla Gradient and Gradient\(\times\)Input | |
| SmoothGrad and SmoothGrad\(\times\)Input | |
| Integrated Gradients | |
| Expected Gradients | |
| Layer-wise relevance propagation (LRP) | |
| Deep learning important features (DeepLift) | |
| Deep Shapley additive explanations (DeepSHAP) | |
| Connection weights method | |
| Local interpretable model-agnostic explanations (LIME) | |
| Shapley values | |
| 
 | Syntactic sugar for object construction | 
| Visualization and getting resultsFunctions and S4 classes for getting and visualizing the results | |
| Get the result of an interpretation method | |
| Get the result of an interpretation method | |
| S4 class for ggplot2-based plots | |
| S4 class for plotly-based plots | |
| 
 | |
| 
 | Generic print, plot and show for  | 
| 
 | Indexing plots of  | 
| Generic add function for  | |
| 
 | |
| 
 | Generic print, plot and show for  | 
| Indexing plots of  | |
| Superclasses and package informationSuper classes intended only for the user’s information | |
| Get the insight of your neural network | |
| Super class for model-agnostic interpretability methods | |
| Super class for gradient-based interpretation methods | |
| Super class for interpreting methods | |