- Directory:
xai - Purpose: Provides XAI (Explainable Artificial Intelligence) service.
- Details:
- Implements various CAM (Class Activation Mapping) methods for different models.
- Utilizes the
ImageNet-Minidataset as input for the analysis.(structure constant for other datasets) - Currently, the XAI explanations are generated and stored in the
resultsfolder. should send to database.
- Directory:
feature transform - Notebook:
masking - Purpose: To transform XAI results into masking images.
- Details:
- Reads from the
resultsdirectory. - Processes the XAI results and generates masking images.
- Masking images are stored in the
evaluationfolder.
- Reads from the
- Directory:
prediction_changes - Notebook:
pcd.ipynb - Purpose: Evaluates the impact of the feature transformations.
- Details:
- Reads from the
evaluationdirectory. - Calculates prediction changes scores based on the masked images.
- Provides a comprehensive statistical analysis of the results.
- Reads from the
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent