Rad4XCNN: a new agnostic method for post-hoc global explanation of CNN-derived features by means of radiomics.
Francesco PrinziCarmelo MilitelloCalogero ZarcaroTommaso Vincenzo BartolottaSalvatore GaglioSalvatore VitabilePublished in: CoRR (2024)
Keyphrases
- post hoc
- feature set
- significant improvement
- extracted features
- prior knowledge
- computational cost
- high accuracy
- high precision
- experimental evaluation
- preprocessing
- similarity measure
- probabilistic model
- support vector machine
- cost function
- detection method
- classification method
- objective function
- feature representation
- neural network
- test images
- similarity function
- classification process
- global shape