Interpretable Cross-Examination Technique (ICE-T): Using highly informative features to boost LLM performance.
Goran MuricBen DelaySteven MintonPublished in: CoRR (2024)
Keyphrases
- highly informative
- low level
- feature space
- specific features
- co occurrence
- decision trees
- feature extraction
- object recognition
- image features
- feature set
- key features
- genetic algorithm
- extracted features
- salient features
- feature detection
- structural information
- neural network
- high resolution
- prior knowledge
- feature vectors
- artificial intelligence
- high level