A new XAI method clusters a network's most-confused labels into a WordNet-labeled hierarchy, and its experiments suggest larger models form more human-aligned concepts.
D.; Ravi, S.; and Ramavajjala, V
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Explainable AI Approach using Near Misses Analysis
A new XAI method clusters a network's most-confused labels into a WordNet-labeled hierarchy, and its experiments suggest larger models form more human-aligned concepts.