MHEX inserts attention-gated deep-supervision heads into ResNet and BERT and derives saliency maps from the product of the head weights, claiming better accuracy and more detailed explanations.
2020.9206626
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Multi-Head Explainer: A General Framework to Improve Explainability in CNNs and Transformers
MHEX inserts attention-gated deep-supervision heads into ResNet and BERT and derives saliency maps from the product of the head weights, claiming better accuracy and more detailed explanations.