Removing tokens whose embeddings diverge most from a source or CLS-token prototype, selected by output entropy, improves 3D point cloud classification under distribution shift without backpropagation.
6 Supplementary Material 6.1 Effect of Batch Size
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Purge-Gate: Backpropagation-Free Test-Time Adaptation for Point Clouds Classification via Token Purging
Removing tokens whose embeddings diverge most from a source or CLS-token prototype, selected by output entropy, improves 3D point cloud classification under distribution shift without backpropagation.