A new regularizer that constrains a distance-based self-similarity score of hidden features improves accuracy on MLP and transformer models by up to 6 points, but the metric itself is not validated.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Self-similarity Analysis in Deep Neural Networks
A new regularizer that constrains a distance-based self-similarity score of hidden features improves accuracy on MLP and transformer models by up to 6 points, but the metric itself is not validated.