A compression method removes near-always-active ReLU neurons and replaces them with shortcut connections, shrinking some fully connected models to one quarter of their size with modest accuracy loss.
Alsaadi, and Xiaohui Liu
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Linearity-based neural network compression
A compression method removes near-always-active ReLU neurons and replaces them with shortcut connections, shrinking some fully connected models to one quarter of their size with modest accuracy loss.