Vision networks whose bottlenecks avoid feature-map interference scale better than MobileNet-style designs at very low parameter counts, and the new NoDepth block demonstrates this on ImageNet.
Softmax Linear Units
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
-
Exploring Superposition and Interference in State-of-the-Art Low-Parameter Vision Models
Vision networks whose bottlenecks avoid feature-map interference scale better than MobileNet-style designs at very low parameter counts, and the new NoDepth block demonstrates this on ImageNet.