Universal pooling is a channel-wise attention-based trainable pooling layer that generalizes existing fixed pooling operations and outperforms them on two benchmark problems by adapting to the data.
Learning multiple layers of features from tiny images
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Universal Pooling -- A New Pooling Method for Convolutional Neural Networks
Universal pooling is a channel-wise attention-based trainable pooling layer that generalizes existing fixed pooling operations and outperforms them on two benchmark problems by adapting to the data.