We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:


References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Vision and Pattern Recognition

Title: Refining activation downsampling with SoftPool

Abstract: Convolutional Neural Networks (CNNs) use pooling to decrease the size of activation maps. This process is crucial to increase the receptive fields and to reduce computational requirements of subsequent convolutions. An important feature of the pooling operation is the minimization of information loss, with respect to the initial activation maps, without a significant impact on the computation and memory overhead. To meet these requirements, we propose SoftPool: a fast and efficient method for exponentially weighted activation downsampling. Through experiments across a range of architectures and pooling methods, we demonstrate that SoftPool can retain more information in the reduced activation maps. This refined downsampling leads to improvements in a CNN's classification accuracy. Experiments with pooling layer substitutions on ImageNet1K show an increase in accuracy over both original architectures and other pooling methods. We also test SoftPool on video datasets for action recognition. Again, through the direct replacement of pooling layers, we observe consistent performance improvements while computational loads and memory requirements remain limited.
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2101.00440 [cs.CV]
  (or arXiv:2101.00440v3 [cs.CV] for this version)

Submission history

From: Alexandros Stergiou MSc [view email]
[v1] Sat, 2 Jan 2021 12:09:49 GMT (57042kb,D)
[v2] Tue, 5 Jan 2021 08:52:31 GMT (57045kb,D)
[v3] Thu, 18 Mar 2021 09:51:42 GMT (39711kb,D)

Link back to: arXiv, form interface, contact.