mlsanityData to 7 Oct 2026

Domino Saliency Metrics: Improving Existing Channel Saliency Metrics with Structural Information

arXiv 2205.02131 ↗First posted 4 May 2022cs.CV cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Channel pruning is used to reduce the number of weights in a Convolutional Neural Network (CNN). Channel pruning removes slices of the weight tensor so that the convolution layer remains dense. The removal of these weight slices from a single layer causes mismatching number of feature maps between layers of the network. A simple solution is to force the number of feature map between layers to match through the removal of weight slices from subsequent layers. This additional constraint becomes more apparent in DNNs with branches where multiple channels need to be pruned together to keep the network dense. Popular pruning saliency metrics do not factor in the structural dependencies that arise in DNNs with branches. We propose Domino metrics (built on existing channel saliency metrics) to reflect these structural constraints. We test Domino saliency metrics against the baseline channel saliency metrics on multiple networks with branches. Domino saliency metrics improved pruning rates in most tested networks and up to 25% in AlexNet on CIFAR-10.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 21 of its 36 references that are in the map. First posted on arXiv 4 May 2022.

Builds on

21 of 36 references.

References · 21

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,540
  3. 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,250
  4. 2015Learning both Weights and Connections for Efficient Neural Networkscited by 1,998
  5. 2016Pruning Filters for Efficient ConvNetscited by 1,129
  6. 2017Channel Pruning for Accelerating Very Deep Neural Networkscited by 697
  7. 2018Exploring the Limits of Weakly Supervised Pretrainingcited by 604
  8. 2019Importance Estimation for Neural Network Pruningcited by 531
  9. 2020ResNeSt: Split-Attention Networkscited by 445
  10. 2017ThiNet: A Filter Level Pruning Method for Deep Neural Network Compressioncited by 438
  11. 2018AMC: AutoML for Model Compression and Acceleration on Mobile Devicescited by 375
  12. 2017A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasetscited by 374
9 more
  1. 2021High-Performance Large-Scale Image Recognition Without Normalizationcited by 359
  2. 2018Soft Filter Pruning for Accelerating Deep Convolutional Neural Networkscited by 313
  3. 2016Dynamic Network Surgery for Efficient DNNscited by 233
  4. 2019Fixing the train-test resolution discrepancycited by 209
  5. 2015Fast ConvNets Using Group-wise Brain Damagecited by 81
  6. 2019Global Sparse Momentum SGD for Pruning Very Deep Neural Networkscited by 70
  7. 2017More is Less: A More Complicated Network with Less Inference Complexitycited by 66
  8. 2016ImageNet pre-trained models with batch normalizationcited by 14
  9. 2019Taxonomy of Saliency Metrics for Channel Pruningcited by 2

Built on by

Not cited yet.