mlsanityData to 14 Oct 2026

Pruning for Feature-Preserving Circuits in CNNs

arXiv 2206.01627 ↗First posted 3 Jun 2022cs.CV cs.LGcited by 4

Not on the map. Why?

Abstract (arXiv)

Deep convolutional neural networks are a powerful model class for a range of computer vision problems, but it is difficult to interpret the image filtering process they implement, given their sheer size. In this work, we introduce a method for extracting 'feature-preserving circuits' from deep CNNs, leveraging methods from saliency-based neural network pruning. These circuits are modular sub-functions, embedded within the network, containing only a subset of convolutional kernels relevant to a target feature. We compare the efficacy of 3 saliency-criteria for extracting these sparse circuits. Further, we show how 'sub-feature' circuits can be extracted, that preserve a feature's responses to particular images, dividing the feature into even sparser filtering processes. We also develop a tool for visualizing 'circuit diagrams', which render the entire image filtering process implemented by circuits in a parsable format.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 45 of its 72 references that are in the map. First posted on arXiv 3 Jun 2022.

Builds on

45 of 72 references.

References · 45

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  4. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  5. 2014Going Deeper with Convolutionscited by 4,241
  6. 2015VQA: Visual Question Answeringcited by 2,506
  7. 2013Describing Textures in the Wildcited by 2,125
  8. 2021Barlow Twins: Self-Supervised Learning via Redundancy Reductioncited by 1,838
  9. 2018SNIP: Single-shot Network Pruning based on Connection Sensitivitycited by 812
  10. 2021Scaling Vision with Sparse Mixture of Expertscited by 747
  11. 2018Explaining Explanations: An Overview of Interpretability of Machine Learningcited by 508
  12. 2020Picking Winning Tickets Before Training by Preserving Gradient Flowcited by 471
33 more
  1. 2015Neural Module Networkscited by 376
  2. 2015Understanding Neural Networks Through Deep Visualizationcited by 334
  3. 2018Soft Filter Pruning for Accelerating Deep Convolutional Neural Networkscited by 315
  4. 2014Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Partscited by 232
  5. 2020How Well Do Self-Supervised Models Transfer?cited by 213
  6. 2020Compositional Explanations of Neuronscited by 172
  7. 2015Conditional Computation in Neural Networks for faster modelscited by 167
  8. 2019Towards Optimal Structured CNN Pruning via Generative Adversarial Learningcited by 146
  9. 2018Net2Vec: Quantifying and Explaining how Concepts are Encoded by Filters in Deep Neural Networkscited by 129
  10. 2017Routing Networks: Adaptive Selection of Non-linear Functions for Multi-Task Learningcited by 126
  11. 2015Grounding of Textual Phrases in Images by Reconstructioncited by 112
  12. 2017Interpreting Deep Visual Representations via Network Dissectioncited by 103
  13. 2020Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Maskscited by 98
  14. 2018AutoPruner: An End-to-End Trainable Filter Pruning Method for Efficient Deep Model Inferencecited by 77
  15. 2020Progressive Skeletonization: Trimming more fat from a network at initializationcited by 64
  16. 2018Modular Networks: Learning to Decompose Neural Computationcited by 62
  17. 2018Explainable Neural Computation via Stack Neural Module Networkscited by 54
  18. 2018Exploiting Kernel Sparsity and Entropy for Interpretable CNN Compressioncited by 40
  19. 2017Exploring the Regularity of Sparse Structure in Convolutional Neural Networkscited by 37
  20. 2017An Entropy-based Pruning Method for CNN Compressioncited by 37
  21. 2019Approximated Oracle Filter Pruning for Destructive CNN Width Optimizationcited by 35
  22. 2021Neural Response Interpretation through the Lens of Critical Pathwayscited by 32
  23. 2020Pruning via Iterative Ranking of Sensitivity Statisticscited by 26
  24. 2021Clusterability in Neural Networkscited by 24
  25. 2017What are the Receptive, Effective Receptive, and Projective Fields of Neurons in Convolutional Neural Networks?cited by 16
  26. 2021Decomposing Convolutional Neural Networks into Reusable and Replaceable Modulescited by 14
  27. 2018Interpreting Layered Neural Networks via Hierarchical Modular Representationcited by 11
  28. 2020Cluster Pruning: An Efficient Filter Pruning Method for Edge AI Vision Applicationscited by 11
  29. 2021Quantifying Local Specialization in Deep Neural Networkscited by 10
  30. 2018Synaptic Strength For Convolutional Neural Networkcited by 5
  31. 2020Hierarchical Group Sparse Regularization for Deep Convolutional Neural Networkscited by 5
  32. 2015Multi-path Convolutional Neural Networks for Complex Image Classificationcited by 4
  33. 2016Improving Neural Network Generalization by Combining Parallel Circuits with Dropoutcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 1; 2025: 0; 2026: 3.

4 papers.

  1. 2024Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficientcited by 40
  2. 2026Certified Circuits: Stability Guarantees for Mechanistic Circuitscited by 5
  3. 2026Engineering Resource-constrained Software Systems with DNN Components: a Concept-based Pruning Approachcited by 0
  4. 2026From Interpretability Methods to Interpretable Modelscited by 0