mlsanityData to 14 Oct 2026

Modelling and analysis of the 8 filters from the "master key filters hypothesis" for depthwise-separable deep networks in relation to idealized receptive fields based on scale-space theory

arXiv 2509.12746 ↗First posted 16 Sep 2025cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

This paper presents the results of analysing and modelling a set of 8 ``master key filters'', which have been extracted by applying a clustering approach to the receptive fields learned in depthwise-separable deep networks based on the ConvNeXt architecture. For this purpose, we first compute spatial spread measures in terms of weighted mean values and weighted variances of the absolute values of the learned filters, which support the working hypotheses that: (i) the learned filters can be modelled by separable filtering operations over the spatial domain, and that (ii) the spatial offsets of the those learned filters that are non-centered are rather close to half a grid unit. Then, we model the clustered ``master key filters'' in terms of difference operators applied to a spatial smoothing operation in terms of the discrete analogue of the Gaussian kernel, and demonstrate that the resulting idealized models of the receptive fields show good qualitative similarity to the learned filters. This modelling is performed in two different ways: (i) using possibly different values of the scale parameters in the coordinate directions for each filter, and (ii) using the same value of the scale parameter in both coordinate directions. Then, we perform the actual model fitting by either (i) requiring spatial spread measures in terms of spatial variances of the absolute values of the receptive fields to be equal, or (ii) minimizing the discrete l1- or l2-norms between the idealized receptive field models and the learned filters. Complementary experimental results then demonstrate the idealized models of receptive fields have good predictive properties for replacing the learned filters by idealized filters in depthwise-separable deep networks, thus showing that the learned filters in depthwise-separable deep networks can be well approximated by discrete scale-space filters.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 22 of its 38 references that are in the map. First posted on arXiv 16 Sep 2025.

Builds on

22 of 38 references.

References · 22

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,845
  4. 2022A ConvNet for the 2020scited by 3,579
  5. 2016Xception: Deep Learning with Depthwise Separable Convolutionscited by 2,332
  6. 2023ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoderscited by 712
  7. 2016Structured Receptive Fields in CNNscited by 32
  8. 2020PDE-based Group Equivariant Convolutional Neural Networkscited by 29
  9. 2021Resolution learning in deep convolutional networks using scale-space theorycited by 26
  10. 2020Scale-covariant and scale-invariant Gaussian derivative networkscited by 25
  11. 2017Normative theory of visual receptive fieldscited by 15
  12. 2022Scale Equivariant U-Netcited by 13
10 more
  1. 2023Scale-Equivariant UNet for Histopathology Image Segmentationcited by 11
  2. 2022Fully trainable Gaussian derivative convolutional layercited by 10
  3. 2022A time-causal and time-recursive scale-covariant scale-space representation of temporal signals and past timecited by 8
  4. 2023Discrete approximations of Gaussian smoothing and Gaussian derivativescited by 7
  5. 2021Frequency learning for structured CNN filters with Gaussian fractional derivativescited by 6
  6. 2024Unveiling the Unseen: Identifiable Clusters in Trained Depthwise Convolutional Kernelscited by 6
  7. 2024Approximation properties relative to continuous scale space for hybrid discretizations of Gaussian derivative operatorscited by 4
  8. 2024Scale generalisation properties of extended scale-covariant and scale-invariant Gaussian derivative networks on image datasets with spatial scaling variationscited by 3
  9. 2024The Master Key Filters Hypothesis: Deep Filters Are Generalcited by 3
  10. 2025The Quest for Universal Master Key Filters in DS-CNNscited by 1

Built on by

Not cited yet.