mlsanityData to 7 Oct 2026

Hidden in Plain Sight -- Class Competition Focuses Attribution Maps

arXiv 2503.07346 ↗First posted 10 Mar 2025cs.CV cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

Attribution methods reveal which input features a neural network uses for a prediction, adding transparency to their decisions. A common problem is that these attributions seem unspecific, highlighting both important and irrelevant features. We revisit the common attribution pipeline and observe that using logits as attribution target is a main cause of this phenomenon. We show that the solution is in plain sight: considering distributions of attributions over multiple classes using existing attribution methods yields specific and fine-grained attributions. On common benchmarks, including the grid-pointing game and randomization-based sanity checks, this improves the ability of 18 attribution methods across 7 architectures up to 2x, agnostic to model architecture.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 32 of its 39 references that are in the map. First posted on arXiv 10 Mar 2025.

Builds on

32 of 39 references.

References · 32

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  4. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  5. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  6. 2016Densely Connected Convolutional Networkscited by 6,691
  7. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,690
  8. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  9. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,040
  10. 2022A ConvNet for the 2020scited by 3,545
  11. 2017Axiomatic Attribution for Deep Networkscited by 3,350
  12. 2016Wide Residual Networkscited by 3,057
20 more
  1. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,348
  2. 2013Visualizing and Understanding Convolutional Networkscited by 2,278
  3. 2017Learning Important Features Through Propagating Activation Differencescited by 1,454
  4. 2018Sanity Checks for Saliency Mapscited by 1,149
  5. 2014Striving for Simplicity: The All Convolutional Netcited by 983
  6. 2018RISE: Randomized Input Sampling for Explanation of Black-box Modelscited by 825
  7. 2020Quantifying Attention Flow in Transformerscited by 759
  8. 2020Transformer Interpretability Beyond Attention Visualizationcited by 620
  9. 2020Captum: A unified and generic model interpretability library for PyTorchcited by 542
  10. 2019Understanding Deep Networks via Extremal Perturbations and Smooth Maskscited by 257
  11. 2019XRAI: Better Attributions Through Regionscited by 116
  12. 2021PartImageNet: A Large, High-Quality Dataset of Partscited by 107
  13. 2022B-cos Networks: Alignment is All We Need for Interpretabilitycited by 88
  14. 2025ConceptAttention: Diffusion Transformers Learn Highly Interpretable Featurescited by 48
  15. 2022Towards Better Understanding Attribution Methodscited by 28
  16. 2020A Framework to Learn with Interpretationcited by 23
  17. 2022Shortcomings of Top-Down Randomization-Based Sanity Checks for Evaluations of Deep Neural Network Explanationscited by 22
  18. 2024IG2: Integrated Gradient on Iterative Gradient Path for Feature Attributioncited by 12
  19. 2025How to Probe: Simple Yet Effective Techniques for Improving Post-hoc Explanationscited by 8
  20. 2024Layer-Wise Relevance Propagation with Conservation Property for ResNetcited by 7

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026What is Missing? Explaining Neurons Activated by Absent Conceptscited by 1