mlsanityData to 14 Oct 2026

Visual Concept Connectome (VCC): Open World Concept Discovery and their Interlayer Connections in Deep Models

arXiv 2404.02233 ↗First posted 2 Apr 2024cs.CVcited by 22

Not on the map. Why?

Abstract (arXiv)

Understanding what deep network models capture in their learned representations is a fundamental challenge in computer vision. We present a new methodology to understanding such vision models, the Visual Concept Connectome (VCC), which discovers human interpretable concepts and their interlayer connections in a fully unsupervised manner. Our approach simultaneously reveals fine-grained concepts at a layer, connection weightings across all layers and is amendable to global analysis of network structure (e.g., branching pattern of hierarchical concept assemblies). Previous work yielded ways to extract interpretable concepts from single layers and examine their impact on classification, but did not afford multilayer concept analysis across an entire network architecture. Quantitative and qualitative empirical results show the effectiveness of VCCs in the domain of image classification. Also, we leverage VCCs for the application of failure mode debugging to reveal where mistakes arise in deep networks.

Key facts. Cited by 22 papers in the map (data through 14 Oct 2026). Builds on 39 of its 61 references that are in the map. First posted on arXiv 2 Apr 2024.

Builds on

39 of 61 references.

References · 39

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  4. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  5. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  6. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  7. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  8. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
  9. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
  10. 2014Going Deeper with Convolutionscited by 4,241
  11. 2015Learning Deep Features for Discriminative Localizationcited by 2,436
  12. 2013Visualizing and Understanding Convolutional Networkscited by 2,290
27 more
  1. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,917
  2. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,432
  3. 2017Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)cited by 1,319
  4. 2018Sanity Checks for Saliency Mapscited by 1,156
  5. 2020Concept Bottleneck Modelscited by 1,032
  6. 2022Toy Models of Superpositioncited by 971
  7. 2017Grad-CAM++: Improved Visual Explanations for Deep Convolutional Networkscited by 939
  8. 2021Multiscale Vision Transformerscited by 897
  9. 2017Network Dissection: Quantifying Interpretability of Deep Visual Representationscited by 708
  10. 2020Transformer Interpretability Beyond Attention Visualizationcited by 627
  11. 2017Interpretation of Neural Networks is Fragilecited by 444
  12. 2014Understanding Deep Image Representations by Inverting Themcited by 403
  13. 2017The (Un)reliability of saliency methodscited by 333
  14. 2019On Completeness-aware Concept-Based Explanations in Deep Neural Networkscited by 239
  15. 2020Concept Whitening for Interpretable Image Recognitioncited by 225
  16. 2016Layer-wise Relevance Propagation for Neural Networks with Local Renormalization Layerscited by 199
  17. 2022CRAFT: Concept Recursive Activation FacTorization for Explainabilitycited by 178
  18. 2022The Disagreement Problem in Explainable Machine Learning: A Practitioner's Perspectivecited by 159
  19. 2020Invertible Concept-based Explanations for CNN Models with Non-negative Concept Activation Vectorscited by 120
  20. 2020Rewriting a Deep Generative Modelcited by 107
  21. 2018Interpreting CNNs via Decision Treescited by 89
  22. 2022B-cos Networks: Alignment is All We Need for Interpretabilitycited by 88
  23. 2017Interpreting CNN Knowledge via an Explanatory Graphcited by 49
  24. 2021A Framework for Learning Ante-hoc Explainable Models via Conceptscited by 45
  25. 2024Understanding Video Transformers via Universal Concept Discoverycited by 23
  26. 2022SegDiscover: Visual Concept Discovery via Unsupervised Semantic Segmentationcited by 13
  27. 2022Deeply Explain CNN via Hierarchical Decompositioncited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 6; 2025: 9; 2026: 7.

22 papers.

Sort:

By area:

  1. 2025Archetypal SAE: Adaptive and Stable Dictionary Learning for Concept Extraction in Large Vision Modelscited by 73
  2. 2025Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignmentcited by 62
  3. 2024Explainable and Interpretable Multimodal Large Language Models: A Comprehensive Surveycited by 52
  4. 2025Visual Large Language Models for Generalized and Specialized Applicationscited by 36
  5. 2024Concept-skill Transferability-based Data Selection for Large Vision-Language Modelscited by 24
  6. 2025Block-Recurrent Dynamics in Vision Transformerscited by 20
  7. 2024DCBM: Data-Efficient Visual Concept Bottleneck Modelscited by 16
  8. 2026Concept Influence: Leveraging Interpretability to Improve Performance and Efficiency in Training Data Attributioncited by 8
  9. 2025Granular Concept Circuits: Toward a Fine-Grained Circuit Discovery for Concept Representationscited by 5
  10. 2026Bi-Orthogonal Factor Decomposition for Vision Transformerscited by 5
  11. 2026Certified Circuits: Stability Guarantees for Mechanistic Circuitscited by 5
  12. 2026Structuring Sparsity: Block-Sparse Featurizers Capture Visual Concept Manifoldscited by 5
10 more
  1. 2025NeurFlow: Interpreting Neural Networks through Neuron Groups and Functional Interactionscited by 3
  2. 2025A Geometric Unification of Concept Learning with Concept Conescited by 3
  3. 2024Multi-Depth Concept Extraction for Post-Hoc Vision Encoder Explanationcited by 2
  4. 2024OMENN: One Matrix to Explain Neural Networkscited by 2
  5. 2024Decompose the model: Mechanistic interpretability in image models with Generalized Integrated Gradients (GIG)cited by 1
  6. 2025FaCT: Faithful Concept Traces for Explaining Neural Network Decisionscited by 1
  7. 2026Seeing Through Circuits: Faithful Mechanistic Interpretability for Vision Transformerscited by 1
  8. 2026From Local to Global to Mechanistic: An iERF-Centered Unified Framework for Interpreting Vision Modelscited by 1
  9. 2025TRANSPORTER: Transferring Visual Semantics from VLM Manifoldscited by 0
  10. 2026SPICE: Simple Polysemantic Feature Interpretation via Clustering-based Explanationcited by 0